AI Agent在科研全流程中的应用趋势与发展机会分析（2026-2029）

深度研究匿名用户发表于 2026年05月06日 14:5986阅读

内容由 AI 生成，仅供参考，请仔细甄别

参考文献

9The promise of artificial intelligence in chemical engineering: Is it here, finally?OpenAlex

Venkat Venkatasubramanian

The current excitement about artificial intelligence (AI), particularly machine learning (ML), is palpable and contagious. The expectation that AI is poised to "revolutionize," perhaps even take over, humanity has elicited prophetic visions and concerns from some luminaries.1-4 There is also a great deal of interest in the commercial potential of AI, which is attracting significant sums of venture capital and state-sponsored investment globally, particularly in China.5 McKinsey, for instance, predicts the potential commercial impact of AI in several domains, envisioning markets worth trillions of dollars.6 All this is driven by the sudden, explosive, and surprising advances AI has made in the last 10 years or so. AlphaGo, autonomous cars, Alexa, Watson, and other such systems, in game playing, robotics, computer vision, speech recognition, and natural language processing are indeed stunning advances. But, as with earlier AI breakthroughs, such as expert systems in the 1980s and neural networks in the 1990s, there is also considerable hype and a tendency to overestimate the promise of these advances, as market research firm Gartner and others have noted about emerging technology.7 It is quite understandable that many chemical engineers are excited about the potential applications of AI, and ML in particular,8 for use in such applications as catalyst design.9-11 It might seem that this prospect offers a novel approach to challenging, long-standing problems in chemical engineering using AI. However, the use of AI in chemical engineering is not new—it is, in fact, a 35-year-old ongoing program with some remarkable successes along the way. This article is aimed broadly at chemical engineers who are interested in the prospects for AI in our domain, as well as at researchers new to this area. The objectives of this article are threefold. First, to review the progress we have made so far, highlighting past efforts that contain valuable lessons for the future. Second, drawing on these lessons, to identify promising current and future opportunities for AI in chemical engineering. To avoid getting caught up in the current excitement and to assess the prospects more carefully, it is important to take such a longer and broader view, as a "reality check." Third, since AI is going to play an increasingly dominant role in chemical engineering research and education, it is important to recount and record, however incomplete, certain early milestones for historical purposes. It is apparent that chemical engineering is at an important crossroads. Our discipline is undergoing an unprecedented transition—one that presents significant challenges and opportunities in modeling and automated decision-making. This has been driven by the convergence of cheap and powerful computing and communications platforms, tremendous progress in molecular engineering, the ever-increasing automation of globally integrated operations, tightening environmental constraints, and business demands for speedier delivery of goods and services to market. One important outcome from this convergence is the generation, use, and management of massive amounts of diverse data, information, and knowledge, and this is where AI, particularly ML, would play an important role. Some of these are application-focused, such as game playing and vision. Others are methodological, such as expert systems and ML—the two branches that are most directly and immediately applicable to our domain, and hence the focus of this article. These are the ones that have been investigated the most in the last 35 years by AI researchers in chemical engineering. While the current "buzz" is mostly around ML, the expert system framework holds important symbolic knowledge representation concepts and inference techniques that could prove useful in the years ahead as we strive to develop more comprehensive solutions that go beyond the purely data-centric emphasis of ML. Many tasks in these different branches of AI share certain common features. They all require pattern recognition, reasoning, and decision-making under complex conditions. And they often deal with ill-defined problems, noisy data, model uncertainties, combinatorially large search spaces, nonlinearities, and the need for speedy solutions. But such features are also found in many problems in process systems engineering (PSE)—in synthesis, design, control, scheduling, optimization, and risk management. So, some of us thought, in the early-1980s, that we should examine such problems from an AI perspective.15-17 Just as it is today, the excitement about AI at that time was centered on expert systems. It was palpable and contagious, with high expectations for AI's near-term potential.18-20 Hundreds of millions of dollars were invested in AI start-ups as well as within large companies. AI spurred the development of special purpose hardware, called Lisp machines (e.g., Symbolics Lisp machines). Promising proof-of-concept systems were demonstrated in many domains, including chemical engineering (see below). In this phase, it was expected that AI would have a significant impact in chemical engineering in the near future. However, unlike optimization and model predictive control, AI did not quite live up to its early promise. So, what happened? Why was not AI as impactful? Before addressing this question, it is necessary to examine the different phases of AI, as I classify them, in chemical engineering. While major efforts to developing AI methods for chemical engineering problems started in the early 1980s, it is remarkable that some researchers (for instance, Gary Powers, Dale Rudd, and Jeff Siirola) were investigating AI in PSE in the late 1960s and early 1970s.21 In particular, the Adaptive Initial DEsign Synthesizer system, developed by Siirola and Rudd22 for process synthesis, represents a significant development. This was arguably the first system that employed AI methods such as means-and-ends analysis, symbolic manipulation, and linked data structures in chemical engineering. Phase I, the Expert Systems Era (from the early 1980s through the mid-1990s), saw the first broad effort to exploit AI in chemical engineering. Expert systems, also called knowledge-based systems, rule-based systems, or production systems, are computer programs that mimic the problem-solving of humans with expertise in a given domain.23, 24 Expert problem-solving typically involves large amounts of specialized knowledge, called domain knowledge, often in the form of rules of thumb, called heuristics, typically learned and refined over years of problem-solving experience. The amount of knowledge manipulated is often vast, and the expert system rapidly narrows down the search by recognizing patterns and by using the appropriate heuristics. The architecture of these systems was inspired by the stimulus–response model of cognition from psychology and pattern-matching-and-search model of symbolic computation, which originated in Emil Post's work in symbolic logic. Building on this work, Simon and Newell in the late 1960s and 1970s devised the production system framework, an important conceptual, representational, and architectural breakthrough, for developing expert systems.25-27 The crucial insight here was that one needs to, and one can, separate domain knowledge from its order of execution, that is, from search or inference, thereby achieving the necessary computational flexibility to address ill-structured problems. In contrast, conventional programs consist of a set of statements whose order of execution is predetermined. Therefore, if the execution order is not known or cannot be anticipated a priori, as in the case of medical diagnosis, for example, this approach will not work. Expert systems programming alleviated this problem by making a clear distinction between the knowledge base and the search or inference strategy. This not only allowed for flexible execution, it also facilitated the incremental addition of knowledge, without distorting the overall program structure. This rule-based knowledge representation and architecture are intuitive, and relatively easy to understand and generate explanations about the system's decisions. This new approach facilitated the development of a number of impressive expert systems, starting with MYCIN, an expert system for diagnosing infectious diseases28 developed at Stanford University during 1972–82. This led to other successful systems such as PROSPECTOR (for mineral prospecting29), R1 (configuring Vax computers30), and so on, in this era. These systems inspired the first expert system application in chemical engineering, CONPHYDE, developed in 1983 by Bañares-Alcántara, Westerberg, and Rychner at Carnegie Mellon16 for predicting thermophysical properties of complex fluid mixtures. CONPHYDE was implemented using Knowledge Acquisition System that was used for PROSPECTOR. This was quickly followed by DECADE, in 1985, again from the same CMU researchers,17 for catalyst design. There was other such remarkable early work in process synthesis, design, modeling, and diagnosis as well. In synthesis and in design, for instance, important conceptual advances were made by Stephanopoulos and his students, starting with Design-Kit,31 and in modeling, MODELL.LA, a language for developing process models.32 In process fault diagnosis, Davis33 and Kramer,34, 35 and their groups, made important contributions in the same period. My group developed causal model-based diagnostic expert systems,36 a departure from the heuristics-based approach, which was the dominant theme of the time. We also demonstrated the potential of learning expert systems, an unusual idea at that time as automated learning in expert systems was not in vogue.37 The need for causal models in AI, a topic that has emerged as very important now,38 was also recognized in those early years.39 This period also saw expert system work commencing in Europe,40 particularly for conceptual design support. An important large-scale program in this era was the Abnormal Situation Management (ASM) consortium, funded at $17 million by the National Institute of Standards and Technology's Advanced Technology Program and by the leading oil companies, under the leadership of Honeywell.41 Three different academic groups, led by Davis (Ohio State), Vicente (University of Toronto), and myself at Purdue, were also involved in the consortium. This program is the forerunner to the current Clean Energy Smart Manufacturing Innovation Institute that was funded in 2016.42 The first course on AI in PSE was developed and taught at Columbia University in 1986, and it was subsequently offered at Purdue University for many years. The earlier offerings had an expert systems emphasis, but as ML advanced, in later years, the course evolved to include topics such as clustering, neural networks, statistical classifiers, graph-based models, and genetic algorithms. In 1986, Stephanopoulos published an article43 titled, "Artificial Intelligence in Process Engineering", in which he discussed the potential of AI in process engineering and outlined a research program to realize it. Coincidentally, in the same issue, I had a article with the same title, which described the Columbia course.44 In my article, I discussed topics from the course, and it mirrored what Stephanopoulos had outlined as the research program. (Curiously, we did not know each other at that time and had written our articles independently, yet with the same title, at the same time, with almost the same content, and had submitted to the same journal for the same issue!) The first AIChE session on AI was organized by Gary Powers (CMU) at the annual meeting held in Chicago in 1985. The first national meeting on AI in process engineering was held in 1987 at Columbia University, co-organized by Venkatasubramanian, Stephanopoulos, and Davis, sponsored by the National Science Foundation, American Association for Artificial Intelligence, and Air Products. The first international conference, Intelligent Systems in Process Engineering (ISPE'95), sponsored by the Computer Aids for Chemical Engineering (CACHE) Corporation, was co-organized by Stephanopoulos, Davis, and Venkatasubramanian, held at Snowmass, CO, in July 1995. The CACHE Corporation had also organized an Expert Systems Task Force in 1985, under the leadership of Stephanopoulos, to develop tools for the instruction of AI in chemical engineering.45 The task force published a series of monographs on AI in process engineering during 1989–1993. Despite impressive successes, the expert system approach did not quite take-off as it suffered from serious drawbacks. It took a lot of effort, time, and money to develop a credible expert system for industrial applications. Furthermore, it was also difficult and expensive to maintain and update the knowledge base as new information came in or the target application changed, such as in the retrofitting of a chemical plant. This approach did not scale well for practical applications (more on this in sections Lack of impact of AI during Phases I and II and Are things different now for AI to have impact?). As the excitement about expert systems waned in the 1990s due to these practical difficulties, interest in another AI technique was picking up greatly. This was the beginning of Phase II, the Neural Networks Era, roughly from 1990 onward. This was a crucial shift from the top-down design paradigm of expert systems to the bottom-up paradigm of neural nets that acquired knowledge automatically from large amounts of data, thus easing the maintenance and development of models. It all started with the reinvention of the backpropagation algorithm by Rumelhart, Hinton, and Williams in 1986 for training feedforward neural networks to learn hidden patterns in input–output data. It had been proposed earlier, in 1974, by Paul Werbos as part of his Ph.D. thesis at Harvard. It is essentially an algorithm for implementing gradient descent search, using the chain rule in calculus, to propagate errors back through the network to adjust the strength (i.e., weights) of connections between nodes iteratively, to make the network learn the patterns. While the idea of neural networks had been around since 1943 from the work of McCulloch and Pitts, and was further developed by Rosenblatt, Minsky, and Papert in the 1960s, these earlier models were limited in scope as they could not handle problems with nonlinearity. The key breakthrough this time was the ability to solve nonlinear function approximation and nonlinear classification problems in an automated manner using the backpropagation learning algorithm. The typical structure of a feedforward neural network from this era is shown in Figure 1, with its input, hidden, and output layers of neurons, and their associated signals, weights and biases. The figure also shows examples of nonlinear function approximation and nonlinear classification problems such networks were able to solve provided enough data were available.46 (a) Architecture of a feedforward neural network. (b) Examples of nonlinear function approximation and classification problems. Adapted from: https://medium.com/@curiousily/tensorflow-for-hackers-part-iv-neural-network-from-scratch-1a4f504dfa8 https://neustan.wordpress.com/2015/09/05/neural-networks-vs-svm-where-when-and-above-all-why/ http://mccormickml.com/2015/08/26/rbfn-tutorial-part-ii-function-approximation/ This novel automated nonlinear modeling ability spurred a tremendous amount of work in a variety of domains including chemical engineering.47 Researchers made substantial progress on addressing challenging problems in modeling,48, 49 fault diagnosis,50-55 control,56, 57 and product design.58 In particular, the recognition of the connection between the autoencoder architecture and the nonlinear principal component analysis by Kramer,48 and the recognition of the nature of the basis function approximation of neural networks through the WaveNet architecture by Bakshi and Stephanopoulos49 are outstanding contributions. There were hundreds of articles in our domain during this phase and only some of the earliest and key articles are highlighted here. While this phase was largely driven by neural networks, researchers also made progress on expert systems (such as the ASM consortium) and genetic algorithms at that time. For instance, we proposed59 directed evolution of engineering polymers in silico using genetic algorithms. This led in subsequent years60 to the multiscale model-based informatics framework called Discovery Informatics61 for materials design. The discovery informatics framework led to the successful development of materials design systems using directed evolution in several industrial applications, such as gasoline additives,62 formulated rubbers,63 and catalyst design.64 During this period, researchers were also beginning to realize the challenges and opportunities in multiscale modeling using informatics techniques.65, 66 Other important advances not using neural networks included research into frameworks and architectures for building AI systems, such as blackboard architectures, integrated problem-solving-and-learning systems, and cognitive architectures. Architectures such as Prodigy and Soar are examples of this work.67 Similarly, there was progress in process synthesis and in design,68 domain-specific representations and languages,32, 69 domain-specific compilers,70 ontologies,71, 72 modeling environments,32, 73 molecular structure search engines,74 automatic reaction network generators,64 and chemical entities extraction systems.74 These references by no means constitute a comprehensive list. All this work, and others along similar lines, performed some two decades ago is still relevant and useful today in the modern era of data science. Building such systems using modern tools presents major opportunities. Despite the surprising success of neural networks in many practical applications, some especially challenging problems in vision, natural language processing, and speech understanding remained beyond the capabilities of the neural nets of this era. Researchers that one would need neural nets with many more hidden not but training these to be So, the was more or for about a or so a breakthrough for training neural thus the current phase which we in Phases of AI in Chemical In of all this effort over two AI was not as in chemical engineering as we had In it is clear this was the First, the problems we were are challenging even Second, we were the powerful and programming to address such challenging problems. Third, we were limited by data. that was was very There were of challenges in Phases I and and While we made progress on the conceptual such as knowledge representation and inference for problems in synthesis, design, diagnosis, and we could not the challenges and involved in practical applications. In there was no as it there was no in the that the in process engineering, in that period, could be more by optimization and by as algorithms and over the years, these well on problems for which we could and solve models. the problems for which such models are difficult to (e.g., diagnosis, analysis, and materials or almost to generate (e.g., speech which computational and data, of which were not during this period. This of practical success led to two one at the of the Expert Systems era and the other at the of the Neural Networks for AI research in computer and in the application This progress even In it typically to take about years for a to and have from discovery to For instance, for the such as to about market it took about years from the time computer of chemical was first proposed in the similar in optimization as for programming and nonlinear programming and for In during Phase I and II, AI as a was only about years It was early to This analysis that one could impact around While predicting and impact is an this given the current of AI. As it for those of us who started on AI in the early-1980s, we were early as as impact is but it was challenging and to these problems. Many of the such as developing AI methods and causal model-based AI systems, are still as I The progress of AI over the last or so has been very and the are largely have been and are also as have started to and from systems, such as Alexa, and more and for a variety of are beginning to and to work It is and to make the In 1985, arguably the most powerful computer was the computational was and it of The million machine million in was and a to it. So, what would it would the In fact, the is more powerful the The at of it is a on the There have been advances in the of algorithms and in programming such as and are the we had to program in Lisp for to what now be in a with a of We have also great progress in The other development is the of tremendous amounts of data, in many domains, which made the stunning advances in ML (more on this below). All this is for this to without for the last years, its expected making these stunning advances As a the is here. The I is also here of the that could be using optimization and have largely been for further for further one go up the and that means going challenging decision-making problems that require solutions. So, now we have a back some years from would that there were early milestones in AI. One is Gary in in the in and the is the surprising by in The AI advances that made these are now poised to have an impact that beyond game In my view, we Phase around the era of Science or This new phase was made by important or neural nets and statistical ML. These are the that are the AI success in game playing, natural language processing, robotics, and vision. neural nets of the 1990s, which typically had only one hidden of neurons, neural nets have hidden as shown in Figure an architecture has the potential to features for complex pattern However, such networks were to using the backpropagation or gradient descent algorithm. The breakthrough came in by using a training with considerable in processing in the form of processing In a called in the training of the neural made such extraction is a in the domain of processing, for features from a noisy the of the network architecture and the such as the and number of a during from a very large data this is a crucial appropriate that to a successful by the neural network from architectural was the neural feedforward neural network has no of and the only it is the current it has been This is not appropriate for problems which have information, such as time series data, where what typically on what has For instance, to the in a one needs to know which came it. networks address such problems by as their not the current example, but also what they have the output on what has the network as if it has This was further by another architectural called the The typical of a an an output and a The over time and the the of information into and of the networks are well for making on time series data, since there be of between important in a time While the key advances here are in the architecture and training of large-scale neural networks, the important be of as a for learning a of to a such as an It is a learning in which an the by its on the it in to its with the is the one to a a where one the with a if it the and it if it this is many one is essentially the patterns to the it the This learning is essentially programming in modern ML For this approach to work well for complex problems, such as the game of one needs millions of millions of to learn the game from of worth of expertise and during a period of a As stunning as this is, one that the game playing domain has the that it almost training data over training with a great deal of This is typically not the case in and engineering, where one is even in this era. But this might be the of the data is a computer as in some materials applications. For the of it is important to that learning from the other dominant learning and In the system the between and output given a set of input–output the other in only a set of is given with no (i.e., no The system is to the in the data on its hence One could that learning for

20Comparative effectiveness of anti-seizure medications in emulated trials using medical informatics.PubMed

Kevin Xie, Jacob Korzun, Daniel J Zhou, et al.

Brain. 2025 Dec 4;148(12):4288-4298. doi: 10.1093/brain/awaf238.

Anti-seizure medications (ASMs) are often prescribed using a trial-and-error approach with a similar sequence for many patients. Comparative effectiveness data beyond the first ASM prescription are limited. Artificial intelligence can automatically extract information from electronic health records (EHRs) and augment clinical trials. We conducted a retrospective cohort comparative effectiveness study of currently available ASMs using emulated clinical trials through a causal inference- and informatics-based framework. We extracted data from EHRs using natural language processing algorithms to collect epilepsy covariates and seizure outcomes. We compared ASMs using weighted Kaplan-Meier analyses and log-rank tests with Bonferroni corrections. We compared ASMs across seizure freedom over 2 years of follow-up, and drug retention rate over 5 years of follow-up. Emulated trials included 1596 patients (8379 patient-years) for seizure freedom as outcome and 2945 patients (14 238 patient-years) for retention as outcome. For all epilepsy types, among first-line ASMs, levetiracetam and lamotrigine were superior to oxcarbazepine for seizure freedom (P < 0.001), and levetiracetam was superior to lamotrigine and oxcarbazepine for retention (P < 0.001); among second-line ASMs, lacosamide had the best retention (P < 0.001) but lower seizure freedom rate than topiramate (P = 0.032); among third-line ASMs, cenobamate had longer retention than brivaracetam, but lower seizure freedom rate than clobazam and brivaracetam (P < 0.001). For focal epilepsies, first-line ASMs achieved similar rates of seizure freedom, but levetiracetam remained superior to lamotrigine and oxcarbazepine for retention (P < 0.001); among second-line ASMs, topiramate and lacosamide were both superior to zonisamide for seizure freedom (P < 0.001), while lacosamide remained superior to topiramate and zonisamide for retention (P < 0.002); among third-line ASMs, cenobamate, brivaracetam and clobazam achieved similar retention and seizure freedom. For generalized and unclassified epilepsies, first-line ASMs achieved similar seizure freedom, but levetiracetam remained superior to lamotrigine and oxcarbazepine for retention (P < 0.02); topiramate had the best seizure freedom among second-line ASMs (P < 0.001), while lacosamide again had better retention than zonisamide (P < 0.006); there were insufficient patients included in third-line ASM trials for this subgroup. In conclusion, emulated clinical trials provide a useful framework for studying the comparative effectiveness of anti-seizure medications using real-world observational data. Our findings of differences in seizure freedom and retention between ASMs help to generate hypotheses that should be tested in future prospective, randomized comparative effectiveness trials.

30Troubling Trends in Biomedical Research Publication: "Publish or Perish" Results in a Propensity for Ethical Violations.PubMed

James H Lubowitz, Elizabeth Matzkin, Michael J Rossi

Arthroscopy. 2025 Apr;41(4):859-862. doi: 10.1016/j.arthro.2024.12.017. Epub 2024 Dec 20.

As a result of the "publish or perish" environment for biomedical journal authors, as well as new developments in open access publication models ("pay to publish") and rapid improvements in artificial intelligence large language models (AI LLMs; e.g., ChatGPT), troubling trends and a propensity for ethical violations now exist. Credit is commonly being taken for authorship by those who fail to meet authorial criteria, which is unethical. Experienced coauthors are providing inadequate diligence in drafting, critical review, and final approval of submitted articles, which is unethical or, at the very least, careless. Research lacking originality ("copycat" studies) is becoming common, which although not unethical, is uninteresting and creates a burden for journal reviewers, editors, and, most of all, readers. Publication of least publishable units (LPU or "salami slicing"), where authors divide a single research publication into a number of papers with small amounts of information in each paper, results in quantity rather than quality and is ethically inappropriate. LPU can result in redundancy, self-plagiarism, publication overlap, and duplicate reporting of patient data that can result in inaccurate conclusions in systematic reviews. Duplicate submission of a paper to more than one journal (in the same or different languages), such that the paper is under peer review by multiple journals at the same time is unethical and can result in two or more journals publishing the same article. Duplicate publication (publication of a paper that overlaps substantially with one already published), without clear reference to the previous publication is unethical. Predatory journals, with low standards of quality or peer review, and predatory practices by publishers and owners of ostensibly nonpredatory journals, can result in solicitation and acceptance of articles (as well as author publication charges or fees) for the purpose of generating revenue, rather than for legitimate editorial reasons based on article quality and content. This is unethical. Opportunities exist to mitigate against these trends, and by naming these trends and identifying opportunities to avoid ethical transgression, our well-intentioned community can publish our scholarship in an unimpeachable manner.

35Navigating academic integrity in biomedical research: the impact of large language models on current practices and future directions.PubMed

Anqi Lin, Zuwei Chen, Aimin Jiang, et al.

Int J Surg. 2026 Feb 1;112(2):4418-4433. doi: 10.1097/JS9.0000000000003839. Epub 2025 Nov 21.

As large language models (LLMs) continue to advance, they have garnered widespread public attention and extensive application across numerous industries and academic disciplines. The proliferation of LLMs has sparked considerable research interest, with studies primarily focusing on their technical characteristics and specific application scenarios. However, systematic research examining the impact of LLMs on academic integrity remains relatively scarce. Academic integrity is of paramount importance in the biomedical field. Therefore, this paper aims to examine both the opportunities and challenges that LLMs present to academic integrity in the biomedical field, and proposes solutions for optimizing the beneficial applications of LLMs. From a positive perspective, LLMs offer substantial benefits to researchers by enhancing research efficiency, improving research quality, and facilitating the generation and dissemination of academic insights. However, they also present numerous challenges, including the potential for promoting academic misconduct, generating content inaccuracies or ambiguous expressions, introducing bias and fairness concerns, compromising peer review mechanisms, facilitating the dissemination of misinformation, and undermining higher education - all of which demand careful attention. To address these issues, we propose solutions and feasible strategies centered on ten core dimensions: establishing policies and regulatory guidelines, enhancing AI literacy and application capabilities, developing and improving relevant technical tools, establishing human-AI collaboration models, reforming peer review procedures and academic evaluation systems, promoting international cooperation and standardization, increasing transparency and strengthening disclosure, reinforcing professional ethics education, and advancing artificial intelligence detection technologies. Overall, while LLMs undoubtedly pose challenges for maintaining academic integrity, their potential for positive impact remains promising. It is anticipated that with technological advancement and improved ethical standards, LLMs will ultimately preserve and strengthen academic integrity.

37ChatGPT Utility in Healthcare Education, Research, and Practice: Systematic Review on the Promising Perspectives and Valid ConcernsOpenAlex

Malik Sallam

ChatGPT is an artificial intelligence (AI)-based conversational large language model (LLM). The potential applications of LLMs in health care education, research, and practice could be promising if the associated valid concerns are proactively examined and addressed. The current systematic review aimed to investigate the utility of ChatGPT in health care education, research, and practice and to highlight its potential limitations. Using the PRIMSA guidelines, a systematic search was conducted to retrieve English records in PubMed/MEDLINE and Google Scholar (published research or preprints) that examined ChatGPT in the context of health care education, research, or practice. A total of 60 records were eligible for inclusion. Benefits of ChatGPT were cited in 51/60 (85.0%) records and included: (1) improved scientific writing and enhancing research equity and versatility; (2) utility in health care research (efficient analysis of datasets, code generation, literature reviews, saving time to focus on experimental design, and drug discovery and development); (3) benefits in health care practice (streamlining the workflow, cost saving, documentation, personalized medicine, and improved health literacy); and (4) benefits in health care education including improved personalized learning and the focus on critical thinking and problem-based learning. Concerns regarding ChatGPT use were stated in 58/60 (96.7%) records including ethical, copyright, transparency, and legal issues, the risk of bias, plagiarism, lack of originality, inaccurate content with risk of hallucination, limited knowledge, incorrect citations, cybersecurity issues, and risk of infodemics. The promising applications of ChatGPT can induce paradigm shifts in health care education, research, and practice. However, the embrace of this AI chatbot should be conducted with extreme caution considering its potential limitations. As it currently stands, ChatGPT does not qualify to be listed as an author in scientific articles unless the ICMJE/COPE guidelines are revised or amended. An initiative involving all stakeholders in health care education, research, and practice is urgently needed. This will help to set a code of ethics to guide the responsible use of ChatGPT among other LLMs in health care and academia.

40Artificial Intelligence Can Generate Fraudulent but Authentic-Looking Scientific Medical Articles: Pandora’s Box Has Been OpenedOpenAlex

Martin Májovský, Martin Černý, Matěj Kasal, et al.

BACKGROUND: Artificial intelligence (AI) has advanced substantially in recent years, transforming many industries and improving the way people live and work. In scientific research, AI can enhance the quality and efficiency of data analysis and publication. However, AI has also opened up the possibility of generating high-quality fraudulent papers that are difficult to detect, raising important questions about the integrity of scientific research and the trustworthiness of published papers. OBJECTIVE: The aim of this study was to investigate the capabilities of current AI language models in generating high-quality fraudulent medical articles. We hypothesized that modern AI models can create highly convincing fraudulent papers that can easily deceive readers and even experienced researchers. METHODS: This proof-of-concept study used ChatGPT (Chat Generative Pre-trained Transformer) powered by the GPT-3 (Generative Pre-trained Transformer 3) language model to generate a fraudulent scientific article related to neurosurgery. GPT-3 is a large language model developed by OpenAI that uses deep learning algorithms to generate human-like text in response to prompts given by users. The model was trained on a massive corpus of text from the internet and is capable of generating high-quality text in a variety of languages and on various topics. The authors posed questions and prompts to the model and refined them iteratively as the model generated the responses. The goal was to create a completely fabricated article including the abstract, introduction, material and methods, discussion, references, charts, etc. Once the article was generated, it was reviewed for accuracy and coherence by experts in the fields of neurosurgery, psychiatry, and statistics and compared to existing similar articles. RESULTS: The study found that the AI language model can create a highly convincing fraudulent article that resembled a genuine scientific paper in terms of word usage, sentence structure, and overall composition. The AI-generated article included standard sections such as introduction, material and methods, results, and discussion, as well a data sheet. It consisted of 1992 words and 17 citations, and the whole process of article creation took approximately 1 hour without any special training of the human user. However, there were some concerns and specific mistakes identified in the generated article, specifically in the references. CONCLUSIONS: The study demonstrates the potential of current AI language models to generate completely fabricated scientific articles. Although the papers look sophisticated and seemingly flawless, expert readers may identify semantic inaccuracies and errors upon closer inspection. We highlight the need for increased vigilance and better detection methods to combat the potential misuse of AI in scientific research. At the same time, it is important to recognize the potential benefits of using AI language models in genuine scientific writing and research, such as manuscript preparation and language editing.

41Managing artificial intelligenceOpenAlex

Paul R. Krausman

By the time you finish reading this editorial about artificial intelligence (AI), it will be outdated; the AI field is growing beyond the imagination of many. One cannot even look at the news without finding something about AI: duplication of voices (even John Lennon's years after his death), use of AI in art and music creation, facial recognition, and others. This new field is readily used in the management of big data in many arenas but is filled with risks and uncertainty (Botes 2023). If the world population declines in number (scientificamerican.com/article/population-decline-will-change-the-world-for-the-better/, accessed 19 Jun 2023), there will be even more demand for AI to replace needed skills once provided by humans. The last part of the sentence is important because AI focuses on things typically done by humans. Artificial intelligence is a real game changer and needs to be managed wherever it is used, including the scientific process. The purpose of this editorial is to briefly indicate how The Journal of Wildlife Management (JWM) will incorporate AI in publishing. Artificial intelligence has been defined as an umbrella term for an array of algorithmic-based technologies, albeit with their own glitches, that solve complex tasks, which previously involved human thought, to whatever has not been done yet (DeWaard 2023). Artificial intelligence is living up to the promise of delivering value, influenced by advances in the availability of relevant data, computation, and algorithms from agriculture (Smith 2018) to zoology (Santangeli et al. 2020, Tuia et al. 2022) and everything in between. Artificial intelligence is also advancing wildlife management and conservation such as the use of drones in combination with AI to locate bird nests (Santangeli et al. 2020) and an array of other uses in wildlife and animal ecology as reviewed by Tuia et al. (2022). Artificial intelligence can also be used for nefarious activities such as Deepfake AI, a type of AI used to create convincing images, audio, and video hoaxes. To combat these uses, countries in The European Union are proposing laws (e.g., the AI Act) to put guardrails on the rapidly growing AI uses for policymakers and others. The United States is doing the same and other countries will not be far behind. Concerns over the use of AI is also growing elsewhere and universities are developing programs to train a new cadre of professionals to deal with the rapid changes and development in AI in all arenas. The rapid emergence and scope of AI generative tools, like chat generative pre-trained transformer (ChatGPT), Scite (an AI tool that provides quantitative and qualitative insight into how scientific publications cite each other), and DALL-E (an AI system named after S. Dali and Pixar's WALL-E movie) that can produce realistic images from text prompts, are at the forefront of many higher education conversations across the nation (e.g., University of Arizona) that are surveying faculty to identify salient topics surrounding emergent AI tools. And, there are a bunch of them. Scientists are surrounded with digital tools and platforms. As the professional academic social networks continue to grow, there have been changes in the publishing world almost as fast as the emergence of AI. For example, many journals are online with open access or are moving in that direction, social media is used to promote research, and with the advent of Covid, societies moved from in-person conferences to virtual conferences. All of these changes and others were assumed to operate with scientific integrity, which should continue with the emergence of AI. In publishing, as in other arenas, AI relates to the ability for machines to learn patterns to do things that have typically been done by humans. Artificial intelligence is the next major frontier in the process and dissemination of scientific knowledge and has arrived without much fanfare, but it is here to stay (Bachanan 2023). Artificial intelligence has been used for analysis and writing, among other parts of the scientific method by drawing on current knowledge and content available online with various combinations of accuracy, biases, and errors (Bachanan 2023). Like Bachanan (2023), I know that AI is here to stay but will have to wait to see how successful it is and how its uses will influence publishing. With the rapid growth of AI, we will not have to wait long. Artificial intelligence is touching all aspects of the publication world including the writing of articles (e.g., PaperPal, Writefull), article submission (e.g., Wiley's ReX, which automatically extracts data from manuscripts), tools to screen manuscripts on submission (e.g., Penelope, RipetaReview), support of peer review (e.g., SciScore) for method checking, and checking scientific images (Proofig, ImageTwin). Need more examples of the emerging role of AI in publishing? Check out Scite.ai to see how citations support arguments in manuscripts and the use of AI in marketing, creating proofs, copyediting, summarization (i.e., Scholarcy), using published material as data, reading (e.g., SemanticScholar summarizes manuscripts in 1–2 sentences), checking on similarities between submissions (e.g., TurnItIn, STM), determining if the scientific process is properly followed, finding referees, and detecting plagiarism (e.g., Content Authenticity Initiative; DeWaard 2023). These are just some of the uses of AI in publishing. There are also pitfalls that need to be addressed including scientific fraud and the potential legal risks for policies and decisions made with AI. One of the greatest challenges in using AI tools in the scientific process is to ensure the truth and validity of results from experiments. If truth is not grounded in scientific evidence or presented with sufficient qualifications, the scientific method would be violated. Artificial intelligence tools are built on probabilistic models that do not have rigorous closed-form mathematical solutions and therefore cannot be independently validated. Further, the algorithms are typically designed and trained to optimize a problem (i.e., to get the highest score possible on a certain task). As such, GenerativeAI models (e.g., ChatGPT) could fabricate scientific references in generating scientific papers to support a conclusion. The obvious danger is that if these references get past reviewers and accepted into the body of scientific literature, they could be referenced by scores of future authors to support a similar conclusion. Yet, at the core, the reference and its corresponding confusion may have never existed in the real world. Imagine how this could negatively influence the scientific process and its effects on the world and humanity. Similar errors could also creep into conclusions when AI is used to detect the number of animals in satellite imagery. Artificial intelligence could be also used to generate purely synthetic imagery of events that never occurred to promote some cause by actors who want to influence the scientific process. For example, what if an endangered species that never existed in an area was shown to exist in the middle of a development project in a peer-reviewed scientific study? Also, how does one address a machine or hold a program responsible for faulty information? And, by combining data from various places and individuals into a single data set, how is diversity of thought maintained? Authors are responsible for the ethical treatment of animals and their data, and must be aware of all aspects of how their data are collected and used. These and other issues must be addressed as AI is incorporated in publishing activities. Thus, the Editor-in-Chief and staff of JWM must embrace the responsibility of managing AI in our publications to maintain research integrity and ensure that AI does not replace the expertise and critical thinking of humans. To foster this idea, JWM staff follows Wiley's guidelines in dealing with AI: “Artificial Intelligence Generated Content (AIGC) tools—such as ChatGPT and others based on large language models (LLMs)—cannot be considered capable of initiating an original piece of research without direction by human authors. They also cannot be accountable for a published work or for research design, which is a generally held requirement of authorship …, nor do they have legal standing or the ability to hold or assign copyright. Therefore—in accordance with COPE's position statement on AI tools—these tools cannot fulfill the role of, nor be listed as, an author of an article. If an author has used this kind of tool to develop any portion of a manuscript, its use must be described, transparently and in detail, in the Methods or Acknowledgements section. The author is fully responsible for the accuracy of any information provided by the tool and for correctly referencing any supporting work on which that information depends. Tools that are used to improve spelling, grammar, and general editing are not included in the scope of these guidelines. The final decision about whether use of an AIGC tool is appropriate or permissible in the circumstances of a submitted manuscript or a published article lies with the journal's editor or other party responsible for the publication's editorial policy.” (https://authorservices.wiley.com/ethics-guidelines/index.html#5, accessed 14 Jun 2023). I will be incorporating similar language in updated publication guidelines to explain how AI-generated material should be treated in JWM. The bottom line is that AI cannot be an author, any use of AI must be acknowledged (except for tools used to improve spelling, grammar, and general editing), and author(s) are responsible for all information in their manuscripts including data generated by AI. As AI develops and is used more in the scientific method and publication process, the policies we follow will also evolve to maintain the foundation of human thought, ethics, and the scientific method in published work. The Wildlife Society Code of Ethics states that members should understand “…human society's proper relationship with natural resources, and in particular for determining the role in wildlife in satisfying human needs and addressing the management of wildlife-related impacts” (https://wildlife.org/wp-content/uploads/2017/07/Code-of-Ethics-May-2017.pdf, accessed 28 Jun 2023). We should all keep abreast of technological advances, including AI, and understand the roles they play in managing our wildlife resources. Keep on publishing but keep AI off the author list. This editorial was improved with reviews from J. A. Bissonette, A. S. Cox, E. H. Merrill, K. A. Norris, and P. M. Wegner. Many thanks.

49Practical Considerations and Ethical Implications of Using Artificial Intelligence in Writing Scientific ManuscriptsOpenAlex

Muhammad Nadeem Yousaf

The growing accessibility and sophistication of artificial intelligence (AI) tools have transformed many areas of research, including scientific writing. AI tools, such as natural language processing models and machine learning-based writing assistants, are increasingly used to help draft, edit, and refine scientific manuscripts. However, the use of AI in the writing process introduces both legal and ethical challenges. Various guidelines and policies have emerged, particularly from academic publishers, aimed at ensuring transparency and maintaining the integrity of scientific work. This editorial aimed to provide guidance for researchers on the ethical and practical considerations regarding the use of AI in writing scientific manuscripts, focusing on institutional policies, authorship accountability, intellectual property concerns, plagiarism issues, and image integrity. INSTITUTIONAL AND JOURNAL POLICIES: TRANSPARENCY AND GUIDELINES As AI tools become more prevalent in scientific writing, academic institutions and journals are establishing clear guidelines regarding their use. The majority of well-known publishers provide guidance to authors, readers, reviewers, and editors concerning the role of AI-assisted technologies in the writing process, but industrywide standards have not yet been fully refined.1 The recently developed framework by American College of Gastroenterology (ACG) and Wolters Kluwer, the publisher of ACG journals emphasizes transparency and accountability, highlighting that although AI can assist in improving the readability and language of a manuscript, it should not replace the core tasks of authorship, such as generating scientific insights or drawing conclusions.2 ACG also requires authors to disclose the use of generative AI and AI-assisted technologies during the writing process. This disclosure fosters trust and ensures compliance with the terms of use of the AI tools. The goal is to provide clarity to readers, reviewers, and editors, helping them understand where and how AI was applied in manuscript preparation. Failure to disclose AI use can lead to ethical breaches, retractions, and potential damage to the professional reputations of both the researcher and the journal. The policy only covers the creation of new content and expressly forbids using AI on previously published material, preventing concerns around self-plagiarism or unauthorized content modification. AUTHORSHIP AND ACCOUNTABILITY: THE ROLE OF HUMAN OVERSIGHT The rise of AI in scientific writing raises fundamental questions about authorship and responsibility. The ACG's authors instructions emphasize that authorship cannot be attributed to AI, as AI cannot take responsibility for the accuracy or integrity of scientific work. Authorship implies a set of ethical and intellectual responsibilities that only human researchers can fulfill. Every listed author must be accountable for the content of the manuscript, and AI cannot be assigned such responsibilities. Moreover, although AI can assist in writing, it cannot generate the intellectual contributions to scientific research. Scientific manuscripts must be the product of human insight and critical thinking. ACG's policy highlights the need for human oversight in the application of AI technologies, emphasizing that all AI-generated content should be thoroughly reviewed and edited by the authors to ensure accuracy, completeness, and lack of bias. The authors are ultimately responsible for ensuring that their work adheres to the highest standards of scientific integrity. This policy aligns with broader academic norms, such as the International Committee of Medical Journal Editors criteria for authorship, which require authors to have made significant intellectual contributions to the research and to be accountable for the final work. Failure to ensure human oversight and control can lead to inaccurate or misleading scientific conclusions, jeopardizing the validity of the research. INTELLECTUAL PROPERTY AND OWNERSHIP OF AI-GENERATED CONTENT The use of AI in scientific writing introduces important questions about intellectual property (IP) ownership. Scientific manuscripts often contain novel ideas and discoveries, and researchers must be careful about the terms and conditions associated with the AI tools they use. Some AI platforms, such as OpenAI's Chat-GPT models, explicitly state that users retain ownership of the content generated through the tool. However, other platforms may have different terms, leading to potential conflicts over content ownership. ACG's policy ensures that authors remain responsible for the originality of their work, cautioning against the use of AI tools in ways that might lead to copyright violations or IP disputes. Because AI models are trained on vast data sets that may include copyrighted material, authors must be vigilant in ensuring that the AI-generated text does not unintentionally replicate existing works without proper attribution. This highlights the importance of understanding the terms of service of the AI platform being used and adhering to proper citation practices. PLAGIARISM AND ETHICAL CONCERNS Plagiarism is a serious ethical violation in academia, and the use of AI tools presents new challenges in this area. AI systems generate text based on large data sets, and although they aim to produce original content, there is always the risk of unintentional plagiarism if the AI-generated text closely resembles existing works.3 Academic journals including all of ACG's journals use plagiarism detection software to monitor submissions, and any AI-generated content that overlaps significantly with previously published works could be flagged as plagiarism. To address this, the authors should practice transparency and proper attribution when AI tools are used. The authors must disclose AI use to avoid accusations of misconduct and to ensure the originality of their work. In addition, because AI can generate authoritative-sounding but incorrect or biased information, researchers must carefully review the content produced by these tools, ensuring that it meets academic standards. THE GENERATION OF FABRICATED CITATIONS AND REFERENCES USING AI-ASSISTED TOOLS One of the most pressing concerns in AI-assisted academic writing is the potential for generating fake citations and references4,5. This problem arises when AI tools produce fictitious or incorrect references that appear authentic but do not correspond to real sources. These generated references often mimic legitimate academic citations, complete with plausible journal titles, author names, and publication dates. However, on closer inspection, the cited works may not exist, or the citation details may be inaccurate, leading to false academic claims. The generation of fake citations by AI tools severely undermines the integrity of the peer-reviewed scientific process. Scholarly research depends on verifiable sources and accurate references that allow readers and reviewers to trace the intellectual lineage of ideas and verify the reliability of the claims being made. When citations are fabricated, this essential foundation of research collapses, leading to a cascade of misinformation. For instance, if one published article contains fake references and is later cited by others, the spread of false information becomes harder to detect and control, potentially polluting the knowledge base of a given field. The dangers of AI-generating fake citations extend beyond simple inaccuracies. Because AI models rely on large-scale data sets to produce content, they may unintentionally fabricate references by amalgamating parts of real sources with incorrect or fictional details. This creates the appearance of a legitimate scholarly foundation, making it difficult for readers and reviewers to identify problematic citations without extensive fact-checking. Researchers using AI tools must remain vigilant and ensure that all citations are factually accurate and correspond to real, verifiable sources. The academic community operates on a system of mutual trust, and violations such as the intentional use of fake citations can cause irreparable harm to the credibility of individual researchers and the broader field. PRACTICAL APPLICATION OF AI TO STREAMLINE RESEARCH WHILE PRESERVING ETHICAL STANDARDS AI is transforming research, offering tools that assist with a variety of tasks, from systematic reviews to advanced data analysis.6 These AI tools have the potential to streamline many stages of the systematic review process, including developing and refining search strategies, screening titles and abstracts based on inclusion or exclusion criteria, extracting essential data from studies, and summarizing findings.6 AI can also efficiently scan vast databases to identify relevant articles and automatically organize metadata from diverse sources. Machine learning algorithms extend these capabilities by uncovering hidden patterns, trends, and correlations in complex datasets, enabling predictive analytics and intuitive visualizations that enrich research insights. However, these tools should not be seen as replacements for human expertise and judgment. Quality and ethical risks, such as biases in training data or inaccuracies in results, remain critical concerns. Researchers must view AI as a supplementary tool that optimizes processes while preserving the importance of critical quality checks, human evaluation, and meaningful intellectual contributions. Human oversight remains indispensable for interpreting findings, evaluating ethical implications, and ensuring the overall integrity of scientific reports. By balancing AI's efficiency with the irreplaceable depth of human analysis, researchers can leverage these technologies to enhance workflows, maintain ethical rigor, and drive meaningful scientific progress. The integration of AI tools into the writing of scientific manuscripts presents both practical benefits and ethical challenges. Although AI can enhance readability and streamline the drafting process, it cannot replace the essential human contributions that define authorship, intellectual insight, and scientific integrity. Clear institutional and publisher policies should emphasize the need for transparency, accountability, and human oversight in the use of AI tools. The generation of fabricated citations and references, IP concerns, and risks related to image integrity highlight the ethical complexities posed by AI. To safeguard the credibility and trustworthiness of academic research, it is imperative that researchers carefully manage AI usage, ensure proper attribution, and maintain the originality of their work. As AI technology evolves, ongoing vigilance, refined policies, and adherence to ethical standards will be crucial in maintaining the integrity of scientific literature. DISCLOSURES Author contributions: MN Yousaf wrote and edited the manuscript and is the article guarantor. Acknowledgments: Neen LeMaster, Assistant Managing Editor of ACG Scholarly Publications. Neen helped in refining ACG policy in the editorial from managing standpoint. Financial disclosure: None to report. Informed consent was obtained for this case report.

50Dignity of science and the use of ChatGPT as a co-authorOpenAlex

Manuel Scimeca, Rita Bonfiglio

‘Science is a reflection of human dignity. It represents the pursuit of truth, the exploration of the nature of the world, and the comprehension of life itself. Science is the beacon that leads us towards a brighter future,’ stated Albert Einstein in the 20th century. Despite this, the question of how science can enhance human life while preserving the human dignity remains a topic of discussion. There can be no denying that technological advancements have greatly improved life on earth, with space exploration, personalized medicine, new drugs, and computing having a significant impact on the lives of people. For example, messenger RNA (mRNA) vaccines have played a crucial role in fighting the current coronavirus disease 2019 (COVID-19) pandemic.1Tregoning J.S. Flight K.E. Higham S.L. Wang Z. Pierce B.F. Progress of the COVID-19 vaccine effort: viruses, vaccines and variants versus efficacy, effectiveness and escape.Nat Rev Immunol. 2021; 21: 626-636Crossref PubMed Scopus (494) Google Scholar,2Fernandes Q. Inchakalody V.P. Merhi M. et al.Emerging COVID-19 variants and their impact on SARS-CoV-2 diagnosis, therapeutics and vaccines.Ann Med. 2022; 54: 524-540Crossref PubMed Scopus (87) Google Scholar Similarly, artificial intelligence (AI) has the potential to revolutionize scientific knowledge and improve the quality of human life, much like the major scientific breakthroughs of the 20th century, including the discovery of DNA, the moon landing, the Theory of Relativity, and quantum mechanics. AI is likely to have a major impact on health care, sustainability, climate change, and environmental issues. Ideally, and through the use of sophisticated AI-based devices, cities may become less congested, less polluted, and generally more livable, and health care systems may also improve. Thus it is the responsibility of the scientific community to invest significant human and economic resources into the development of AI science. Nevertheless, the advancement of AI in science poses significant ethical dilemmas.3Kargl M. Plass M. Müller H. A literature review on ethics for AI in biomedical research and biobanking.Yearb Med Inform. 2022; 31: 152-160Crossref Scopus (2) Google Scholar, 4Schuklenk U. On the ethics of AI ethics.Bioethics. 2020; 34: 146-147Crossref Scopus (2) Google Scholar, 5Kazim E. Koshiyama A.S. A high-level overview of AI ethics.Patterns (N Y). 2021; 2100314Google Scholar Some of these include questions about accountability for AI’s errors, the impact of machines on human interactions, and how to address unintended consequences of AI. There is an urgent need for a ‘behavioral code’ to ensure that AI is used in a humane manner. Therefore scientists, doctors, and philosophers are developing principles of ‘algo-ethics’, which guide the ethical use of AI based on the principles of transparency, inclusion, responsibility, impartiality, reliability, security, and privacy, so that AI remains in service of people. Technology often advances faster than the scientific community’s ability to evaluate it. Before determining which applications of AI bring benefits and which pose risks, it is essential to understand the potential and challenges of new technologies. This creates a ‘gray zone’, in which researchers face ethical challenges without adequate training or support. In recent days, there has been a noteworthy discussion about the utilization of ChatGPT and other AI tools as coauthors in scientific papers.6Zhavoronkov A. ChatGPT Generative Pre-trained TransformerRapamycin in the context of Pascal’s Wager: generative pre-trained transformer perspective.Oncoscience. 2022; 9: 82-84Crossref PubMed Google Scholar, 7Curtis N. ChatGPT To ChatGPT or not to ChatGPT? The impact of artificial intelligence on academic publishing.Pediatr Infect Dis J. 2023; 42: 275Crossref Scopus (13) Google Scholar, 8King M.R. chatGPT A conversation on artificial intelligence, chatbots, and plagiarism in higher education.Cell Mol Bioeng. 2023; 16: 1-2Crossref Scopus (44) Google Scholar ChatGPT (Chat Generative Pre-trained Transformer) is a large language model developed by OpenAI (San Francisco, CA). It is built on the GPT-3 language model family and has been fine-tuned using both supervised and reinforcement learning methods.9The Lancet Digital Health. ChatGPT: friend or foe? Lancet Digit Health. 2023:S2589-7500(23)00023-7.Google Scholar In scientific research, it can be considered a useful tool for producing well-written, sometimes comically absurd, mini essays in response to requests. It can also be used for composing short computer programs, conducting literature research, analyzing data for statistical purposes, and detecting plagiarism and mistakes in scientific texts. Some authors have expressed their opposition to the use of ChatGPT or similar AI tools as coauthors of scientific publications due to their inability to meet editorial standards.10Thorp H.H. ChatGPT is fun, but not an author.Science. 2023; 379: 313Crossref PubMed Scopus (130) Google Scholar,11Stokel-Walker C. ChatGPT listed as author on research papers: many scientists disapprove.Nature. 2023; 613: 620-621Crossref PubMed Scopus (112) Google Scholar Nonsentient devices such as ChatGPT cannot take responsibility for the content and integrity of scientific papers or give consent to terms of use and distribution rights. Here, we want to emphasize the potential negative consequences of using ChatGPT as an author on both the scientific community and humanity as a whole. Concerning the scientific community, the use of ChatGPT as a coauthor in a scientific publication raises serious concerns. Equating the dignity of researchers with that of a machine is unacceptable. According to the United Nations (UN), human dignity is defined as the intrinsic and inalienable value of every person, and is recognized as a universal and inalienable right that must be respected and protected by all societies and institutions. However, it is clear that machines, such as ChatGPT, cannot possess dignity. By acknowledging ChatGPT as a coauthor, we would be denying authors their dignity as humans and scientists. Thus, by losing our dignity, will we have the strength to defend the good? Will it make sense to promulgate what is right and stigmatize what is wrong? Will we have the authority to defend science from conspiracists and fake news? These three questions prompt us to examine the potential impact of using ChatGPT as a coauthor on human beings. It would be important to assess how the public might react if they were to learn that an AI device was one of the authors of a scientific discovery that impacts their lives. In the field of biomedical research, patients may ask the AI device for a diagnosis or treatment instead of a doctor, because both the physician and the AI device could have coauthored the same study. If this were the case, could we reject our nonsentient coauthor? This would be a challenge because we should explain an undeniable reality that we ourselves are denying, the superiority and complexity of human thought compared with that of machines. In our opinion, these and other considerations should alarm the scientific community. Researchers and publishers should promptly preclude the use of ChatGPT as an author for scientific publication, thus providing the necessary time to discuss about this issue. We have the responsibility to drive technological progress by both preserving the dignity of the science and pursuing the good of humanity. PS: We used ChatGPT to detect possible plagiarisms and mistakes. It is an exceptional tool, certainly not an author. None declared.

51Addressing student use of generative AI in schools and universities through academic integrity reportingOpenAlex

Steven A. Peterson

The aim of this theoretical article is to explore frameworks contributing to reasonable and applicable definitions of artificial intelligence (AI), human intelligence, and generative artificial intelligence (GenAI), as well as propose a framework for an efficient, transparent, and scalable approach to assess and address inappropriate use of GenAI resources in an academic environment.The integration of artificial intelligence technologies in academic environments has transformed how students engage with learning materials, complete assignments, and demonstrate formative knowledge. The opportunities AI-enhanced resources offer for personalized learning and robust academic experiences are comingled with significant challenges related to appropriate student use. Although automated content generators can enhance productivity and understanding, they also raise complex questions about academic integrity and the role educators must play in shaping its ethical use. AI-powered tools provide students with grammar correction, citation management, language translation, research abstracts, and a wide-range of enhanced learning support when used appropriately (Dwivendi et al., 2021). These tools also assist non-native speaking students with accessibility to learning modalities throughout the world (Karakas, 2023). It is when these tools are used to circumvent academic effort that concerns quickly emerge. Large language models (LLMs) capable of generating human-quality text and creative content have blurred the lines of authorship and academic integrity. Students have unprecedented access to resources capable of completing assignments, writing essays, and solving complex problems with minimal personal effort. This raises strong opinion about the assessment of student learning (Luo, 2024), determination of plagiarism (Bittle et al., 2025), and efficacy of performance outcomes (Weng et al., 2024).Luo's critical policy analysis published in 2024 examined the institutional frameworks guiding the use of GenAI in assessment at twenty world-leading universities. Employing Bacchi's "What's the problem represented to be" (WPR) methodology, the research sought to critically analyze how these institutions articulate the challenges posed by GenAI within an evolving academic landscape.The core critique identified a dominant, nearly universal policy paradigm that frames GenAI as a potential threat to academic integrity and the intellectual originality of student submissions. By designating these resources as a form of external assistance separate from the student's contribution, this prevailing policy structure suggests a critical silence regarding the increasingly distributed and collaborative nature of modern, technology-mediated knowledge production (Luo, 2024). This research suggests a redefinition of "originality" to effectively accommodate and integrate human-AI collaborative endeavors.The potential effects of these prevailing integrity-focused policies are significant: they risk stigmatizing students utilizing GenAI resources for legitimate purposes and may transform faculty into punitive "gatekeepers" whose focus is policing misconduct rather than empowering robust student learning experiences (Luo, 2024). Notwithstanding its theoretical utility, Luo's study exhibits key research gaps. A primary limitation is the absence of empirical data concerning the lived effects of the policies, specifically lacking both student perspectives on their resultant impact and real-world faculty implementation data across varied disciplines. This reliance on a limited, purposed sample of elite universities undermines the generalizability of findings, neglecting distinct policy challenges faced by diverse institutional types, such as community colleges, proprietary and non-profit institutions. Finally, a key absence is the inclusion of alternative policy models, specifically in-depth case studies on institutions that have successfully moved beyond the punitive model to implement progressive, integrated GenAI policies.A systematic literature review was conducted to effectively assess how GenAI technologies influence the demonstration of formative knowledge, balancing both educational benefits and associated risks to academic honesty (Bittle et al., 2025). This comprehensive analysis encompassed 41 studies gathered from key databases, including the IEEE Xplore and JSTOR.Core findings suggest a profound impact of GenAI resources in higher education environments. The opportunities AI-enhanced resources offer for customized learning and robust educational engagement are comingled with significant challenges related to appropriate student use. Large language models capable of generating human-quality assignments and creative content have blurred the lines of authorship, enabling the evasion of conventional plagiarism tools and raising complex questions about academic integrity. This review suggests that immediate actions are required to effectively manage GenAI's influence. This primarily involves enhancing digital literacy among both students and faculty, and developing more robust detection tools that can assess and address AI-generated content. A goal of this review was to guide future efforts in developing evidence-based practices for the responsible integration of these transformative tools.The current body of work lacks diversity in study design and scope, often exhibiting a bias toward studies originating in Western contexts, and potentially missing perspectives from social science (Bittle et al., 2025). is a comprehensive analysis that the of GenAI its potential for academic the lacks empirical data regarding the efficacy of studies that the impact of policies on student knowledge and the of detection tools as GenAI models 2024 review conducted by et how educators are student learning in environments GenAI is a identified learning outcomes that have and the research in this evolving academic This analysis and studies a The findings concerning assessment identified primary is often in the of GenAI educators are and assessment such as and The model involves AI-enhanced resources are a and of the (Weng et al., integration of artificial intelligence technologies has the of and learning are increasingly on and learning including critical and digital to a technology-mediated current research that the of studies this significant in the research identified a critical is a for studies to beyond work and provide empirical concerning the impact of GenAI on student the literature lacks research on the efficacy of assessment and how to effectively integrate assessment into a the outcomes are a in of the assessment and these work to and for these (Weng et al., 2024). beyond to and of intellectual core ethical 2023). such as academic regarding intellectual and on work as the is on critical to for writing in This a assessment citation from from a punitive model to a approach a approach 2023). on and academic in citation and an integrity future social work to ethical in academic and of a of academic a strong of plagiarism et al., 2021). The study also plagiarism detection detection tools are used and for the use of these tools a in within the research community et al., key is the plagiarism is and in their and work et al., 2021). The propose that plagiarism effectively institutions to beyond detection tools and they a focus on educational These must the ethical of originality and in and et al., 2021). This is for knowledge into responsible research from to the is The et review literature to primary for student This systematic review of the on with a from a of significant is the accessibility of by tools content The is of a critical in academic and citation and knowledge in research writing and the of a also policy these are into and often an for in design This is among and the is in research and must integrity et al., of a of to in institutional and risk This often from for faculty may This the study empirical findings into integrity rather than on on citation and ethical from research to assignments and analysis These design the of by and policy developing and for faculty as this and and a for academic must and to and provide a et al., from digital access and of is often as studies suggest from A research ethical to and research at of plagiarism detection citation knowledge, and A key was that the of students have strong of plagiarism and its This suggests a in efforts have data a critical in research and writing 2021). The primary is must from and punitive to that The study that into The for plagiarism was research primary risk is a to and 2021). This the that plagiarism is primarily a is a Although research as the is that institutional policy from a punitive policing model to must integrated into the core an The goal to with the and to and knowledge real-world are key 2021). on academic and This must by risk from language by language support to on text and a distinct the role as the for on citation detection and developing review and formative on to and as a rather than and studies to the efficacy of The key for is a in the of student by ethical that focus on to address the a to knowledge theoretical article a research methodology, institutional policy and a framework to and of in the use of GenAI by students within the of academic integrity in higher The approach is to current research and institutional policies also how these academic environments and academic in a comprehensive review of academic policy and the of technologies in educational the aim is to explore frameworks contributing to reasonable and applicable definitions of artificial intelligence (AI), human intelligence, and generative artificial intelligence (GenAI), as well as propose a framework for an efficient, transparent, and scalable approach to assess and address inappropriate use of GenAI resources in an academic analysis was conducted of academic integrity policies from a of institutions. This analysis to and institutional related to such as of and the integration of artificial intelligence tools in academic this research is of engagement with academic integrity and twenty as a and academic across institutions. This work the and of academic integrity policies, the of plagiarism and in design that and this framework definitions that articulate and the of human intelligence, artificial intelligence, and generative artificial intelligence These definitions are used to critically assess how institutions are to the and ethical challenges by GenAI this approach of literature policy and this article to a of academic integrity with for policy of the studies to the of in the human in by and a model was used to demonstrate how and This the for in research and and and within the of intelligence was as the of is in the of This with to complex by for in The was a of These to for the of of data and for and This was the of learning 2025). learning language models (LLMs) that to problems on and definitions propose such as integration and of and external et al., 2025), to and on of human intelligence et al., knowledge and learning et al., to actions that human et al., that and 2025), and capable of for within et al., the of this article a of artificial intelligence, by research and models and identified by a that and in such a that human intelligence by and to across human et al., et al., et al., et al., et al., intelligence as to to and to of intelligence the framework the of universal and knowledge from of are primary contributing to of human intelligence and in the of of of frames as a of and from the and learning and for intellectual of the of of intelligence are to the and of intelligence as knowledge of of in and The of the with of such as of problem solving and the to knowledge, and to the of this article a of human intelligence, by and of a academic enabling the to from to and and use knowledge to their and that generative artificial intelligence from of artificial intelligence is its more engagement et al., 2025). that on such as data GenAI unprecedented across a of from creative and The of this with the to content and to a in the of technologies et al., 2025). GenAI has an to to and in that more human GenAI and from a data utilizing and proprietary to content at the of a human et al., the of has that separate from of a approach to language on and an generating language model et al., and from to its approach et al., the of this article a of generative artificial intelligence, from the literature and in for a of artificial intelligence to content on from and models and in et al., et al., et al., et al., the and these of intelligence is an when academic and of on a of the literature and key identified the a of of the in a comprehensive review of academic policy and the of technologies in educational the aim is to explore frameworks contributing to reasonable and applicable definitions of artificial intelligence (AI), human intelligence, and generative artificial intelligence (GenAI), as well as propose a framework for an efficient, transparent, and scalable approach to assess and address inappropriate use of GenAI resources in an academic analysis was conducted of academic integrity policies from a of institutions. This analysis to and institutional related to such as of and the integration of artificial intelligence tools in academic this research is of engagement with academic integrity and twenty as a and academic across institutions. This work the and of academic integrity policies, the of plagiarism and in design that and this framework definitions that articulate and the of human intelligence, artificial intelligence, and generative artificial intelligence These definitions are used to critically assess how institutions are to the and ethical challenges by GenAI this approach of literature policy and this article to a of academic integrity with for policy The potential for in learning can when is on and their AI-generated bias can and both theoretical and applicable and 2023). This is also by that article are are essays, of generative text tools on and of the data This is by of that as inappropriate use of and and and of writing as a to as a is text tools also to conventional human and such can a of such as and use of also as well as an of and of throughout the language models (LLMs) have a to and 2024). This often in and the the are in missing has of the and et al., 2024). A is to the of the a a when into the the a the is the a that is from the by the student's and with and with support this of a was into a of intelligence with and was into The by GenAI they in and of the GenAI was to a of the of of a potential of the academic integrity policy is an demonstration of plagiarism academic This is an plagiarism an that the of from the student with in writing this has to the student's and they have the to provide an and the is within the student's work is to the personal than that of the the student has the for for a and the has This to the institutional the of academic and a determination to the appropriate the faculty to provide a a as review of the academic integrity policies of on an universities and proprietary of was conducted to the of the institutional policies regarding the use of GenAI by its to proprietary policies and the institutions The universities and the and institutions from of the and universities and from and their academic policies to and universities a academic policy in with regarding student use of and of the academic integrity policies in their academic specifically related to inappropriate use of The policies in the used the of of the universities specifically that and of the may by This was with the The and policy across institutions was that missing in with GenAI policies that content is AI-generated that to students review content to and of of the student's and represented the that and from must and to policies faculty to and for use in the institutional policies that questions regarding use of GenAI to policy with the appropriate to is to students about AI-powered the ethical data authorship and appropriate use in academic This how to critically AI-generated the of academic integrity in the digital and use tools and as learning rather than for their students to the of and to academic integrity 2025). A to students on the faculty are in to and assess inappropriate integration of automated content suggests that on artificial intelligence and practices students to integrate GenAI tools into their 2025). study this a literacy to students in that this approach content and of study also among with educational et al., is the of the of as of content the they are plagiarism of in and This the as to the use of AI-powered tools in that assessment the determination of a learning a must to the of It is to that the of an as are and to and as such is for of plagiarism within the of both educational and and specifically to the academic is as an to work that from is to provide appropriate to the of the is to appropriate regarding as that the a can a with the that future This is learning a problem and The appropriate are and the is This an for required for an of plagiarism are and of is must the academic is key academic and writing in the and the of the is an ethical in writing is a from a of primarily is as a of core ethical 2023). It beyond This more such as plagiarism involves from a and the of intellectual 2023). The is the to work as educators a for distinct that suggest the student is to the use of external These writing in the of 2023). that for such can academic regarding intellectual the focus on the ethical of originality et al., 2021). for often this ethical et al., a writing is by by a to and 2021). The primary risk is a This from that a of contributing is of and citation a in regarding academic et al., 2025). This often an for in design rather than It is among such as and students may have et al., 2025). contributing to this on knowledge of research and et al., to address writing is punitive the is in research and institutions must integrity et al., 2025). must focus on in and to transform knowledge into responsible research practices et al., academic review concerns of plagiarism and the of their can to the and student use of intelligence and digital must to the of with and of and to a of the academic integrity a of and for students as an is of to and of the learning a student's academic in of an when and have and the student has a of to a and the of in a with the student about and concerns within the and for their on This is the in the absence of this is to the of the is appropriate the to was absence of a These are faculty can guide the student and are academic integrity is appropriate when of are in the student's and this is a on the student the determination can to to an of a student an faculty a of the and as an for determination of plagiarism has This is the role of the academic review is and at the from that the student has has of academic is to for This is for the that with the student from the appropriate institutional use such for on on have on how to address have an to the and the is that these concerns for the appropriate It is that the has to the of the student in a with a for their the student and of to and the is identified and is a to to the academic review as a and for The is to demonstrate the faculty in potential academic integrity the faculty in this as an and of successfully within the and that the student has the to from this with the the that future plagiarism concerns within the is of and as such for to a review for of the efforts have of a student to an at a of plagiarism is by an This to the student in a and A of to this is for institutional academic review and with a to a of for the with the that to the on the of the and to the faculty This that a has identified and is review the of a of student academic performance and academic integrity concerns in review as has in the artificial intelligence increasingly in educational its potential to enhance learning and the of academic integrity must and with The integration of in is a is an and ethical that challenges about authorship, and and universities from a transparent, and approach to and student use of intelligence and digital in their institutions. intelligence is a to enhance personalized educational and of content. The critical to is its use is academic student and the of academic are in the of of offer a comprehensive framework for institutions to address this evolving by policies that are and also By from punitive models to institutions can academic also student education tools to and the for ethical engagement with and that and rather than and for and integration of in education in in must with and the and to students to use tools and to and policies related to GenAI use of a robust and scalable academic integrity institutions from an students to provide more and language and models and data language and models use of to often in absence of critical analysis and 2024). an institutional are that can that are to intelligence and opportunities for student of artificial intelligence such as more learning critical integrate real-world and a robust that an empirical study the efficacy of to enhance student writing and the critical required for of student learning research a to student performance on distinct writing assignments and of implementation The integration of the on successfully transformed how students with the a significant in This key that the of the successfully critical as by higher performance The suggests that when students are with they can effectively and for assignments that methodology, is with significant research challenges and gaps. an the generalizability of the findings to academic environments is the research lacks the data to a the of critical with assessment the study the regarding the address the missing and student a framework for future involves the integration of writing It is when these are used in with that a assessment of the learning and the of critical can beyond a focus on the students to analyze their experiences and learning The key that students in the writing their critical from to the the writing significant in the critical of analysis and Although in a for future design critical and the faculty to raising complex questions about It is when these assignments are used to about complex that concerns quickly regarding the the research must data text analysis to how the writing and enhance and these are across academic articulate a and real-world of aim of a work by and was to explore frameworks contributing to reasonable and applicable models for as well as propose for an efficient, transparent, and scalable approach to with the goal of community impact and social The integration of into academic environments has transformed how students engage with learning materials, complete assignments, and demonstrate formative knowledge by collaborative with The opportunities for robust academic experiences and social challenges are comingled with significant challenges related to its the case studies and such as the and on can enhance and community understanding, they also raise complex questions about its and The literature faculty and students a of enhanced support for into diverse from to It is when these are used to student that concerns quickly about the for the The a core in the of studies capable of community for is a research concerning the complex and and in It was that future work focus on systematic of faculty to this that institutional support for community engagement is with its social integration of into academic and environments has transformed how research is both and of knowledge et al., The opportunities for generating knowledge that is both and are comingled with significant challenges related to and Although its methodology, for a can enhance community also raises complex questions about its and the role must play in shaping a research conducted with a of core and a wide-range of enhanced case studies It is when the of of and community is used critical on that concerns quickly emerge. The article suggests that is a for complex social and to the is by is on how institutional policies can address the the nature of and the of academic a significant in developing and impact capable of the of social across diverse et al., critical and intellectual integration of assessment in academic environments has the potential to transform how students engage with learning materials, complete assignments, and demonstrate formative knowledge et al., as by an empirical study is to explore a framework contributing to an and applicable assessment for as well as propose an approach students are in their questions within a The opportunities this learning for enhanced critical and learning are comingled with significant challenges related to generalizability and implementation the core of questions students to analyze and can enhance performance and also raises questions about the at It is when the research on the that concerns about the of the study suggests a critical is the for an efficient, transparent, and scalable approach to assess the of the questions the of these questions to a with grammar et al., 2023). the nature of the case study the lines of future work across diverse educational to the integration of in academic within an has transformed how students engage with primary complete assignments, and demonstrate formative knowledge. The opportunities offer for learning and robust academic experiences are comingled with significant challenges related to the of learning Although the writing to a case and can enhance knowledge and also raises concerns about the of study data the efficacy of the with student performance and et al., 2021). It is when these are used to learning that concerns quickly about the the The research that writing enhanced student the of the the benefits of significant in the for a and scalable approach to the the learning from the case the from completing a The nature of the content the lines of generalizability to empirical across diverse academic and and of and an empirical study frameworks contributing to an and scalable approach to collaborative learning to enhance both and The integration of technologies in academic specifically for and has transformed how students engage with collaborative and complete complex assignments 2025). The opportunities AI-enhanced resources offer for personalized learning and robust academic experiences significant challenges related to generalizability and core on and data for students with an enhanced learning support It is when these complex on that concerns about ethical and A significant in the for an efficient, transparent, and scalable approach to and how this data is and by the The 2025). Finally, the study complex questions about the and the required for to provide a critical analysis for institutions this integration of assessment and in academic environments has transformed how students engage with learning and demonstrate formative knowledge. Although the assessment its both and a bias when to the as well as challenges related to and analysis conducted by and educators with data on the that the of is students the for and assessment that questions about the to their The research that the of the assessment a strong with student the of the a significant in the for an efficient, transparent, and scalable approach to developing that can the the of the The study complex questions about the the and the generalizability of these beyond this assignments on personal and personal theoretical article by et frameworks contributing to

53The future (Industry 4.0) is closer than we think. Will it also be ethical?OpenAlex

Pavel Fobel, Aleksandra Kuzior

We live in an era when we are significantly confronted with new social trends which affect the living environment, sustainable life, migration processes, global social changes, and economic innovations, as well as new technologies and more efficient use of artificial intelligence. We perceive the future not only as a scientific and technological challenge, but also as an issue of ethical importance and potential ethical risks. Therefore the civilisation changes, such as the adaptation to the parameters of a new society under Industry 4.0, call for ethical caution and moral sensibility in order to reduce, or even eliminate, potential negative impacts on humans and their existential conditions. The use of robotics and artificial intelligence in various areas, such as in the field of law, education, construction of smart cities or public administration, gene technologies, housing, productivity, social services, industry, and the like also poses a new requirement. For example, the demand will grow for creative people, professionals having understanding for a human in a new environment, in everyday contact with artificial intelligence, new functioning of institutions, business transformation, and the entire social systems. They should be individuals able to response, in a creative manner, to specific situations and needs, new forms of self-realization. There will be a radical change in the area of personalisation, this including both the education and the personalised and individualised service, responses to individual requirements of a citizen, client, or consumer. Modern era was based on mass production and social revolutions. The social changes and shifts in values are mirrored by intellectual authorities, such as G. Lipovetsky who emphasises the necessity to adopt new ethical approach in relation to the new social paradigm. "Postmodern age is obsessed with information and self-expression" (Lipovetsky, 1998, p. 19). Individual ethics will gain its momentum, the ethics of other (third) type will be necessary, e.g. the ethics as presented by G. Lipovetsky: "Our era does not restore the rule of the "good old morality", it abandons it. In this sense, it is not possible to expect any changes of laws, any exploration of new moral values. Its ambition is to participate in solutions and changes, implement mechanisms of ethical prevalence, or the instruments of ethical supervision within social plans being prepared or already implemented.Its mission is to enrich the dialogue in the area of new trends with ethical questions, to extend the interdisciplinary discourse, to enter the dialogue within innovative projects and be an active player in such dialogues. Each change affects human and human’s integrity. Underestimation of professional opinions of ethical nature may generate an irreversible or dangerous situation which could put humans under threat. Solution of consequences without setting responsibilities, assessment of ethical risks may lead to serious social issues and delayed responses which would rather stem from searching of conscience.What is more, it needs to be emphasised that new technologies are the outcome of scientific production, implementation of science, and are associated with the activities of research teams. And this is the aspect that is pointed out by the representatives of the Technology Assessment concept1, the importance of applying ethical criteria to technology assessment. The significance and risks of contemporary science are also addressed by Ulrich Beck in his Risk Society. It is evident that the issue deserves wide interdisciplinary discourse across all areas aiming to overcome particularised approaches to understanding and solutions of dilemmas. In this interdisciplinary discourse, it is necessary to emphasise the ethical context and value and contextual parameters. Initiatives associated with the transformation to the new paradigm Industry 4.0 start emerging also in Slovakia. The initiative originated at the Ministry of Economy and was approved by the Slovak Government in October 2018. Action Plan was prepared in cooperation with the representatives of individual departments, industry, associations, and the academic circles. This national concept perceives the process in conjunction with other social components and stakeholders. National strategies and conceptions tend to underestimate the ethical aspect, not taking it as an important part of innovative approaches, mitigation of risks, or prevention. We hold the opinion that Industry 4.0 constitutes a fundamental turning point that deserves ethical appreciation and solutions. The peculiarities of this paradigm should also be explored within ethics and enter, in a constructive manner, the discourse in the area of science and research, both within professional socialisation and within the area of institutionalisation of ethical instruments in order to minimise, to a maximum possible extent, the ethical risks and potential negative consequences of new technologies and use of digital data in relation to customers and partners.

AI Agent在科研全流程中的应用趋势与发展机会分析（2026-2029）

深度研究匿名用户发表于 2026年05月06日 14:5986阅读

发起深度研究

内容由 AI 生成，仅供参考，请仔细甄别

参考文献

9The promise of artificial intelligence in chemical engineering: Is it here, finally?OpenAlex

Venkat Venkatasubramanian

20Comparative effectiveness of anti-seizure medications in emulated trials using medical informatics.PubMed

Kevin Xie, Jacob Korzun, Daniel J Zhou, et al.

Brain. 2025 Dec 4;148(12):4288-4298. doi: 10.1093/brain/awaf238.

30Troubling Trends in Biomedical Research Publication: "Publish or Perish" Results in a Propensity for Ethical Violations.PubMed

James H Lubowitz, Elizabeth Matzkin, Michael J Rossi

Arthroscopy. 2025 Apr;41(4):859-862. doi: 10.1016/j.arthro.2024.12.017. Epub 2024 Dec 20.

35Navigating academic integrity in biomedical research: the impact of large language models on current practices and future directions.PubMed

Anqi Lin, Zuwei Chen, Aimin Jiang, et al.

Int J Surg. 2026 Feb 1;112(2):4418-4433. doi: 10.1097/JS9.0000000000003839. Epub 2025 Nov 21.

37ChatGPT Utility in Healthcare Education, Research, and Practice: Systematic Review on the Promising Perspectives and Valid ConcernsOpenAlex

Malik Sallam

40Artificial Intelligence Can Generate Fraudulent but Authentic-Looking Scientific Medical Articles: Pandora’s Box Has Been OpenedOpenAlex

Martin Májovský, Martin Černý, Matěj Kasal, et al.

41Managing artificial intelligenceOpenAlex

Paul R. Krausman

49Practical Considerations and Ethical Implications of Using Artificial Intelligence in Writing Scientific ManuscriptsOpenAlex

Muhammad Nadeem Yousaf

50Dignity of science and the use of ChatGPT as a co-authorOpenAlex

Manuel Scimeca, Rita Bonfiglio

51Addressing student use of generative AI in schools and universities through academic integrity reportingOpenAlex

Steven A. Peterson

53The future (Industry 4.0) is closer than we think. Will it also be ethical?OpenAlex

Pavel Fobel, Aleksandra Kuzior

Suppr 超能文献

Suppr 超能文献

AI Agent在科研全流程中的应用趋势与发展机会分析（2026-2029）

1. AI Agent赋能科研的发展基础与现状概述

1.1 AI Agent的核心技术特性与科研场景适配逻辑

1.2 全球科研领域AI Agent的落地应用进展

2. AI Agent在科研核心流程的典型应用与演进趋势

2.1 多源文献智能检索与知识体系梳理

2.2 定向化实验方案设计与可行性预评估

2.3 多模态科研数据自动化分析与规律挖掘

2.4 结构化论文写作辅助与期刊适配优化

2.5 同行评审环节的智能辅助与效率提升

3. AI Agent科研应用伴生的科研诚信风险与防控难点

3.1 内容生成类风险

3.2 流程合规类风险

3.3 现有科研监管体系的适配短板

参考文献

AI Agent在科研全流程中的应用趋势与发展机会分析（2026-2029）

1. AI Agent赋能科研的发展基础与现状概述

1.1 AI Agent的核心技术特性与科研场景适配逻辑

1.2 全球科研领域AI Agent的落地应用进展

2. AI Agent在科研核心流程的典型应用与演进趋势

2.1 多源文献智能检索与知识体系梳理

2.2 定向化实验方案设计与可行性预评估

2.3 多模态科研数据自动化分析与规律挖掘

2.4 结构化论文写作辅助与期刊适配优化

2.5 同行评审环节的智能辅助与效率提升

3. AI Agent科研应用伴生的科研诚信风险与防控难点

3.1 内容生成类风险

3.2 流程合规类风险

3.3 现有科研监管体系的适配短板

参考文献

Suppr 超能文献

Suppr 超能文献

AI Agent在科研全流程中的应用趋势与发展机会分析（2026-2029）

1. AI Agent赋能科研的发展基础与现状概述

1.1 AI Agent的核心技术特性与科研场景适配逻辑

1.2 全球科研领域AI Agent的落地应用进展

2. AI Agent在科研核心流程的典型应用与演进趋势

2.1 多源文献智能检索与知识体系梳理

2.2 定向化实验方案设计与可行性预评估

2.3 多模态科研数据自动化分析与规律挖掘

2.4 结构化论文写作辅助与期刊适配优化

2.5 同行评审环节的智能辅助与效率提升

3. AI Agent科研应用伴生的科研诚信风险与防控难点

3.1 内容生成类风险

3.2 流程合规类风险

3.3 现有科研监管体系的适配短板

参考文献

1A Survey of Multi-AI Agent Collaboration: Theories, Technologies and ApplicationsOpenAlex

2Research on Intelligent Agent Technology and Applications Based on Large ModelsOpenAlex

3An In-depth Survey of Large Language Model-based Artificial Intelligence AgentsOpenAlex

4Artificial intelligence agents in cancer research and oncology.PubMed

5Empowering biomedical discovery with AI agents.PubMed

6Artificial intelligence guided discovery of a barrier-protective therapy in inflammatory bowel disease.PubMed

7ChatGPT: Bullshit spewer or the end of traditional assessments in higher education?OpenAlex

8<scp>ChatGPT</scp> and a new academic reality: <scp>Artificial Intelligence‐written</scp> research papers and the ethics of the large language models in scholarly publishingOpenAlex

9The promise of artificial intelligence in chemical engineering: Is it here, finally?OpenAlex

10Artificial Intelligence (AI): Multidisciplinary perspectives on emerging challenges, opportunities, and agenda for research, practice and policyOpenAlex

11Artificial Intelligence for Student Assessment: A Systematic ReviewOpenAlex

12The Impact of AI and LMS Integration on the Future of Higher Education: Opportunities, Challenges, and Strategies for TransformationOpenAlex

13Systematic review of research on artificial intelligence applications in higher education – where are the educators?OpenAlex

14Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AIOpenAlex

15Opinion Paper: “So what if ChatGPT wrote it?” Multidisciplinary perspectives on opportunities, challenges and implications of generative conversational AI for research, practice and policyOpenAlex

16LITERAS: Biomedical literature review and citation retrieval agentsOpenAlex

17An AI Agent-Based System for Retrieving Compound Information in Traditional Chinese MedicineOpenAlex

18Digital Twin: Values, Challenges and Enablers From a Modeling PerspectiveOpenAlex

19Increasing acceptance of AI-generated digital twins through clinical trial applications.PubMed

20Comparative effectiveness of anti-seizure medications in emulated trials using medical informatics.PubMed

21Personalized medicine in acromegaly: insights from the ACROFAST clinical trial.PubMed

22Development of a differential treatment selection model for depression on consolidated and transformed clinical trial datasets.PubMed

23Personalised selection of medication for newly diagnosed adult epilepsy: study protocol of a first-in-class, double-blind, randomised controlled trial.PubMed

24Advancing innovation in financial stability: A comprehensive review of ai agent frameworks, challenges and applicationsOpenAlex

25Review of deep learning: concepts, CNN architectures, challenges, applications, future directionsOpenAlex

26Wordcraft: Story Writing With Large Language ModelsOpenAlex

27PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviewsOpenAlex

28AI-assisted peer reviewOpenAlex

29Use of artificial intelligence and the future of peer reviewOpenAlex

30Troubling Trends in Biomedical Research Publication: "Publish or Perish" Results in a Propensity for Ethical Violations.PubMed

31The transformative impact of large language models on medical writing and publishing: current applications, challenges and future directions.PubMed

32Fraud in Medical Publications.PubMed

33Artificial Intelligence in Detecting Statistical Errors: Implications for Authors, Reviewers, and Editors.PubMed

34Detecting fabrication in large-scale molecular omics data.PubMed

35Navigating academic integrity in biomedical research: the impact of large language models on current practices and future directions.PubMed

36Open AI Chat Fabricated Anti–Electroconvulsive Therapy Statements and ReferencesOpenAlex

37ChatGPT Utility in Healthcare Education, Research, and Practice: Systematic Review on the Promising Perspectives and Valid ConcernsOpenAlex

38Artificial intelligence-assisted academic writing: recommendations for ethical useOpenAlex

39Navigating the Landscape of Personalized Medicine: The Relevance of ChatGPT, BingChat, and Bard AI in Nephrology Literature SearchesOpenAlex

40Artificial Intelligence Can Generate Fraudulent but Authentic-Looking Scientific Medical Articles: Pandora’s Box Has Been OpenedOpenAlex

41Managing artificial intelligenceOpenAlex

42Ethics and artificial intelligence.PubMed

43Generative artificial intelligence and ethical considerations in health care: a scoping review and ethics checklist.PubMed

44Ethical considerations for artificial intelligence in dermatology: a scoping review.PubMed

45ChatGPT in medicine: an overview of its applications, advantages, limitations, future prospects, and ethical considerationsOpenAlex

46Advancing AI in healthcare: A comprehensive review of best practices.PubMed

47AI: from rational agents to socially responsible agentsOpenAlex

48Academic Integrity considerations of AI Large Language Models in the post-pandemic era: ChatGPT and beyondOpenAlex

49Practical Considerations and Ethical Implications of Using Artificial Intelligence in Writing Scientific ManuscriptsOpenAlex

50Dignity of science and the use of ChatGPT as a co-authorOpenAlex

51Addressing student use of generative AI in schools and universities through academic integrity reportingOpenAlex

52On the creativity of large language modelsOpenAlex

53The future (Industry 4.0) is closer than we think. Will it also be ethical?OpenAlex

AI Agent在科研全流程中的应用趋势与发展机会分析（2026-2029）

1. AI Agent赋能科研的发展基础与现状概述

1.1 AI Agent的核心技术特性与科研场景适配逻辑

1.2 全球科研领域AI Agent的落地应用进展

2. AI Agent在科研核心流程的典型应用与演进趋势

2.1 多源文献智能检索与知识体系梳理

2.2 定向化实验方案设计与可行性预评估

2.3 多模态科研数据自动化分析与规律挖掘

2.4 结构化论文写作辅助与期刊适配优化

2.5 同行评审环节的智能辅助与效率提升