1
Within five years, do you expect AI to perform parts of expert scientific review at the level of a credentialed expert in your field?
Results
(229 Answers)
Answer Explanations
- Yes — some partsuser-154906Technically, AI will be capable of doing that and even more. However, scientific review processes would need to continue being supervised by humans in most of the process.
- Yes — most of ituser-597773Nkw a day technology especially AI and robotics are largeky fabricated and developed to help and decrease the intense of human labor even they lower in a high budget and labor with accurate work.
It might be participate on revuew orocess. Why not since it used in medicine - Yes — some partsuser-568782Plagiarism check for example.
- Nouser-887652AI might be useful to flag either unclear language OR over-reliance on AI, but it is incapable of adequately dealing with anything unprecedented.
- Nouser-441029AI will perform scientific reviews, as it is performing it now, but it won't be on the level of a quality expert due to low ability to evaluate the reliability of data it uses to perform a review. Though, many human reviewers also suffer from that, so...
- Yes — most of ituser-153764Much of the US EPA protocols for hazardous waste site-specific human health risk assessments are already computerized by many different consultant organizations. One such model developed >10 years ago was CalTox but it was never widely adopted. There is little benefit to having state and federal EPA employees doing these evaluations over and over.
- Yes — most of ituser-498547I expect that AI will perform parts of expert scientific review at the level of a credentialed expert in any field because the expert scientific review requires knowledge of the specific science field. The expert does not need to perform actual experiments and assays to provide the review.
- Nouser-157767AI will never outsmart a person.
- Yes — some partsuser-701806Data collection and first draft can be delivered. Quality check, conclusion and sign off still require expert judgement
- Yes — most of ituser-460380AI is progressing day by day. I believe 5 years is enough to have a highly performing AI model.
- Yes — some partsuser-267027Artificial intelligence should be regarded as a beneficial tool, just like iThenticate, EndNote, Zotero, translation software, or statistical software. It helps produce higher-quality work, detect inconsistencies, and draw attention to aspects that we may have overlooked. In this sense, AI enables experts to work at a higher level without replacing scientific reasoning. Assessing originality, scientific significance, methodological rigour, the interpretation of results in their proper context, and the overall contribution of a study still requires human expertise. AI does not contribute scientific judgement; it enhances the expert's ability to exercise that judgement. I would simply say that it is like wearing glasses to see better. Glasses do not see for us; vision can only be guided by the person wearing them. Likewise, only the expert can exercise critical judgement, drawing on experience, disciplinary knowledge, and scientific understanding. AI should therefore be regarded as a powerful support tool that assists scientific reviewers rather than replacing them.
- Yes — some partsuser-109201Yes, particularly for literature synthesis, methodological assessment, statistical evaluation, reference verification, and identifying inconsistencies.
- Yes — some partsuser-958242Yes, beyond doubt
- Yes — most of ituser-649290Currently, most of the reviewers using this without declaration.
- Nouser-301040No because most researchers/experts use AI to write and perform other things, so using AI to review what it has written will not produce reliable result.
- Yes — some partsuser-582664There is no doubt that artificial intelligence is used these days in many fields and will be used more widely in the future to encompass all aspects of life. This has had many positive effects, but it has also had a significant negative impact on the human mind, as people have used artificial intelligence for the simplest tasks and no longer use their minds to think and search for information. In the field of scientific research, it has been used extensively. I do not deny that it has facilitated access to information, but it has been used by unscrupulous individuals to conduct entire research projects. However, in terms of its use in the field of medical sciences, it will be used more widely, but of course, humans cannot replace it, as we know that these sciences are partly and fundamentally dependent on humans.
- Yes — most of ituser-759482Fear from change will be replaced by more wise and useful non biased reviews.
- Yes — some partsuser-508016The IA can be used for review process of the methodology aspect , structure of manuscript , style, bibliographic references, cites and plagie, but I think that the novelty, scienftific contribution or relevance is more dificult of training to IA, althogth is not imposible, should keep in hands of researcher
- Yes — some partsuser-890708现在的人工智能发展很快，有很多功能已与人体接近，已开发了人机接口。
- Nouser-378118AI is good at collecting and analyzing written data. AI is not good at knowing which data is true which is scientifically not true.
- Yes — most of ituser-8664331. It has been observed that proper training of AI produces much closer results to ground truth ie like real expert.
2. Even in some places AI review can be much better than an expert specially in Mathematical, statistical and physical papers as it can verify the computations, equations and solutions whereas an ordinary expert may not have enough time to do so. - Yes — most of ituser-364111The next five years, AI is likely to perform at a high level in several aspects of scientific peer review, including summarization, language assessment, consistency checks, and initial methodological screening. Expert assessment of originality, scientific significance, domain-specific methodology, and contextual interpretation, however, will still require human reviewers. AI should support, not replace, expert scientific judgment.
- Yes — some partsuser-953899Honestly, in five years, AI can do many sections of the literature review and screening in a very short time which actually takes many months for humans to do, but still once the review, formatting and writing the paper is finished still the judgment and making of how much effective and save it to the patient on the bedside needs a human review.
- Yes — some partsuser-876636I have used perplexity and Claude for identifying similar case reports or article published in the literature (Literature Review).
- Nouser-188852From my experience with AI, I believe it excels at managing structured tasks as a useful tool. It is a valuable resource; however, in scientific review, there remains a need for expert judgment and reasoning that AI cannot yet offer.
- Yes — some partsuser-139009it can help by editing some grammatical errors that a science can mistake and also to clear and correct some English lanquage patterns. However, the science is the one who guides fully all the scientific paper exclusively.
- Yes — some partsuser-951744I expect that journal publishers will use AI to check for written and typographical errors and for similarity. Publishers may also use AI to turn reviewer feedback into constructive action points or to check for fairness/bias.
- Yes — some partsuser-539924The easier the task is to standardize, the sooner AI will perform it. The harder the task is to define, the more important human experts remain.
- Yes — some partsuser-813332It is the easiest way for managers to get information they think will answer their situations and seem to be expert.
- Yes — some partsuser-487971I don't think AI to perform parts of the expert scientific review at the level of a credential expert in my field. This is because their out put still need reliable verification with the subject matter specialists, especially in high-stakes scientific contexts. However those AI models that are consistently accurate, transparent and well reasoned in their answer might be do quarter of expert scientific community in the upcoming years.
- Yes — some partsuser-544312Yes, I expect that in the future many aspects, including scientific review, data analysis support, and evaluations, will be performed by artificial intelligence. Despite this development, I believe that AI will remain limited in making final decisions and cannot fully replace managerial decision-making. However, it can serve as a decision-support tool that assists managers in making more informed decisions.
- Yes — some partsuser-814722AI can already assist with literature searching, summarizing evidence, identifying relevant studies, extracting data, and drafting portions of reviews. Therefore, within five years, AI is likely to perform some parts of expert scientific review at an expert level, but not necessarily the entire process.
- Yes — most of ituser-194414Yes, I believe so. When provided with well-defined parameters and targeted queries in database searches, AI is exceptionally effective at synthesizing existing knowledge. Furthermore, it efficiently condenses findings and offers valuable support in refining and advancing research methodology.
- Nouser-616424The problem word here is "level". If you mean in terms of linguistic style and picking out contradictions, sure. But not in terms of intuition
- Nouser-316248Não, porque ela pode deixar passar muito coisa.
- Yes — most of ituser-871577My experience shows that the AI I am currently using has a good grasp of the essential message of scientific papers and can access a large number of relevant papers. I am uncertain as to how partial this is, it may depend on the field. But sofar I am very impressed, with the competence and speed and accuracy of any scientific opinions sought. I certainly have come across inaccuracies but this is rare on on further questioning this is corrected.
- Yes — some partsuser-481327I expect it to be "Yes - some parts", but I do not believe it is the correct, ethical, smart, nor scientific option.
- Nouser-228600These are literally language models. Expert review requirea experts
- Nouser-551482The AI will likely be applied automatically and reviewers will apply human judgement on top of that
- Yes — some partsuser-516400I think that specific portions of expert scientific review can be helped by AI. We have to be careful of how AI performs this task, what questions are asked to AI, and what actually are the responses.
- Yes — some partsuser-377267Some parts of a scientific review are based on routine work and can be performed by AI if adequately trained. One example is the systematic review of the literature.
- Yes — some partsuser-762565This is because it is beyond the reach of the human mind.
- Yes — some partsuser-234128Could be used to sort through differences between reviewers and provide a more "objective" viewpoint. Perhaps as a tie-breaker of sorts. It should not replace human judgment.
- Nouser-365645Actually, researcher and scientists first finding the unique things and then it will come to AI. Suppose I have new findings which are validated thoroughly. Now, establishment and testing takes time. What I can judge relevant to my field that can't be judged by trained model. These models are running through fed data and results. Many times, I found that there is conceptual mistakes. That is dangerous. So I feel it can't do full. But for small things may be reviewed but as per my experience, I find severe technical flaws in the AI generated results. Although, some people are using, but it won't be dependable fully. Therefore, sensitive information to be judged manually through learned expert. Then anybody may run AI to recheck.
- Yes — some partsuser-812675Most AI tools are learning the art of publishing because most of the manuscripts that are been written are encouraging the use of AI
- Yes — some partsuser-395064The question should be phrased as "do you expect researchers/similar to use AI models..." as the current phrasing implies that his can be done without a human in the loop, a popular misconception. I expect people to use AI models, generally LLMs, to perform selected parts of expert scientific review at the level of a paediatric oncology specialist, e.g tasks assessing study design, identifying statistical inconsistencies, checking whether conclusions are supported by the data, summarising relevant literature, and detecting omissions or internal contradictions. It may be especially useful in fields and locales where evidence is limited by small sample sizes, heterogeneous populations and differences in treatment capacity between settings. I do not expect AI to replace expert review as a whole. Judging clinical relevance, understanding the realities of care in LMIC, recognising when technically sound recommendations are not feasible or equitable and weighing the significance of findings for children and families still require experience and contextual judgement. I see AI becoming a co-reviewer rather than an independent substitute for an expert.
- Yes — some partsuser-432588AI can identify patterns, review grammar and scientific writing, and evaluate study design, methodology and statistical analyses. However, I believe that human expertise remains essential for assessing the scientific relevance, clinical implications and overall validity of research, as well as for providing a more humanized and profound bioethical perspective that AI cannot offer
- Yes — some partsuser-984622I don't see AI being good enough in that time frame to complete a full review, although the benefit of rapid review and objectivity is intriguing. I do think that AI could screen out poorly done applications or manuscripts which would save some time.
- Yes — some partsuser-490306Scientific researches require human efforts and their judgments based on the evidence and experimental observations. So, the AI might assist human but I think it cannot replace humans.
- Yes — some partsuser-477751AI is not able to explain the context of effects in toxicity studies, helpful in summarizing results and generating presentations. Limited ability to synthesize complex datasent into new hypothese
- Uncertainuser-468267To reach the level of a credentialed expert there needs to be set a specialized quality assurance filter in AI. This may not be attainable in machine learning protocols (at least not in five years).
- Nouser-439854I don't think AI can apply the expert judgement and it has no institutional knowledge
- Yes — some partsuser-77091There will be resistance to AI from various professionals, and their comments will be valid. AI is not as perfect as some would have us believe.
- Yes — some partsuser-328992I'm already using LLMs on a daily basis for my research and adminstration work. For reseach, LLMs and agents can very nicely plan the entire workflow from hypthesis formulation, math derivation, numerical verification plan, code generation, simulation, results analysis and finally paper writting. In a few years, this will be even stronger.
- Yes — some partsuser-550102AI can create hypotheses that hadn't been considered before, so with new technologies it can create more complex tasks such as participating in a review. In fact, I think it's already working on reviewing the arguments one can make by asking AI.
- Yes — some partsuser-746645
I believe that in the future, AI will be able to help researchers quickly summarize the literature and identify relevant references. It can also assist in checking whether a manuscript complies with a journal's submission requirement and improve English grammar and writing style, especially for authors whose first language is not English
- Yes — some partsuser-851286I am an expert in biological taxonomy and ecology.
I believe that the analytical aspects of ecology will eventually be replaced by AI.
On the other hand, I do not believe that taxonomy will be replaced by AI. - Yes — some partsuser-151126I expect AI to be able to identify part of potential scientific flaws, but it will need to be overlooked by a human do to the guessing nature of AI models which does not fit with scientific standards.
- Yes — some partsuser-164084The large language model is being refined and fine tuned with the intelligence expanding symmetrically. We are getting to a stage where those LLM can handle some of the review processes.
- Yes — some partsuser-532952Life moves on. Research needs advanced models with capacity for large datasets
- Yes — some partsuser-756416I believe that certain aspects can be automated with AI, but they should always remain subject to human supervision, quality control, and validation by experts.
- Yes — most of ituser-923069Within five years, I think AI will play an important role in scientific review. Since most reviewers aren't native English speakers, AI can translate content into their own language, and it can also verify images and references to catch fraud or inconsistencies. But a human expert is still needed to know whether the finding is truly novel, the methodology sound, and to valorate the clinical relevance.
- Yes — most of ituser-616480The challenge of getting sufficient peer reviewers combined with the ability of (good) AI systems to identify methodological flaws. There would always (hopefully
) be an appropriately skilled human to check/confirm AI-based reviews - Yes — most of ituser-531244AI is rapidly evolving to development of agents capable of independent work. Objective criteria for scientific review will mot only be spelled out for agents but will be improved by them as the gain experience with review.
Debate (1 Comment)
Please log in to comment.
user-267027