science research papers

Tomorrow´s Research Today

SSRN provides 1,448,162 preprints and research papers from 1,834,323 researchers in over 65 disciplines.

New Networks

Research disciplines, applied sciences.

APPLIED SCIENCES are those disciplines, including applied and pure mathematics, that apply existing scientific knowledge to develop practical applications.

Health Sciences

HEALTH SCIENCES are those disciplines that address the use of science and technology to the delivery of healthcare.

HUMANITIES are those disciplines that investigate human constructs, cultures and concerns, using critical and analytical approaches.

Life Sciences

LIFE SCIENCES are those disciplines that study living organisms, their life processes, and their relationships to each other and their environment.

Physical Sciences

PHYSICAL SCIENCES are those disciplines that study natural sciences, dealing with nonliving materials.

Social Sciences

SOCIAL SCIENCES are those disciplines that study (a) institutions and functioning of human society and the interpersonal relationships of individuals as members of society; (b) a particular phase or aspect of human society.

Products and Services

Recent announcements.

SSRN is devoted to the rapid worldwide dissemination of preprints and research papers and is composed of a number of specialized research networks.

Special thanks to:

Current Selection: All SSRN Networks Modify

To use this feature, you must allow cookies or be signed in.

MEMBER SIGN IN

First-time user? Free Registration

Forgot ID or Password? | Contact Us

Every print subscription comes with full digital access

Science News

More than 100 bacteria species can flourish in microwave ovens

Swabs of 30 microwave ovens in different settings identified over 100 bacterial species, some of which could be pathogenic or cause food-borne disease.

National Geographic’s ‘OceanXplorers’ dives into the ocean’s mysteries

50 years ago, antibiotic resistant bacteria became a problem outside hospitals

New COVID-19 booster shots have been approved. When should you get one?

A newly approved ‘living drug’ could save more cancer patients’ lives

Can scientists make fruits and veggies resilient to climate change?

Zapping sand to create rock could help curb coastal erosion

Trending stories.

A long-awaited cancer treatment reaches patients

Scientists are getting serious about UFOs. Here’s why

The world’s fastest microscope makes its debut

Spotlight on health.

Why mpox is a global health emergency — again

The WHO made the declaration as a potentially more infectious version of the deadly virus has emerged and mpox cases are rapidly rising across Africa.

More than 4 billion people may not have access to clean water

A hunger protein reverses anorexia symptoms in mice, from the archives.

Quarks, Timelike Photons Do Not Mix

August 17, 1974 Vol. 106 No. #7

Science News Magazine

August 10, 2024 Vol. 206 No. 02

Psilocybin temporarily dissolves brain networks

The last woolly mammoths offer new clues to why the species went extinct.

Featured Media

A risk-tolerant immune system may enable house sparrows’ wanderlust

Birds that are willing to eat seed spiked with chicken poop have higher expression levels of a gut immunity gene, a new study finds.

A new element on the periodic table might be within reach 

Twisters asks if you can 'tame' a tornado. We have the answer

Freeze-drying turned a woolly mammoth’s DNA into 3-D ‘chromoglass’

Bird flu has been invading the brains of mammals. Here’s why

Follow Science News

• Follow Science News on X
• Follow Science News on Facebook
• Follow Science News on Instagram

More Stories

This protist unfolds its ‘neck’ up to 30 times its body length to scout prey

The largest known genome belongs to a tiny fern, it’s a big year for cicadas. here’s what to know about this year’s emergence.

World record speeds for two Olympics events have fallen over time. We can go faster

Does social status shape height, rain bosworth studies how deaf children experience the world.

The world’s record-breaking hot streak has lasted 14 months. When will it end?

Extraordinary heat waves have readers asking how a/c affects greenhouse gas emissions, zigzag walls could help buildings beat the heat.

The nearest midsized black hole might instead be a horde of lightweights

A distant quasar may be zapping all galaxies around itself, some meteors leave trails lasting up to an hour. now we may know why, paper cut physics pinpoints the most hazardous types of paper, can light spark superconductivity a new study reignites debate, health & medicine.

Expanding antibiotic treatment in sub-Saharan Africa could save kids’ lives

Your medications might make it harder for you to beat the heat, your face’s hot spots may reveal how well you are aging.

‘Turning to Stone’ paints rocks as storytellers and mentors

Why japan issued its first-ever mega-earthquake alert, squall line tornadoes are sneaky, dangerous and difficult to forecast, science & society.

‘After 1177 B.C.’ describes how societies fared when the Bronze Age ended

Scientists are fixing flawed forensics that can lead to wrongful convictions, language models may miss signs of depression in black people’s facebook posts.

Subscribers, enter your e-mail address for full access to the Science News archives and digital editions.

Not a subscriber? Become one now .

• Find journals
• My journals

Register Sign in

Register or sign-in in order to manage your journal lists

Sign in or register to save a journal

To save a journal and create lists, you need to sign in to your Elsevier account.

Find the right journal for your research

Looking for the best journal match for your paper? Search the world's leading source of academic journals using your abstract or your keywords and other details.

Check if you're eligible for open access (OA) savings.

arXiv's Accessibility Forum starts next month!

Help | Advanced Search

Computer Science > Machine Learning

Title: kan 2.0: kolmogorov-arnold networks meet science.

Abstract: A major challenge of AI + Science lies in their inherent incompatibility: today's AI is primarily based on connectionism, while science depends on symbolism. To bridge the two worlds, we propose a framework to seamlessly synergize Kolmogorov-Arnold Networks (KANs) and science. The framework highlights KANs' usage for three aspects of scientific discovery: identifying relevant features, revealing modular structures, and discovering symbolic formulas. The synergy is bidirectional: science to KAN (incorporating scientific knowledge into KANs), and KAN to science (extracting scientific insights from KANs). We highlight major new functionalities in the pykan package: (1) MultKAN: KANs with multiplication nodes. (2) kanpiler: a KAN compiler that compiles symbolic formulas into KANs. (3) tree converter: convert KANs (or any neural networks) to tree graphs. Based on these tools, we demonstrate KANs' capability to discover various types of physical laws, including conserved quantities, Lagrangians, symmetries, and constitutive laws.

Submission history

Access paper:.

• HTML (experimental)
• Other Formats

References & Citations

• INSPIRE HEP
• Google Scholar
• Semantic Scholar

BibTeX formatted citation

Bibliographic and Citation Tools

Code, data and media associated with this article, recommenders and search tools.

• Institution

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .

• Advanced search
• Peer review

Discover relevant research today

Advance your research field in the open

Reach new audiences and maximize your readership

ScienceOpen puts your research in the context of

Publications

For Publishers

ScienceOpen offers content hosting, context building and marketing services for publishers. See our tailored offerings

• For academic publishers  to promote journals and interdisciplinary collections
• For open access journals  to host journal content in an interactive environment
• For university library publishing  to develop new open access paradigms for their scholars
• For scholarly societies  to promote content with interactive features

For Institutions

ScienceOpen offers state-of-the-art technology and a range of solutions and services

• For faculties and research groups  to promote and share your work
• For research institutes  to build up your own branding for OA publications
• For funders  to develop new open access publishing paradigms
• For university libraries to create an independent OA publishing environment

For Researchers

Make an impact and build your research profile in the open with ScienceOpen

• Search and discover relevant research in over 95 million Open Access articles and article records
• Share your expertise and get credit by publicly reviewing any article
• Publish your poster or preprint and track usage and impact with article- and author-level metrics
• Create a topical Collection  to advance your research field

Create a Journal powered by ScienceOpen

Launching a new open access journal or an open access press? ScienceOpen now provides full end-to-end open access publishing solutions – embedded within our smart interactive discovery environment. A modular approach allows open access publishers to pick and choose among a range of services and design the platform that fits their goals and budget.

Continue reading “Create a Journal powered by ScienceOpen”   

What can a Researcher do on ScienceOpen?

ScienceOpen provides researchers with a wide range of tools to support their research – all for free. Here is a short checklist to make sure you are getting the most of the technological infrastructure and content that we have to offer. What can a researcher do on ScienceOpen? Continue reading “What can a Researcher do on ScienceOpen?”   

ScienceOpen on the Road

Upcoming events.

• 15 June – Scheduled Server Maintenance, 13:00 – 01:00 CEST

Past Events

• 20 – 22 February – ResearcherToReader Conference
• 09 November – Webinar for the Discoverability of African Research
• 26 – 27 October – Attending the Workshop on Open Citations and Open Scholarly Metadata
• 18 – 22 October – ScienceOpen at Frankfurt Book Fair.
• 27 – 29 September – Attending OA Tage, Berlin .
• 25 – 27 September – ScienceOpen at Open Science Fair
• 19 – 21 September – OASPA 2023 Annual Conference .
• 22 – 24 May – ScienceOpen sponsoring Pint of Science, Berlin.
• 16-17 May – ScienceOpen at 3rd AEUP Conference.
• 20 – 21 April – ScienceOpen attending Scaling Small: Community-Owned Futures for Open Access Books .

What is ScienceOpen?

• Smart search and discovery within an interactive interface
• Researcher promotion and ORCID integration
• Open evaluation with article reviews and Collections
• Business model based on providing services to publishers

Live Twitter stream

Some of our partners:.

A new ‘AI scientist’ can write science papers without any human input. Here’s why that’s a problem

Dean, School of Computing Technologies, RMIT University, RMIT University

Disclosure statement

Karin Verspoor receives funding from the Australian Research Council, the Medical Research Future Fund, the National Health and Medical Research Council, and Elsevier BV. She is affiliated with BioGrid Australia and is a co-founder of the Australian Alliance for Artificial Intelligence in Healthcare.

RMIT University provides funding as a strategic partner of The Conversation AU.

View all partners

Scientific discovery is one of the most sophisticated human activities. First, scientists must understand the existing knowledge and identify a significant gap. Next, they must formulate a research question and design and conduct an experiment in pursuit of an answer. Then, they must analyse and interpret the results of the experiment, which may raise yet another research question.

Can a process this complex be automated? Last week, Sakana AI Labs announced the creation of an “AI scientist” – an artificial intelligence system they claim can make scientific discoveries in the area of machine learning in a fully automated way.

Using generative large language models (LLMs) like those behind ChatGPT and other AI chatbots, the system can brainstorm, select a promising idea, code new algorithms, plot results, and write a paper summarising the experiment and its findings, complete with references. Sakana claims the AI tool can undertake the complete lifecycle of a scientific experiment at a cost of just US$15 per paper – less than the cost of a scientist’s lunch.

These are some big claims. Do they stack up? And even if they do, would an army of AI scientists churning out research papers with inhuman speed really be good news for science?

How a computer can ‘do science’

A lot of science is done in the open, and almost all scientific knowledge has been written down somewhere (or we wouldn’t have a way to “know” it). Millions of scientific papers are freely available online in repositories such as arXiv and PubMed .

LLMs trained with this data capture the language of science and its patterns. It is therefore perhaps not at all surprising that a generative LLM can produce something that looks like a good scientific paper – it has ingested many examples that it can copy.

What is less clear is whether an AI system can produce an interesting scientific paper. Crucially, good science requires novelty.

But is it interesting?

Scientists don’t want to be told about things that are already known. Rather, they want to learn new things, especially new things that are significantly different from what is already known. This requires judgement about the scope and value of a contribution.

The Sakana system tries to address interestingness in two ways. First, it “scores” new paper ideas for similarity to existing research (indexed in the Semantic Scholar repository). Anything too similar is discarded.

Second, Sakana’s system introduces a “peer review” step – using another LLM to judge the quality and novelty of the generated paper. Here again, there are plenty of examples of peer review online on sites such as openreview.net that can guide how to critique a paper. LLMs have ingested these, too.

AI may be a poor judge of AI output

Feedback is mixed on Sakana AI’s output. Some have described it as producing “ endless scientific slop ”.

Even the system’s own review of its outputs judges the papers weak at best. This is likely to improve as the technology evolves, but the question of whether automated scientific papers are valuable remains.

The ability of LLMs to judge the quality of research is also an open question. My own work (soon to be published in Research Synthesis Methods ) shows LLMs are not great at judging the risk of bias in medical research studies, though this too may improve over time.

Sakana’s system automates discoveries in computational research, which is much easier than in other types of science that require physical experiments. Sakana’s experiments are done with code, which is also structured text that LLMs can be trained to generate.

AI tools to support scientists, not replace them

AI researchers have been developing systems to support science for decades. Given the huge volumes of published research, even finding publications relevant to a specific scientific question can be challenging.

Specialised search tools make use of AI to help scientists find and synthesise existing work. These include the above-mentioned Semantic Scholar, but also newer systems such as Elicit , Research Rabbit , scite and Consensus .

Text mining tools such as PubTator dig deeper into papers to identify key points of focus, such as specific genetic mutations and diseases, and their established relationships. This is especially useful for curating and organising scientific information.

Machine learning has also been used to support the synthesis and analysis of medical evidence, in tools such as Robot Reviewer . Summaries that compare and contrast claims in papers from Scholarcy help to perform literature reviews.

All these tools aim to help scientists do their jobs more effectively, not to replace them.

AI research may exacerbate existing problems

While Sakana AI states it doesn’t see the role of human scientists diminishing, the company’s vision of “a fully AI-driven scientific ecosystem” would have major implications for science.

One concern is that, if AI-generated papers flood the scientific literature, future AI systems may be trained on AI output and undergo model collapse . This means they may become increasingly ineffectual at innovating.

However, the implications for science go well beyond impacts on AI science systems themselves.

There are already bad actors in science, including “paper mills” churning out fake papers . This problem will only get worse when a scientific paper can be produced with US$15 and a vague initial prompt.

The need to check for errors in a mountain of automatically generated research could rapidly overwhelm the capacity of actual scientists. The peer review system is arguably already broken , and dumping more research of questionable quality into the system won’t fix it.

Science is fundamentally based on trust. Scientists emphasise the integrity of the scientific process so we can be confident our understanding of the world (and now, the world’s machines) is valid and improving.

A scientific ecosystem where AI systems are key players raises fundamental questions about the meaning and value of this process, and what level of trust we should have in AI scientists. Is this the kind of scientific ecosystem we want?

• Artificial intelligence (AI)
• Computer science
• Research integrity
• Paper mills

Head of Evidence to Action

Supply Chain - Assistant/Associate Professor (Tenure-Track)

Education Research Fellow

OzGrav Postdoctoral Research Fellow

Casual Facilitator: GERRIC Student Programs - Arts, Design and Architecture

• Library of Congress
• Research Guides
• Science & Technology

Technical Reports & Standards Collection Guide

Introduction.

• Technical Reports Collections
• Standards Collection
• American Documentation Institute (ADI)
• Office of Scientific Research and Development (OSRD) Collection
• Synthetic Rubber Project
• Technical Translations (TT) Series
• Locating Technical Reports and Standards
• Research Assistance and Reproductions
• Online Resources and Databases
• Using the Library of Congress
• Jennifer Harbster, Head, Science Section, Researcher Engagement & General Collections Division
• Sean Bryant, Reference Librarian, Researcher Engagement & General Collections Division
• Ashley Fielder,  Librarian for Medicine and Life Science. Science Section, Researcher Engagement & General Collections
• Created:  September 22, 2023

Last Updated: May 7, 2024

Science & Technical Reports : Ask a Librarian

Have a question? Need assistance? Use our online form to ask a librarian for help.

Get connected to the Library’s large and diverse collections related to science, technology, and business through our Inside Adams Blog. This blog also features upcoming events and collection displays, classes and orientations, new research guides, and more.

The Library of Congress is completing a project to update and modernize Library reading room websites. As a part of the process, “The Technical Reports and Standards Collection” is in the process of being updated and migrated to this new platform. The process has not yet been completed and the guide remains subject to change.

Researchers with questions about the collection are encouraged to contact a science or business librarian using the Ask-a-Librarian: Science and Technical Reports or Ask a Librarian: Business online form, by phone, at (202) 707-5639, or in person, at the reference desk, in the Science and Business Reading Room, on the fifth floor of the Library's John Adams Building.

Technical Reports

Technical reports are designed to quickly alert researchers to recent findings and developments in scientific and technical research. These reports are issued for a variety of purposes:

• to communicate results or describe progress of a research project
• to convey background information on an emerging or critical research topic
• to provide lists of instructions or procedures for current practices
• to determine the feasibility of a technology and recommend if research should be continued (and how to evaluate any further progress made)
• to detail technical specifications (materials, functions, features, operation, market potential, etc.)

Technical reports first appeared in the early part of the 20th century. The U.S. Geological Survey (USGS) published a series of professional papers beginning in 1902, and the National Advisory Committee for Aeronautics (NACA) issued its first report in 1915. But, the format gained importance during World War II, emerged in the postwar era, and remains, to this day, a major tool for reporting progress in science and technology, as well as in education, business, and social sciences research. The names given to series of these publications vary, but are often such generic terms as "technical reports," "working papers," "research memoranda," "internal notes," "occasional papers," "discussion papers" or "gray (or grey) literature." In the physical and natural sciences, "technical report" seems to be the preferred designation. For reports dealing with business, education, and the social sciences, on the other hand, the terms "working paper," "occasional paper," and "memorandum" are often the designations of choice. Other, more specific types of technical reports include "preprints" and "reprints." Preprints generally are versions of papers issued by researchers before their final papers are published by commercial publishers. Preprints allow researchers to communicate their findings quickly, but usually have not been peer reviewed. Reprints are typically released to heighten awareness of the research being conducted in a particular field or at a single institution. The term, "technical report" encompasses all of these designations.

Since many of these publications are intended to provide just a temporary snapshot of current research in a particular field or topic, they may contain the some of following distinctions:

• Rapid communication of new research results
• Dissemination to a targeted audience.
• Detailed methodologies, in order to facilitate review of research results by others
• No peer review, though there is often another selection process for publication (grant, contract, or institutional affiliation)
• Not published by typical commercial publishers (instead reports are issued or sponsored by government agencies, professional associations, societies, councils, foundations, laboratories, universities, etc.)
• Corporate authorship, where present, is typically emphasized

Unfortunately, uncertain availability, limited print runs, and decentralized distribution patterns with little bibliographic information are also often characteristics of this literature.

The Federal Government issues many different types of technical reports. An overview of some of these can be found in a May 2001 GAO report, " Information Management: Dissemination of Technical Reports ." Government issued or sponsored reports contain an additional characteristic - they may be subject to distribution restrictions linked to their classification status. Although references to classified reports may be found in technical reports literature, the security status or limited distribution of reports may make them unavailable to the general public and to the Library as well, as the Library holds only titles in the public domain. Those interested in locating such materials can consult the U.S. Department of Justice's Freedom of Information Act  site for guidance in obtaining these reports.

To enable them to be identified and located, technical reports are assigned report codes by agencies or organizations involved in their production or distribution. These codes may be referred to as "accession numbers," "agency report series numbers," "contract numbers," "grant numbers" or by other names, and include dates and individual report numbers. Typically, reports are assigned multiple codes and these codes help to identify the sponsoring agency, the organization performing the research or the organization disseminating the report.  Most technical reports held by the Library of Congress are not cataloged, and, for these reports, one or more report codes is required for Library staff to check the collections for a report or to locate and retrieve it. For more information about the current Standard Technical Report Number format (STRN) see ANSI/NISO Z39.23- 1997 (S2015) Standard Technical Reports Number Format and Creation . 

Standards are specifications which define products, methods, processes or practices, and are known to have existed as early as 7000 B.C., when cylindrical stones were used as units of weight in Egypt. According to  Office of Management and Budget (OMB) Circular A-119 , as revised in 2016, the term "standard" or "technical standard" refers to:

• common and repeated use of rules, conditions, guidelines or characteristics for products or related processes and production methods, and related management systems practices;
• the definition of terms; classification of components; delineation of procedures; specification of dimensions, materials, performance, designs, or operations; measurement of quality and quantity in describing materials, processes, products, systems, services, or practices; test methods and sampling procedures; or descriptions of fit and measurements of size or strength; and
• terminology, symbols, packaging, marking or labeling requirements as they apply to a product, process, or production method.

Technical standards are not "professional standards of personal conduct; or institutional codes of ethics." (p. 15).

Standards are typically generated by governments or by professional associations and organizations interested in or affected by the subject matter of particular standards. For example, U.S. government standards mandated by the  Fair Packaging & Labeling Act (FPLA)  have standardized the labeling required for packaging in which consumer commodities is sold. Standards set the basis for determining consistent and acceptable minimum levels of reliability and safety, and are adhered to either voluntarily or as mandated by law. For a more complete overview, see the NIST report  " The ABC's of Standards Activities " by Maureen A. Breitenberg (2009).

The Library of Congress standards collection includes military and other federal standards, industry standards, and a few older international standards from Russia, China, and South Africa. Material from the collection is available in various formats, including digital, print, and microform materials. The majority of the Library's standards collection held in the Science Section's Technical Reports and Standards Collection. The collection remains largely uncatalogued, and as a result, most items from this collection are not discoverable in the Library's online catalog. Inquires on Library holdings can be sent to the Science Section using the Science and Technical Reports Ask-a-Librarian form . Some standards, however, are housed in the Library's general collections and discoverable by searching the  online catalog -- the ASTM standards are one example. Other standards are in custody of appropriate specialized research centers, such as the Law Library , which maintains  OSHA standards and some building codes.

About the Science Section

Part of the  Science & Business Reading Room  at the Library of Congress, the Science Section is the starting point for conducting research at the Library of Congress in the subject areas of science, medicine and engineering. Here, reference specialists in specific subject areas of science and engineering  assist patrons in formulating search strategies and gaining access to the information and materials contained in the Library's rich collections of science, medicine, and engineering materials.

• Next: Technical Reports Collections >>
• Last Updated: Jul 3, 2024 11:51 AM
• URL: https://guides.loc.gov/technical-reports

share this!

August 21, 2024

This article has been reviewed according to Science X's editorial process and policies . Editors have highlighted the following attributes while ensuring the content's credibility:

fact-checked

trusted source

written by researcher(s)

A new 'AI scientist' can write science papers without any human input—here's why that's a problem

by Karin Verspoor, The Conversation

Scientific discovery is one of the most sophisticated human activities. First, scientists must understand the existing knowledge and identify a significant gap. Next, they must formulate a research question and design and conduct an experiment in pursuit of an answer. Then, they must analyze and interpret the results of the experiment, which may raise yet another research question.

Can a process this complex be automated? Last week, Sakana AI Labs announced the creation of an "AI scientist"—an artificial intelligence system they claim can make scientific discoveries in the area of machine learning in a fully automated way.

Using generative large language models (LLMs) like those behind ChatGPT and other AI chatbots, the system can brainstorm, select a promising idea, code new algorithms, plot results, and write a paper summarizing the experiment and its findings, complete with references. Sakana claims the AI tool can undertake the complete lifecycle of a scientific experiment at a cost of just US$15 per paper—less than the cost of a scientist's lunch.

These are some big claims. Do they stack up? And even if they do, would an army of AI scientists churning out research papers with inhuman speed really be good news for science ?

How a computer can 'do science'

A lot of science is done in the open, and almost all scientific knowledge has been written down somewhere (or we wouldn't have a way to "know" it). Millions of scientific papers are freely available online in repositories such as arXiv and PubMed .

LLMs trained with this data capture the language of science and its patterns. It is therefore perhaps not at all surprising that a generative LLM can produce something that looks like a good scientific paper—it has ingested many examples that it can copy.

What is less clear is whether an AI system can produce an interesting scientific paper. Crucially, good science requires novelty.

But is it interesting?

Scientists don't want to be told about things that are already known. Rather, they want to learn new things, especially new things that are significantly different from what is already known. This requires judgment about the scope and value of a contribution.

The Sakana system tries to address interestingness in two ways. First, it "scores" new paper ideas for similarity to existing research (indexed in the Semantic Scholar repository). Anything too similar is discarded.

Second, Sakana's system introduces a "peer review" step—using another LLM to judge the quality and novelty of the generated paper. Here again, there are plenty of examples of peer review online on sites such as openreview.net that can guide how to critique a paper. LLMs have ingested these, too.

AI may be a poor judge of AI output

Feedback is mixed on Sakana AI's output. Some have described it as producing " endless scientific slop ."

Even the system's own review of its outputs judges the papers weak at best. This is likely to improve as the technology evolves, but the question of whether automated scientific papers are valuable remains.

The ability of LLMs to judge the quality of research is also an open question. My own work (soon to be published in Research Synthesis Methods ) shows LLMs are not great at judging the risk of bias in medical research studies, though this too may improve over time.

Sakana's system automates discoveries in computational research, which is much easier than in other types of science that require physical experiments. Sakana's experiments are done with code, which is also structured text that LLMs can be trained to generate.

AI tools to support scientists, not replace them

AI researchers have been developing systems to support science for decades. Given the huge volumes of published research, even finding publications relevant to a specific scientific question can be challenging.

Specialized search tools make use of AI to help scientists find and synthesize existing work. These include the above-mentioned Semantic Scholar, but also newer systems such as Elicit , Research Rabbit , scite and Consensus .

Text mining tools such as PubTator dig deeper into papers to identify key points of focus, such as specific genetic mutations and diseases, and their established relationships. This is especially useful for curating and organizing scientific information.

Machine learning has also been used to support the synthesis and analysis of medical evidence, in tools such as Robot Reviewer . Summaries that compare and contrast claims in papers from Scholarcy help to perform literature reviews.

All these tools aim to help scientists do their jobs more effectively, not to replace them.

AI research may exacerbate existing problems

While Sakana AI states it doesn't see the role of human scientists diminishing, the company's vision of "a fully AI-driven scientific ecosystem" would have major implications for science.

One concern is that, if AI-generated papers flood the scientific literature, future AI systems may be trained on AI output and undergo model collapse . This means they may become increasingly ineffectual at innovating.

However, the implications for science go well beyond impacts on AI science systems themselves.

There are already bad actors in science, including "paper mills" churning out fake papers . This problem will only get worse when a scientific paper can be produced with US$15 and a vague initial prompt.

The need to check for errors in a mountain of automatically generated research could rapidly overwhelm the capacity of actual scientists. The peer review system is arguably already broken, and dumping more research of questionable quality into the system won't fix it.

Science is fundamentally based on trust. Scientists emphasize the integrity of the scientific process so we can be confident our understanding of the world (and now, the world's machines) is valid and improving.

A scientific ecosystem where AI systems are key players raises fundamental questions about the meaning and value of this process, and what level of trust we should have in AI scientists. Is this the kind of scientific ecosystem we want?

Explore further

Feedback to editors

How hardware contributes to the fairness of artificial neural networks

19 hours ago

Resin made from biomass advances method for manufacturing recyclable wind turbine blades

Aug 23, 2024

Sprayable gels could protect buildings during wildfires

Researchers propose a smaller, more noise-tolerant quantum factoring circuit for cryptography

Researchers unleash machine learning in designing advanced lattice structures

Aug 22, 2024

Redox-active metal-organic framework developed for Li batteries in freezing conditions

Flexible nanogenerator with enhanced power density could one day rival the power of solar panels

New method allows AI to learn indefinitely

Self-improving AI method increases 3D-printing efficiency

Scientists invent a hot-emitter transistor for future high-performance, low-power, multifunctional devices

Related stories.

'AI Scientist' model designed to conduct scientific research autonomously

Aug 15, 2024

New benchmarking tool evaluates the factuality of LLMs

Aug 21, 2024

Flood of 'junk': How AI is changing scientific publishing

Aug 10, 2024

Top science editor defends peer-review system in climate row

Sep 15, 2023

New tool detects fake, AI-produced scientific articles

Aug 5, 2024

AI researchers introduce an LLM capable of generating text outputs of up to 10,000 words

Aug 16, 2024

Recommended for you

The first tensor processor chip based on carbon nanotubes could lead to energy-efficient AI processing

Researchers train a robot dog to combat invasive fire ants

Let us know if there is a problem with our content.

Use this form if you have come across a typo, inaccuracy or would like to send an edit request for the content on this page. For general inquiries, please use our contact form . For general feedback, use the public comments section below (please adhere to guidelines ).

Please select the most appropriate category to facilitate processing of your request

Thank you for taking time to provide your feedback to the editors.

Your feedback is important to us. However, we do not guarantee individual replies due to the high volume of messages.

E-mail the story

Your email address is used only to let the recipient know who sent the email. Neither your address nor the recipient's address will be used for any other purpose. The information you enter will appear in your e-mail message and is not retained by Tech Xplore in any form.

Your Privacy

This site uses cookies to assist with navigation, analyse your use of our services, collect data for ads personalisation and provide content from third parties. By using our site, you acknowledge that you have read and understand our Privacy Policy and Terms of Use .

E-mail newsletter

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts

Publish with Scientific Reports

We're an open-access journal publishing rigorously peer-reviewed research from across the natural sciences, psychology, medicine and engineering.

Announcements

Journal Metrics

• *2023 Journal Citation Reports® Science Edition (Clarivate Analytics, 2024).

Calls for Papers

• The median time from submission to acceptance for papers submitted to our Guest Edited Collections is 102 days.

Top 100 Collections

• *Data obtained from SN Insights which is based on Digital Science's Dimensions.

• Engineering

We are delighted to announce that Scientific Reports is now expanding its scope to include engineering disciplines.

Advertisement

Browse articles

Physical sciences.

Effect of clays incorporation on properties of thermoplastic starch/clay composite bio-based polymer blends

• Soledad Cecilia Pech-Cohuo
• Mario Adrián de Atocha Dzul-Cervantes
• Yamile Pérez‑Padilla

Practical preparation of unsaturated very-long-chain fatty acids (VLCFAs) and very-long-chain alkene pollinator attractants

• Björn Bohman
• Aylin J. Bersch
• Philipp M. Schlüter

Experimental study of dual-cycle thermal management system for engineering radiator

• Wenbao Zhang
• Huimin Zhao

Computational investigation on physical properties of lead based perovskite RPbBr 3  (R = Cs, Hg, and Ga) materials for photovoltaic applications

• Muhammad Khuram Shahzad
• Shoukat Hussain
• Waqar Azeem

Rotation insensitive implantable wireless power transfer system for medical devices using metamaterial-polarization converter

• Tarakeswar Shaw
• Bappaditya Mandal
• Robin Augustine

Subjects within Physical sciences

• Astronomy and planetary science
• Energy science and technology
• Materials science
• Mathematics and computing
• Nanoscience and technology
• Optics and photonics

Earth and environmental sciences

The impact of the digital economy on land transfer-out decisions among Chinese farmers: evidence from CHFS micro-data

• Peijiang Zheng

Infertility-related stress is associated with quality of life through negative emotions among infertile outpatients

• Li-ping Shi
• Yao-guo Geng
• Jing-jing Gu

Spatial spillover effect and driving factors of urban carbon emissions in the Yellow River Basin using nighttime light data

• Mingjuan Ma
• Yumeng Wang

Research on food security issues considering changes in rainfall

• SiMan Jiang
• Shuyue Chen

Drought prediction using artificial intelligence models based on climate data and soil moisture

• Mhamd Saifaldeen Oyounalsoud
• Abdullah Gokhan Yilmaz
• Abdulrahman Abdeljaber

Subjects within Earth and environmental sciences

• Biogeochemistry
• Climate sciences
• Environmental sciences
• Environmental social sciences
• Natural hazards
• Ocean sciences
• Planetary science
• Solid Earth sciences
• Space physics

Biological sciences

Mesenchymal stem cells from adipose tissue prone to lose their stemness associated markers in obesity related stress conditions

• Sura Hilal Ahmed Al-Sammarraie
• Şerife Ayaz-Güner
• Servet Özcan

Risk assessment and transmission of fluoroquinolone resistance in drug-resistant pulmonary tuberculosis: a retrospective genomic epidemiology study

• Vijayalakshmi Jawaharlal Nehru
• Maria Jose Vandakunnel
• Muthuraj Muthaiah

Self-reference and emotional reaction drive aesthetic judgment

• Sara Salgues
• Amélie Jacquot
• Marco Sperduti

Subjects within Biological sciences

• Biochemistry
• Biological techniques
• Biotechnology
• Cell biology
• Chemical biology
• Computational biology and bioinformatics
• Developmental biology
• Drug discovery
• Microbiology
• Molecular biology
• Neuroscience
• Plant sciences
• Structural biology
• Systems biology

Health sciences

Malnutrition stratified by marasmus and kwashiorkor in adult patients with heart failure

• Akiomi Yoshihisa
• Yasuchika Takeishi

Efficacy of various techniques in calcium silicate-based intracanal medicament removal: a micro-CT analysis

• Rahaf A. Almohareb
• Reem M. Barakat
• Hanan Balto

The association between dietary habits and self-care behavior of pregnant women with pregnancy complications

• Mehdi Karimi
• Maryam Mofidi Nejad
• Leila Azadbakht

Effect of nutrition education on the nutritional status of pregnant women in Robe and Goba Towns, Southeast Ethiopia, using a cluster randomized controlled trial

• Girma Beressa
• Susan J. Whiting
• Tefera Belachew

Predictive validity of resource-adjusted Korean Triage and Acuity Scale in pediatric gastrointestinal tract foreign body patients

• Jin Hee Lee
• Jin Hee Jung

Subjects within Health sciences

• Endocrinology
• Gastroenterology
• Health care
• Health occupations
• Medical research
• Molecular medicine
• Pathogenesis
• Rheumatology
• Risk factors
• Signs and symptoms

Trending - Altmetric

The influence of exploration activities of a potential lithium mine to the environment in Western Serbia

Patients recovering from COVID-19 who presented with anosmia during their acute episode have behavioral, functional, and structural brain alterations

Estimating the prevalence of Non-Verbal Learning Disability (NVLD) from the ABCD sample

Daily supplementation with the Lab4P probiotic consortium induces significant weight loss in overweight adults

Science jobs, senior researcher-experimental leukemia modeling, mullighan lab.

Memphis, Tennessee

St. Jude Children's Research Hospital (St. Jude)

Assistant or Associate Professor (Research-Educator)

The Center for Molecular Medicine and Genetics in the Wayne State University School of Medicine (http://genetics.wayne.edu/) is expanding its high-...

Detroit, Michigan

Wayne State University

Postdoctoral Fellow – Cancer Immunotherapy

Tampa, Florida

H. Lee Moffitt Cancer Center & Research Institute

Postdoctoral Associate - Specialist

Houston, Texas (US)

Baylor College of Medicine (BCM)

Postdoctoral Associate- CAR T Cells, Synthetic Biology

This journal is a member of and subscribes to the principles of the Committee on Publication Ethics.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

China beats US in top global scientific ‘hot papers’ ranking: report

• The number of ‘hot papers’ – those that receive instant recognition – published by Chinese scientists has topped a global list
• China also came first in six major disciplines on a chart tracking international citations of scientific papers

A key US-China science cooperation deal is about to end

Hot papers are often measured in bimonthly periods rather than years. They need to be no more than two years old and have citations among the top 0.1 per cent in their field over the past two months.

In comparison, another indicator known as highly cited papers are based on 10 years of publication data, and refer to the top 1 per cent of papers which often lead to fundamental breakthroughs in their field.

In terms of highly cited papers, the report found China remained in second place with 57,900 papers, while the US took the lead with a total of 76,600 highly cited papers. This amounted to 41 per cent of the global total.

The report also noted a significant increase in the proportion of high-quality, high-influence Chinese papers published in domestic, rather than Western, journals.

“The impact of China’s core scientific and technological journals continues to grow, and their ability to attract high-level papers has been going up steadily,” the report said.

Last year, domestic journals published nearly 42,000 research papers with regard to national major projects and key research and development programmes, covering a wide range of disciplines including clinical medicine, agriculture, environmental sciences, electronics, communications and earth sciences, the report said.