Empowering biomedical discovery with AI agents

Shanghua Gao; Ada Fang; Yepeng Huang; Valentina Giunchiglia; Ayush Noori; Jonathan Richard Schwarz; Yasha Ektefaie; Jovana Kondic; Marinka Žitnik

doi:10.1016/j.cell.2024.09.022

AI summary · DOI: 10.1016/j.cell.2024.09.022 · Medicine

Empowering biomedical discovery with AI agents

Shanghua Gao (Harvard University), Ada Fang (Harvard University), Yepeng Huang (Harvard University), Valentina Giunchiglia (Harvard University, Imperial College London), Ayush Noori (Harvard University), Jonathan Richard Schwarz (Harvard University), Yasha Ektefaie (Harvard University), Jovana Kondic (Massachusetts Institute of Technology),

and 1 more author

Marinka Žitnik (Broad Institute, Harvard University)

Original paper Published on 01 Oct. 2024 Cell License: CC BY

How to read this summary? Illustrated guide

The essentials

The authors envision collaborative AI agents that combine AI capabilities and biomedical tools with human expertise to support biomedical discovery.

Based on the abstract only: check the details in the original paper.

Question

How might AI agents contribute to biomedical research without removing humans from the discovery process? What capabilities and biomedical applications could these systems have?

Method

This paper presents a vision for biomedical AI agents rather than reporting a specific experiment or dataset. It describes agents that combine large language and generative models, structured memory, machine learning tools, and experimental platforms to plan workflows, analyze data, and assess gaps in their knowledge.

Findings

Authors’ interpretation The authors propose that biomedical AI agents should collaborate with human researchers rather than replace them.
See the passage in the paper
Rather than taking humans out of the discovery process, biomedical AI agents combine human creativity and expertise with AI's ability to analyze large datasets, navigate hypothesis spaces, and execute repetitive tasks.
Authors’ interpretation The envisioned agents would plan discovery workflows and assess their own knowledge gaps.
See the passage in the paper
AI agents are poised to be proficient in various tasks, planning discovery workflows and performing self-assessment to identify and mitigate gaps in their knowledge.
Authors’ interpretation The proposed systems would use structured memory and machine learning tools to support learning and incorporate scientific knowledge.
See the passage in the paper
These agents use large language models and generative models to feature structured memory for continual learning and use machine learning tools to incorporate scientific knowledge, biological principles, and theories.
Authors’ interpretation The authors identify several potential areas of impact, including cell simulation, phenotype control, cellular circuit design, and therapy development.
See the passage in the paper
AI agents can impact areas ranging from virtual cell simulation, programmable control of phenotypes, and the design of cellular circuits to developing new therapies.

Limitations

The paper offers a vision of possible agent capabilities and applications, but the provided text does not report empirical evaluations of them.

Why it matters

The authors argue that collaborative AI agents could expand biomedical discovery by combining human creativity and expertise with AI's capacity to analyze data, explore hypotheses, and perform repetitive tasks. They frame these systems as a potential aid across diverse areas of biomedical research.

Written by AI from the paper. Passages are checked word for word against the text we analysed; the rest may contain errors.

AI-generated keywords: AI agents biomedical discovery human–AI collaboration experimental platforms scientific reasoning

Read further

Comprehensive Summary
Layman's Summary
Blog article
Original abstract

The paper presents a vision for integrating increasingly capable AI systems into biomedical research. Its central proposal is to develop “AI scientists”: collaborative systems that combine AI models and biomedical tools with experimental platforms while keeping human researchers involved. Rather than replacing scientists, these agents would complement human creativity and expertise by processing large datasets, exploring broad hypothesis spaces, and performing repetitive tasks. The authors outline a possible approach rather than report a specific experiment. In their framework, agents would plan discovery workflows, use large language and generative models, and maintain structured memories to support continual learning. They would also draw on machine-learning tools to incorporate scientific knowledge, biological principles, and theories. Another proposed capability is skeptical reasoning and self-assessment: agents should identify gaps or uncertainty in their knowledge and take steps to address them, rather than merely generate plausible answers. The proposed systems would collaborate with researchers and connect computational reasoning to experimental work. Potential applications include virtual cell simulation, programmable control of cellular phenotypes, cellular-circuit design, and therapy development. Across these areas, agents could help organize and carry out complex discovery processes by proposing and evaluating possibilities, coordinating tools and analyses, and assisting with laborious or repetitive tasks. The broader aim is to improve the capabilities and efficiency of biomedical research by combining the complementary strengths of humans and AI. The supplied text contains only the abstract and reports no experiments, quantitative findings, or numerical results. Consequently, it provides no performance measures, sample sizes, comparisons, or effect estimates. The article should therefore be understood as a perspective on future research, not evidence that an AI-agent system has already achieved the proposed capabilities. The authors argue that AI agents could become valuable collaborators in biomedical discovery if they can integrate models, tools, scientific knowledge, and experimental platforms while planning work and assessing their own limitations. Their vision emphasizes human–AI collaboration over fully autonomous research. However, the abstract does not provide enough detail to evaluate challenges such as reliability, validation, safety, accountability, or practical deployment. Nor does it specify how skeptical reasoning, continual memory, or self-assessment would be measured or demonstrated. These capabilities and applications are presented as research directions, not established results.

Empowering Biomedical Discovery with AI Agents

Biomedical research increasingly involves analyzing vast datasets, connecting evidence across fields, and testing complex ideas through experiments. In “Empowering biomedical discovery with AI agents,” Shanghua Gao, Ada Fang, Yepeng Huang, Valentina Giunchiglia, Ayush Noori, Jonathan Richard Schwarz et al. present a vision for how AI might help meet these demands. Their proposal centers on “AI scientists”: collaborative systems that combine AI models and computational tools with experimental platforms, while keeping human researchers involved.

The paper is a perspective on a possible future for biomedical research, not a report of a completed experiment. It outlines what such systems might do and the capabilities they would need. The central idea is not to replace scientists, but to pair human creativity and expertise with AI’s ability to process large datasets, explore many possible hypotheses, and handle repetitive tasks.

What an AI scientist might do

In the authors’ framework, an AI agent would help organize and carry out a research workflow. It could help plan a sequence of discovery tasks, use language and generative models, and coordinate computational tools and analyses. Structured memory could help the agent retain useful information and support learning over time, rather than treating each research question as entirely separate.

The proposed agents would also draw on machine-learning tools that incorporate scientific knowledge, biological principles, and theories. This is important because biomedical discovery requires more than producing plausible text or identifying patterns in data. Useful research assistance must be grounded in what is known about biology and able to connect computational ideas to experimental work.

Another proposed capability is skeptical reasoning. Instead of presenting every answer with equal confidence, an agent should recognize gaps or uncertainty in its knowledge and take steps to address them. That could mean seeking additional information, suggesting an analysis, or helping researchers decide what experiment might clarify an unresolved question. The goal is an assistant that can assess the limits of its own conclusions, not simply generate convincing-sounding responses.

Working with researchers and experiments

Human involvement is central to the vision. Researchers would contribute scientific judgment, creativity, and knowledge of experimental context, while AI agents could help manage broad searches, coordinate tools, and take on laborious work. This division of effort could make it easier to explore complicated questions without expecting an agent to independently direct an entire research program.

Connecting computational reasoning to experimental platforms is another key part of the proposal. An agent might help turn a hypothesis into a plan involving analyses and experiments, then support researchers as they evaluate the resulting evidence. In this model, AI is part of a larger discovery process: it helps generate and assess possibilities, while people remain involved in decisions and interpretation.

Possible areas of application

The paper points to several areas where AI agents could eventually assist biomedical research:

Virtual cell simulation, to help researchers explore how cells behave under different conditions.
Programmable control of cellular phenotypes, supporting investigations into how cellular characteristics might be changed.
Cellular-circuit design, where computational tools could help develop and evaluate biological systems.
Therapy development, by helping organize and assess complex discovery workflows.

Across these areas, the proposed contribution is not a single automated breakthrough. Agents could instead help researchers coordinate analyses, evaluate alternatives, connect relevant knowledge, and manage repetitive steps. If effective, this kind of support might improve the capacity and efficiency of biomedical research.

A vision, not a demonstrated system

The distinction between an ambitious proposal and demonstrated results matters. The supplied paper summary reports no experiments, quantitative findings, sample sizes, comparisons, performance measures, or effect estimates. It therefore does not show that an AI-agent system has already achieved the capabilities described. The suggested applications and features should be understood as research directions, not established outcomes.

The summary also leaves important questions open. It does not explain how reliability, validation, safety, accountability, or practical deployment would be handled. Nor does it specify how researchers could measure whether an agent truly reasons skeptically, learns continually, or accurately assesses its own limitations. These are central challenges: a system used in scientific work must be judged not only by how much it can do, but also by whether its outputs can be checked and trusted.

The promise of collaboration

“Empowering biomedical discovery with AI agents” offers a framework for thinking about how AI could become a more active collaborator in biomedical science. Its promise rests on bringing together models, tools, scientific knowledge, memory, and experimental platforms, while preserving meaningful human participation. The proposal is compelling as a direction for future research, but its practical value will depend on whether these capabilities can be built, tested, and validated in real research settings.

For now, the paper is best read as a vision rather than evidence of a working AI scientist. It invites researchers to consider how AI could help explore complex questions and reduce routine burdens, while making clear—through the open questions its framework raises—that reliable collaboration will require careful development and evaluation.

Created on 03 Nov. 2026

Was this synthesis useful?

Score: 0

Ask questions about this paper to our AI assistant

How to chat with this paper? Illustrated guide

You can also chat with multiple papers at once here.

AI assistant instructions?

Tools for this paper

Explain it differently 3 free rewrites left today

Save to a collection Group papers in collections to compare them, chat with them and export them. Create a free account Log in

Export BibTeX · RIS (Zotero, Mendeley, EndNote) · Obsidian (Markdown) · Notion (Markdown)

Advanced tools Premium

For researchers who follow a field closely: citation alerts, journal-club slides, version comparison and Zotero export. Discover Premium

Zotero

Citation alerts Citation alerts guide

Who works on it? The people and institutions who publish on its topic

Journal club Premium+ Slides to present it Propose for the journal club

Talk about it in public Premium+ Press kit: release, questions, hype check Press kit guide

Papers around this one

See them on a map

Ranked by how close their title and abstract are to this paper’s. Select some to read them together.

Closest papers

Very close Review
Agentic AI for Scientific Discovery: A Survey of Progress, Challenges, and Future Directions
Mourad Gridach, Jay Nanavati, Khaldoun Zine El Abidine et al. · 2025 · Computation and Language · preprint, not peer reviewed · cited 128 times · AI summary available

Agentic AI for Scientific Discovery surveys agent systems, tools, evaluation, and challenges across biology and other scientific fields.

Abstract
The integration of Agentic AI into scientific discovery marks a new frontier in research automation. These AI systems, capable of reasoning, planning, and autonomous decision-making, are transforming how scientists perform literature review, generate hypotheses, conduct experiments, and analyze results. This survey provides a comprehensive overview of Agentic AI for scientific discovery, categorizing existing systems and tools, and highlighting recent progress across fields such as chemistry, biology, and materials science. We discuss key evaluation metrics, implementation frameworks, and commonly used datasets to offer a detailed understanding of the current state of the field. Finally, we …
Very close Another approach to the problem
The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery
Chris Lu, Cong Lu, Robert Tjarko Lange et al. · 2024 · Artificial Intelligence · preprint, not peer reviewed · cited 1,355 times · AI summary available

The AI Scientist presents a framework for fully autonomous research, whereas this paper envisions agents collaborating with human biomedical researchers.

Abstract
One of the grand challenges of artificial general intelligence is developing agents capable of conducting scientific research and discovering new knowledge. While frontier models have already been used as aids to human scientists, e.g. for brainstorming ideas, writing code, or prediction tasks, they still conduct only a small part of the scientific process. This paper presents the first comprehensive framework for fully automatic scientific discovery, enabling frontier large language models to perform research independently and communicate their findings. We introduce The AI Scientist, which generates novel research ideas, writes code, executes experiments, visualizes results, describes its …
Close Broader context
Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems
Bang Liu, Xinfeng Li, Jiayi Zhang et al. · 2025 · Artificial Intelligence · preprint, not peer reviewed · cited 178 times · AI summary available

Advances and Challenges in Foundation Agents surveys agent architectures and challenges across domains rather than focusing on biomedical discovery.

Abstract
The advent of large language models (LLMs) has catalyzed a transformative shift in artificial intelligence, paving the way for advanced intelligent agents capable of sophisticated reasoning, robust perception, and versatile action across diverse domains. As these agents increasingly drive AI research and practical applications, their design, evaluation, and continuous improvement present intricate, multifaceted challenges. This survey provides a comprehensive overview, framing intelligent agents within a modular, brain-inspired architecture that integrates principles from cognitive science, neuroscience, and computational research. We structure our exploration into four interconnected parts.…
Close Review
Transforming Science with Large Language Models: A Survey on AI-assisted Scientific Discovery, Experimentation, Content Generation, and Evaluation
Steffen Eger, Yong Cao, Jennifer D'Souza et al. · 2025 · Computation and Language · cited 65 times · AI summary available

Transforming Science with Large Language Models surveys AI support across research tasks, including experimentation and content generation.

Abstract
With the advent of large multimodal language models, science is now at a threshold of an AI-based technological transformation. Recently, a plethora of new AI models and tools has been proposed, promising to empower researchers and academics worldwide to conduct their research more effectively and efficiently. This includes all aspects of the research cycle, especially (1) searching for relevant literature; (2) generating research ideas and conducting experimentation; generating (3) text-based and (4) multimodal content (e.g., scientific figures and diagrams); and (5) AI-based automatic peer review. In this survey, we provide an in-depth overview over these exciting recent developments, whic…
Close Shared data, tool or method
DISCOVERYWORLD: A Virtual Environment for Developing and Evaluating Automated Scientific Discovery Agents
Peter Jansen, Marc-Alexandre Côté, Tushar Khot et al. · 2024 · Artificial Intelligence · cited 128 times · AI summary available

DISCOVERYWORLD provides a virtual benchmark for evaluating agents on end-to-end scientific discovery, rather than proposing biomedical agent capabilities.

Abstract
Automated scientific discovery promises to accelerate progress across scientific domains. However, developing and evaluating an AI agent's capacity for end-to-end scientific reasoning is challenging as running real-world experiments is often prohibitively expensive or infeasible. In this work we introduce DISCOVERYWORLD, the first virtual environment for developing and benchmarking an agent's ability to perform complete cycles of novel scientific discovery. DISCOVERYWORLD contains a variety of different challenges, covering topics as diverse as radioisotope dating, rocket science, and proteomics, to encourage development of general discovery skills rather than task-specific solutions. DISCOV…
Close Same method, another problem
DrugAgent: Explainable Drug Repurposing Agent with Large Language Model-based Reasoning
Yoshitaka Inoue, Tianci Song, Tianfan Fu · 2024 · Artificial Intelligence · AI summary available

DrugAgent uses specialized AI agents, machine learning, and biomedical knowledge sources to identify drug-repurposing opportunities.

Abstract
Drug repurposing offers a promising avenue for accelerating drug development by identifying new therapeutic potentials of existing drugs. In this paper, we propose a multi-agent framework to enhance the drug repurposing process using state-of-the-art machine learning techniques and knowledge integration. Our framework comprises several specialized agents: an AI Agent trains robust drug-target interaction (DTI) models; a Knowledge Graph Agent utilizes the drug-gene interaction database (DGIdb), DrugBank, Comparative Toxicogenomics Database (CTD), and Search Tool for Interactions of Chemicals (STITCH) to systematically extract DTIs; and a Search Agent interacts with biomedical literature to an…
Close Broader context
AI and Generative AI for Research Discovery and Summarization
Mark Glickman, Yi Zhang · 2024 · Computation and Language · cited 53 times · Not summarized yet

AI and Generative AI for Research Discovery and Summarization discusses AI tools for literature search and summarization, not experimental biomedical discovery.

Abstract
AI and generative AI tools, including chatbots like ChatGPT that rely on large language models (LLMs), have burst onto the scene this year, creating incredible opportunities to increase work productivity and improve our lives. Statisticians and data scientists have begun experiencing the benefits from the availability of these tools in numerous ways, such as the generation of programming code from text prompts to analyze data or fit statistical models. One area that these tools can make a substantial impact is in research discovery and summarization. Standalone tools and plugins to chatbots are being developed that allow researchers to more quickly find relevant literature than pre-2023 sear…
Close Same method, another problem
Paper2Agent: Reimagining Research Papers As Interactive and Reliable AI Agents
Jiacheng Miao, Joe R. Davis, Yaohui Zhang et al. · 2025 · Artificial Intelligence · preprint, not peer reviewed · cited 34 times · AI summary available

Paper2Agent uses multiple agents to turn research papers and their code into interactive assistants, applying AI agents to research reuse.

Abstract
We introduce Paper2Agent, an automated framework that converts research papers into AI agents. Paper2Agent transforms research output from passive artifacts into active systems that can accelerate downstream use, adoption, and discovery. Conventional research papers require readers to invest substantial effort to understand and adapt a paper's code, data, and methods to their own work, creating barriers to dissemination and reuse. Paper2Agent addresses this challenge by automatically converting a paper into an AI agent that acts as a knowledgeable research assistant. It systematically analyzes the paper and the associated codebase using multiple agents to construct a Model Context Protocol (…

Go further

How are these papers found?

We compare the title and abstract of this paper with those of every paper summarized on this site, and with the papers Semantic Scholar recommends, those it cites and those citing it (for a much-cited paper, its most recent citations and, from OpenAlex, the most cited ones). Among the prior work and the follow-ups, the papers that use each other in depth, a method it uses and the much-cited follow-ups come a little higher. The levels (very close, close, related) come from the similarity of the texts, not from a judgment of quality; * marks a paper without an abstract, compared by its title. The sentence under a paper is written by an AI from the two abstracts: check it in the papers. A paper marked not relevant by three readers leaves the list.

Updated on 3 November 2026.

Disclaimer: The AI-based summarization tool and virtual assistant provided on this website may not always provide accurate and complete summaries or responses. We encourage you to carefully review and evaluate the generated content to ensure its quality and relevance to your needs.

Empowering biomedical discovery with AI agents

The essentials

Question

Method

Findings

Limitations

Why it matters

Read further

Definitions

Empowering Biomedical Discovery with AI Agents

What an AI scientist might do

Working with researchers and experiments

Possible areas of application

A vision, not a demonstrated system

The promise of collaboration

Ask questions about this paper to our AI assistant

Tools for this paper

Advanced tools Premium

Papers around this one

Closest papers

Go further

Cite this paper (BibTeX)