Peter Clark
Dr. Peter Clark is a Senior Research Director and founding member of the Allen Institute for AI (AI2), and also served as Interim CEO from 2022-2023. His research spans the areas of automated scientific discovery, knowledge representation and reasoning, commonsense reasoning, and natural language understanding. He co-leads AI2's Asta Project, a large-scale initiative developing agentic frameworks for both assisted and automated scientific discovery. He received his Ph.D. in 1991 and has worked in AI for over 35 years. He has published over 250 papers, and has received several awards, including five Best Paper awards (AAAI, EMNLPx3, AKBC), a Boeing Associate Technical Fellowship (2004), and Senior Membership of AAAI.
Publications
Recent Talks
- Towards Autonomous Scientific Discovery: Addressing the "Outer Loop" of Long-Horizon Research (slides) Invited talk at the Automated Scientific Discovery of Mind and Brain Workshop, Oct 2nd, 2025.
- Towards Semi-Automated Scientific Discovery. (slides) Invited talk at the NAACL Workshop on AI and Scientific Discovery, May 3, 2025.
- What do our Machines Dream Of? (slides). Invited talk at the University of Rochester, Nov 2024.
- Towards Integrated Scientific Discovery Systems (slides). Talk at the AAAI Fall Symposium on Integrated Approaches to Computational Scientific Discovery, Nov 2024.
- What do our Machines Believe? (slides). Invited talk at the ACL Workshop on Knowledgeable Language Models, Aug 2024.
- Navigating the Evolution: The Rising Tide of Large Language Models for AI and Education (slides). Keynote at AIED'24, July 2024.
- Structured Reasoning with Language (slides). Keynote at ESWC'24, May 2024.
- What just happened? Deciphering the sudden rise of AI (slides). Invited talk at the Rainier Club, Jan 2024.
- Invited talk at Google NY. Oct 2023.
- Small Steps to Grand Challenges: Lessons from the AI2 Reasoning Challenge (ARC) (slides, video (at minute 17:00)) Invited talk at the NAS Workshop on Scientific Discovery, Oct 2023.
- The Role of Structured Reasoning in the age of GPT (slides). Invited talk at Megagon Labs, Oct 2023.
- The role of NL reasoning in the age of GPT (slides, video). Keynote at the ACL Workshop on NL Reasoning and Structured Explanations, July 2023.
- What do our Machines Dream of? (slides, video). Invited Talk at Santa Fe Institute (SFI), March 2023.
- Systematic Reasoning with Language Models (slides). Keynote talk at the NeurIPS MATH-AI Workshop, Dec 2022.
- Towards Teachable Reasoning Systems (slides). Invited Seminar at Univ Notre Dame, Nov 2022.
- Systematic Reasoning with Language Models (slides, video). Keynote talk at AKBC, Oct 2021.
- Reasoning, Question-Answering, and Teachable Systems (slides). AI Class Lecture at Univ Pennsylvania, June 2022.
- Towards Teachable Reasoning Systems (slides, video). Invited webinar at AI Camp, April 2022.
- Systematic Reasoning and Explanation over Natural Language (slides). Invited Talk at Univ Pennsylvania, Feb 2022.
- Systematic Reasoning and Explanation over Natural Language (slides). Invited Talk at Univ Texas, Dec 2021.
- Invited Talk at Univ North Carolina (UNC) at Chapel Hill, Nov 2021.
- General-Purpose Question Answering with AI2's Macaw (slides, video). Invited webinar at AI Camp, Nov 2021.
Current Research: (Semi-)Automated Scientific Discovery
- The Inner Loop (Single Experiments)
- CodeScientist - End-to-End Semi-Automated Scientific Discovery with Code-based Experimentation (ACL'25)
- Panda - A plan-and-act discovery agent
- Data Voyager - Data-driven discovery and analysis (blog, ICML'24)
- The Outer Loop (Long-Horizon Research)
- AutoDS (Autonomous Discovery via Surprisal) - Open-ended Scientific Discovery via Bayesian Surprise (NeurIPS'25)
- Genesys - Discovery of new transformer architectures (NeurIPS'25)
- Asta - Our Interactive Research Agent
- Asta - Main Web Site
- Asta On-Line System - Interactive use of (a few) of our research tools
- Evaluation Benchmarks
- AstaBench - A comprehensive suite of benchmarks for science agents
- DiscoveryBench - A benchmark for data-driven discovery (ICLR'25)
- Discovery World - A simulated discovery environment (NeurIPS'24)
Additional Recent Research Areas
- Structured Reasoning with Language
- Language Models with Rationality - Architectures combining structured and neural reasoning (EMNLP'23)
- EntailmentBank and Entailer - Entailment-based reasoning (EMNLP'21, '22)
- Towards Teachable Reasoning Systems - interactive correction of an LLM's reasoning mistakes (EMNLP'22)
- RuleTaker, RuleTaker demo, ProofWriter - Transformers as soft reasoners over language (IJCAI'20, ACL'21)
- Agent-Based Systems
- CLIN - A Continually Learning Language Agent (ICLR'24)
- Skill Set Optimization - Dynamically learning skills in an interactive environment (ICML'24)
Contact
peterc@allenai.org