PeopleofSuperintelligence

Researcher

Ashish Sabharwal

Ashish Sabharwal is an AI2 researcher whose work spans automated reasoning, science question answering, and language models. He earned a Bachelor of Technology at IIT Kanpur in 1998, completed MS and PhD degrees at the University of Washington in 2005, and joined Cornell as a postdoctoral associate. His 2006 AAAI paper with Carla Gomes and Bart Selman on model counting won an Outstanding Paper Award; he worked at IBM Watson from 2010 to 2014 and joined AI2 in May 2014. His projects included ARC, Aristo, and UnifiedQA, which address science questions and question answering across formats. In 2025, he coauthored Olmo 3, an open family of language models; he was a principal research scientist in 2024 and remains on AI2's team as of October 2026.

Works
260
Citations
8,833
h-index
46
Badges

Publication numbers from OpenAlex, as of 18 October 2026.

Timeline

  1. 1994–2005

    Degrees at IIT and Washington

    He earns a Bachelor of Technology at IIT Kanpur in 1998 and completes MS and PhD degrees at the University of Washington in 2005.

  2. 2005–2008

    Postdoctoral associate at Cornell

    He works as a postdoctoral associate in computer science at Cornell from 2005 through 2008.

  3. 2006

    An AAAI model-counting award

    His paper with Carla P. Gomes and Bart Selman, “Model Counting: A New Strategy for Obtaining Good Bounds,” wins an Outstanding Paper Award at AAAI in 2006.

  4. October 2010–April 2014

    Research scientist at IBM Watson

    He works as a research scientist at IBM Watson Research Center from October 2010 through April 2014.

  5. May 2014

    Senior research scientist at AI2

    He joins AI2 as a senior research scientist in May 2014.

  6. 2018–2020

    ARC, Aristo and UnifiedQA

    In 2018, he coauthors ARC, the AI2 Reasoning Challenge, a question set and text corpus for advanced question answering and reasoning research. In 2019, he coauthors an Aristo report on a system that scores more than 90% on Grade 8 New York Regents multiple-choice science questions without diagrams. In 2020, he coauthors UnifiedQA, a single pre-trained QA model that performs well across 19 datasets spanning four formats.

  7. 2025

    Coauthoring Olmo 3

    In 2025, he coauthors Olmo 3, an open family of 7B and 32B language models whose release includes model stages, checkpoints, data, and dependencies.

  8. As of October 2026

    AI2 research team member

    In 2024, he serves as a Principal Research Scientist at AI2. As of October 2026, he is part of the institute's team.

Career

  1. Since 2014Allen Institute for AI (Ai2)Senior research scientist, Principal Research Scientist
  2. 2021–2022University of Washington
  3. 2011–2014IBM
  4. 2010–2014IBM ResearchResearch scientist
  5. 2005–2008Cornell UniversityPostdoctoral associate

Education

  1. 2005University of WashingtonMS
  2. 1998Indian Institute of Technology KanpurBachelor of Technology

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering2018 · Conference on Empirical Methods in Natural Language Processing (EMNLP) · 666 citations
  2. Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models2022 · 569 citationsWith Abhinav Rastogi, Abubakar Abid, Andrea Madotto, Chris Callison-Burch, Christopher D. Manning and Christopher Potts
  3. UNIFIEDQA: Crossing Format Boundaries with a Single QA System2020 · 511 citations
  4. Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge2018 · 461 citationsWith Oren Etzioni
  5. SciTaiL: A Textual Entailment Dataset from Science Question Answering2018 · Proceedings of the AAAI Conference on Artificial Intelligence · 416 citations

From OpenAlex, as of 18 October 2026.

  1. The company share of AI researchers’ top papers rose from 3% to 32%Includes Allen Institute for AI (Ai2). We estimate where leading AI researchers worked when their most-cited papers appeared. The company share rose from 3% in 1975–1979 to 32% in 2020–2024.
  2. Google has the highest combined h-index among AI organizationsIncludes Allen Institute for AI (Ai2). Our estimates of combined and median h-index at AI organizations, using current affiliations and OpenAlex records.
  3. 44% of AI career moves go from one university to anotherIncludes Allen Institute for AI (Ai2). We estimate how 12,774 AI career moves link universities, companies, labs and government research centers, with net flows and the most common paths.
  4. Google alumni founded or led 82 other organizations, the most among the labs we compareIncludes IBM. We trace 189 organizations founded or led by AI lab alumni, and 196 career moves between the labs.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Ashish Sabharwal yet. Send us a free-licensed photo.

  1. LCI Forum Talk cs.ubc.ca. Checked 18 October 2026.
  2. Ashish Sabharwal's Homepage at Cornell University cs.cornell.edu. Checked 18 October 2026.
  3. Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge arxiv.org. Checked 18 October 2026.
  4. From 'F' to 'A' on the N.Y. Regents Science Exams: An Overview of the Aristo Project arxiv.org. Checked 18 October 2026.
  5. UNIFIEDQA: Crossing Format Boundaries with a Single QA System aclanthology.org. Checked 18 October 2026.
  6. Olmo 3 arxiv.org. Checked 18 October 2026.
  7. Organizers | PERSONALIZE @ EACL 2024 genpersonalization.github.io. Checked 18 October 2026.
  8. Team Directory | Ai2 allenai.org. Checked 18 October 2026.