Eric Horvitz

Career

Before moving into the role of Chief Scientific Officer, Horvitz served as director of Microsoft Research, overseeing its research labs. As Microsoft's Chief Scientific Officer, Horvitz now leads initiatives that connect science, technology and society, with emphases on AI, the biosciences and healthcare.¹²

Horvitz founded and chairs Microsoft's Aether committee on AI, effects and ethics in engineering and research. Outside Microsoft, Horvitz established the One Hundred Year Study on AI (AI100) at Stanford University, co-founded the Partnership on AI and served as its board chair, served as president of AAAI, and sits on the advisory boards of the Allen Institute for AI and Stanford's Institute for Human-Centered AI.¹²

Research

Horvitz's research has used probability and utility in machine learning and reasoning, developed models of bounded rationality, and built systems that perceive and act by interpreting streams of multisensory information. Horvitz has also written on principles of human-AI collaboration, and the work has led to fielded systems in healthcare, transportation, e-commerce, operating systems and aerospace.¹²

In 2023, Horvitz co-authored studies of large language models in medicine. One evaluated GPT-4 on medical competency examinations and benchmark datasets, and the Medprompt paper showed how prompting strategies can unlock specialist capabilities in a generalist foundation model on medical benchmarks.³⁴

Work and contributions

• Can Generalist Foundation Models Outcompete Special-Purpose Tuning? (Medprompt) Published research

  Co-author of the 2023 paper studying whether prompting strategies can steer a generalist foundation model to perform specialist medical question answering.⁴

• Capabilities of GPT-4 on Medical Challenge Problems Published research

  Co-author of the 2023 evaluation of GPT-4 on medical competency examinations and benchmark datasets.³

Timeline

1. 2014Wrote the framing memo for the One Hundred Year Study on Artificial Intelligence (AI100).²
2. Nov 2022Delivered a Tanner Lecture titled "Humanity and its Tools of Reason and Imagination".²
3. Mar 2023Co-authored "Capabilities of GPT-4 on Medical Challenge Problems".³
4. May 2023Published the AI Anthology on AI and the future of human flourishing.²
5. Nov 2023Co-authored the Medprompt paper on steering generalist foundation models for medicine.⁴

Other documented roles

Roles the sources document that are not confirmed as current.

• Director of Microsoft Research, Microsoft Research¹
• Co-founder and former board chair, Partnership on AI¹

Honors and recognition

• Member of the National Academy of Engineering · National Academy of Engineering¹
• ACM Fellow · Association for Computing Machinery¹
• AAAI Fellow · Association for the Advancement of Artificial Intelligence¹
• AAAS Fellow · American Association for the Advancement of Science¹
• Fellow of the American Academy of Arts and Sciences · American Academy of Arts and Sciences¹
• Fellow of the American College of Medical Informatics · American College of Medical Informatics¹
• Member of the American Philosophical Society · American Philosophical Society¹

Where they work

• Microsoft 75 people in the wiki

Related people

• Margaret MitchellAlso at Microsoft
• Su Lin BlodgettAlso at Microsoft
• Christopher BishopAlso at Microsoft
• Pushmeet KohliAlso at Microsoft
• Peter LeeAlso at Microsoft
• Eric Michael SmithSafety and alignment · Evaluation and benchmarks
• Jacob SteinhardtSafety and alignment · Evaluation and benchmarks
• Peter MattsonEvaluation and benchmarks · Safety and alignment
• Maarten SapSafety and alignment · Evaluation and benchmarks
• Beth BarnesEvaluation and benchmarks · Safety and alignment