PeopleofSuperintelligence

Researcher

Eric Michael Smith

Also known as Eric Smith

Eric Michael Smith is a Meta research scientist whose work connects conversational AI, bias measurement, language models, and AI safety. He studied at Columbia and Princeton before joining Meta’s FAIR team. He coauthored work on BlendedSkillTalk, HolisticBias, BlenderBot 3, Llama 2, Llama Guard 3 Vision, and backtracking for safer text generation. Since April 2023, he has been a Research Scientist in Generative AI at Meta.

Timeline

  1. 2008–2015

    Research assistant at Columbia and Princeton

    He studies at Columbia and Princeton, serving as an undergraduate research assistant at Columbia from October 2008 to May 2010. At Princeton, he serves as a graduate research assistant from July 2010 to May 2015.

  2. 2018–2023

    Research Engineer at FAIR

    He joins FAIR at Meta as a Research Engineer in March 2018. In April 2023, he becomes a Research Scientist in Generative AI at Meta.

  3. 2020–2022

    BlendedSkillTalk, HolisticBias and BlenderBot

    In 2020, he coauthors a study that introduces BlendedSkillTalk to examine how conversational agents combine skills in natural dialogue. Its experiments compare training schemes and model architectures for blended conversation. In 2022, he coauthors HolisticBias, a dataset with nearly 600 descriptor terms across 13 demographic axes, and BlenderBot 3, a 175-billion-parameter dialogue model with internet access and long-term memory; its team releases weights and code and deploys a public webpage for conversation.

  4. 2023

    Llama 2 models

    In 2023, he coauthors Llama 2, a family of pretrained and fine-tuned language models from 7 billion to 70 billion parameters, including Llama 2-Chat models optimized for dialogue.

  5. 2024

    Llama Guard 3 Vision

    In November 2024, he coauthors Llama Guard 3 Vision, a multimodal safeguard for conversations involving image understanding. The model can classify multimodal prompts and responses to help safeguard those conversations.

  6. 2024–2026

    Backtracking and Muse Spark safety

    In 2024, he coauthors Backtracking Improves Generation Safety, which introduces a method for language models to recover from unsafe partial generations. In 2026, he coauthors Meta's Muse Spark Safety & Preparedness Report, which evaluates catastrophic-risk domains under Meta's Advanced AI Scaling Framework.

  7. Since 2023

    Research Scientist at Meta

    He continues as a Research Scientist in Generative AI at Meta in September 2026.

  1. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI from 1961 to 2025, with a year-by-year chart.
  2. The United States employs an estimated 53% of people in our data with a known employment countryOur estimates of where AI careers are concentrated now and how countries’ shares of dated positions changed over time.
  3. 46% of living people in our data with known birth and employment countries work outside their birth countryOur estimate of cross-border careers in AI, from country of birth and the country of each current organization.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionOur estimates of where the people shaping AI studied and where their careers led, from education records in cited biographies.

Sources and credits

Photo: We have no published photo of Eric Michael Smith yet. Send us a free-licensed photo.

  1. Eric Smith's email & phone number - Research Scientist, Generative AI at Meta | SignalHire signalhire.com. Checked 19 October 2026.
  2. Can You Put it All Together: Evaluating Conversational Agents’ Ability to Blend Skills - ACL Anthology aclanthology.org. Checked 19 October 2026.
  3. "I'm sorry to hear that": Finding New Biases in Language Models with a Holistic Descriptor Dataset arxiv.org. Checked 19 October 2026.
  4. BlenderBot 3: a deployed conversational agent that continually learns to responsibly engage arxiv.org. Checked 19 October 2026.
  5. Llama 2: Open Foundation and Fine-Tuned Chat Models arxiv.org. Checked 19 October 2026.
  6. Llama Guard 3 Vision: Safeguarding Human-AI Image Understanding Conversations arxiv.org. Checked 19 October 2026.
  7. Backtracking Improves Generation Safety arxiv.org. Checked 19 October 2026.
  8. Muse Spark Safety & Preparedness Report arxiv.org. Checked 19 October 2026.