Keith Duggar

Also known as Keith H. Duggar, Keith Howard Duggar

Keith Duggar is an AI podcast co-interviewer and technology executive whose public work spans applied research, language models and AI alignment. He holds a PhD from MIT and coauthored a flow-cytometry model of mouse embryonic stem-cell population dynamics in 2004. On Machine Learning Street Talk, he discussed POET and GPT-3 in 2020, joined a Noam Chomsky panel on language models in 2022, and explored the fractured entangled representation hypothesis with Kenneth Stanley in 2025. By May 2023, he and Tim Scarfe were joint CTOs of XRAI Glass, whose app combines extended reality and AI for real-time transcription and translation; in October 2024, he debated Liron Shapira about Turing machines, finite automata and LLM limits. As of October 2026, he is XRAI Glass's CTO and often co-interviews guests on Machine Learning Street Talk.

Now at: XRAI Glass

Timeline

By 2026

A PhD at MIT

By 2026, he holds a PhD from MIT.
2004

Modeling stem cell populations

In November 2004, he coauthors with Wendy Prudhomme and Douglas Lauffenburger a model for mouse embryonic stem-cell population dynamics. It uses flow-cytometry marker distributions to distinguish rates of commitment to differentiation, proliferation of differentiated cells and self-renewal.
2020

POET and GPT-3 episodes

In April 2020, he appears as a special guest in a Machine Learning Street Talk episode about open-ended algorithms and POET, introduced as a new YouTube channel by Tim Scarfe, Connor Shorten and Yannic Kilcher. In November 2020, he joins Scarfe and Kilcher to discuss GPT-3 with Gary Marcus and Connor Leahy.
July 2022

A panel with Noam Chomsky

In July 2022, he joins Tim Scarfe and Walid Saba on an MLST panel with Noam Chomsky. They discuss large language models, connectionism and Yann LeCun's position paper on AGI.
May 2023

Joint CTOs at XRAI Glass

In May 2023, he and Tim Scarfe interview Robert Miles about AI alignment as joint CTOs of XRAI Glass. Founded in 2022, XRAI Glass combines extended reality and AI in an app for real-time transcription and translation.
October 2024

Turing machines and LLM limits

In October 2024, he debates Liron Shapira on whether his distinction between Turing machines and discrete finite automata helps explain limits of current LLMs. They compare views on alignment, orthogonality and instrumental convergence.
July 2025

Fractured entangled representations

In July 2025, he joins Tim Scarfe and Kenneth Stanley in an MLST episode about the fractured entangled representation hypothesis, which describes AI trained with brute-force SGD as producing impressive results with a disordered internal understanding. He compares formula memorization in his high-school physics classes with deriving answers through calculus.
As of October 2026

CTO and MLST co-interviewer

His career includes stints at IBM Research, Wall Street and Microsoft. In September 2025, he co-hosts an MLST episode with NYU's Andrew Gordon Wilson; in May 2026, he spends about eighty minutes pushing back on Brad Carson, who heads Americans for Responsible Innovation. As of October 2026, he is chief technology officer of XRAI Glass and often co-interviews guests on Machine Learning Street Talk.