Skip to content

Paige Bailey

Google DeepMind

  • 1 episode on Chain of Thought
  • 9 topics covered
  • 1 full transcript
  • Jan 2026 first on the show
Paige Bailey

At the time of the conversation, Paige Bailey was AI Developer Relations Lead at Google DeepMind, after starting as a geophysicist and helping ship GitHub Copilot. On Chain of Thought she broke down Google’s AI ecosystem, from Gemini 3 Pro and Flash to the open Gemma models and developer tools like Jules and Gemini CLI.

Cover art for Gemini 3 & Robot Dogs: Inside Google DeepMind's AI Experiments | Paige Bailey

Latest episode with Paige Bailey · EP 47

Gemini 3 & Robot Dogs: Inside Google DeepMind's AI Experiments | Paige Bailey

Paige Bailey, Google DeepMind ·

Show notes & transcript → All listening options →

Episodes

What the conversations cover

These notes summarize the episodes featuring Paige Bailey, including other speakers in panel conversations. Follow a source link to read the discussion in context.

EP 47 Gemini 3 & Robot Dogs: Inside Google DeepMind's AI Experiments | Paige Bailey

  • Gemini 3 Flash is priced at roughly 25% of Gemini 3 Pro on both input and output tokens — and a simple Python coding task costs less than a penny — making Flash the natural pick for latency-sensitive or high-volume tasks, with thinking optionally enabled when you need it. Episode takeaways
  • The Palm 2 model that preceded Gemini was built by a team of only “40 people, 42 ish people,” with Paige Bailey herself on a 20% assignment. The Gemini effort now takes multiple pages just to list the team — a direct illustration of what the Google Brain–DeepMind merger unlocked. Episode takeaways
  • Google’s open-source Gemma 3 family (27B params) and Gemma 3n (4B, with 1B–2B variants small enough to run in a browser) are available free via the Gemini API at ai.google.dev — and Paige recommends them explicitly for summarization, translation, and toxicity detection to offload the Gemini flagship models. Episode takeaways

In their words

Whereas grad students … obviously it's boring work, which is not the best, but it's also very, very easy for humans to make mistakes. So I think the more that AI is incorporated into the scientific process, both for designing experiments, but also executing on them, I think the better off we're all gonna be.

Paige Bailey Gemini 3 & Robot Dogs: Inside Google DeepMind's AI Experiments | Paige Bailey · 41:52

Transcript excerpts; ellipses mark omitted words. Some automatically selected quotes have filler words removed.

Essay featuring Paige Bailey