Ed H. Chi

Research VP @ GoogleDeepMind. ACM Fellow.

[Ed H. Chi @edchi]

Hanging out with @lmthang and @YiTayML for multiple days in Vietnam and Google Singapore.
And great event with @newturing on AI and humanity.

Thank you for hosting!!

[Ed H. Chi @edchi]

@kamalikac Welcome!

[Demis Hassabis @demishassabis]

Gemini 3.5 Flash is amazing!

- Performs better than 3.1 Pro on coding & agentic tasks
- 4x faster than other frontier models
- 12x faster in @antigravity - 800 tokens/sec!
- Often at less than half the cost

And Pro to come…

Try it in @antigravity, @GeminiApp & more - enjoy!

[Ed H. Chi @edchi]

@Azaliamirh Very relevant to the research I'm doing at GDM. Maybe I should go to ICLR this year!

[New Turing Institute @newturing]

Dr. @edchi will join GStar Summit 2026 for a keynote on foundation models and a panel discussion on frontier AI.

Website for program details and ticket registration: https://t.co/9JtMDBX24q

Be part of the conversation!

[Ed H. Chi @edchi]

and from 2017: https://t.co/KIptgAdM1X

[Ed H. Chi @edchi]

Related: https://t.co/enDIFfsok1

"Claude Code skill & Codex plugin that makes agent talk like caveman — cutting ~75% output tokens while keeping full technical accuracy. Now with 文言文 mode, terse commits, one-line code reviews, and compression tool that cuts ~46% tokens."

[Ed H. Chi @edchi]

A Hypothesis: Agents of the future could communicate to each other using thought tokens.

Corollary: Humans will have to decode those thought tokens in order to find out what the agents are doing with each other.

Note: GenZers are already doing this with their slang. No Cap.

[Ed H. Chi @edchi]

@datawarmup being able to invent new rewards that are interesting might be the real definition of AGI. :D

[Ed H. Chi @edchi]

A Hypothesis: Agents of the future could communicate to each other using thought tokens.

Corollary: Humans will have to decode those thought tokens in order to find out what the agents are doing with each other.

Note: GenZers are already doing this with their slang. No Cap.

[Ed H. Chi @edchi]

While Agentic AI is brewing (and exploding), are you preparing for the next wave? What's the next wave of new ideas?

[Ed H. Chi @edchi]

@spacegrep @denny_zhou @quocleix Imho, the attention paid over the length of CoT helps sharpen the model toward the correct decoding path. This CoT reasoning path helps the model perform 'next idea prediction' instead of just doing 'next token prediction'. That has been my intuition.

[Ed H. Chi @edchi]

Thanks for this recognition, Mike.
/cc @denny_zhou @quocleix

[Mike Knoop @mikeknoop]

The original CoT paper ("lets think step by step") from Jan 2022 is equally important as the Transformer ("attention is all you need"). https://t.co/YRKESO6OQE

[Ed H. Chi @edchi]

@hyhieu226 @OpenAI @xai Good luck Hieu. Still remember the days when we worked together. Hope you recover well!

[Thang Luong @lmthang]

Thrilled to share: #Aletheia, our math research agent, just solved 6/10 notoriously hard FirstProof problems autonomously, the best result in the inaugural challenge! To me, this is even bigger than our historic IMO-gold achievement last year; these problems challenge even top mathematicians. We share our results transparently, see paper and full thoughts in the thread. 👇

[Ed H. Chi @edchi]

@spacegrep @denny_zhou @quocleix That's why I work on Project Astra. Situated intelligence in the real world is absolutely part of the research agenda.

[Ed H. Chi @edchi]

@spacegrep @denny_zhou @quocleix yes, major bugs IMHO:
- the current models are generally fixed minds, and only learns and compresses new knowledge during gradient descent.
- the other learning / memory mechanism is in-context learning with CoT, but the model forgets it right after.
Clearly insufficient.

[Ed H. Chi @edchi]

In the social media era, kids actually feel more loneliness---ironically.
As a former social computing researcher, this is deeply depressing to me.
https://t.co/UBMYybK6X8

h/t Kristina Lerman #WSDM 2026 keynote

[Ed H. Chi @edchi]

@denny_zhou @quocleix in my not-so-humble opinion:
- 1995-2015 most important 3 ideas are: reverse indexing with mapreduce, vector space models, deep learning.
- 2015-2025 most important 3 ideas: seq2seq learning/transduction with transformers, CoT fine-tuning, and refinement using RL.

[Ed H. Chi @edchi]

@zzlccc @GoogleDeepMind @YiTayML @quocleix Welcome!

[Ed H. Chi @edchi]

@JustinAngel @denny_zhou @_jasonwei @xuezhiw @MaartenBosma @brian_ichter @xf1280 @quocleix Sadly, Kind of far for us in the south bay. That's like 1hr each way or more these days.