@growwithsanch Currently, the available evidences suggest that agent frameworks can only improve creativity performance up to a threshold before it appears to plateau. I would surmise that both are necessary in the future.
English
Ed H. Chi
10.5K posts
Ed H. Chi
@edchi
Research VP @ GoogleDeepMind. ACM Fellow.
California Katılım Ekim 2007
3.8K Takip Edilen13K Takipçiler
Increasingly, I'm focusing my research on the creativity question: how can reasoning and chain-of-thought (schemata) might lead to "creativity" in large models.
Answer is more complicated than I expected---better memory mechanisms, planning, connecting novel knowledge.
Taipei City, Taiwan 🇹🇼 English
A few months ago, @qx_dong turned me from a pure IC into a manager. 🤗
Was a really fun experience so far watching people join one by one, building the team culture together and going up on the frontline. 🫠
Unprecedented moment in the history of Singapore to have so many talented researchers in one place. I'm glad to be building this legendary team. 🫡
Qingxiu Dong@qx_dong
Overdue update: I’ve joined @GoogleDeepMind as a research scientist to work on 💎Gemini post-training. Feeling incredibly fortunate to be cooking models with such a brilliant team, under the leadership of Yi Tay (@YiTayML) and Quoc Le (@quocleix). In the past a few month, we've found so much enjoyment in working with our own models and keeping pushing their boundaries. Ps, there's a special joy in creating recipes and internally naming them after our favorite drinks🧋 Looking forward to keep enjoying research, pushing the frontiers of Gemini and seeing the magic unfold on the path to AGI!
English
@LichangChen2 In specific domains, current frontier models long surpassed human intelligence. We don't call them AGI-capable already is because the models still make simple reasoning mistakes that are often unreasonable.
English
@edchi I think we are on the AGI: Model’s IQ is above the level of general humans.😆 Curious about your definition of AGI.
English
@JohnYue122333 Agree. Many problems are under-specified and ambiguous, and there are not enough environmental constraints known to fully specify which solution is the best.
English
@edchi Deeeeeeeepeeeeeest insigt！
We can use uncertainty reasoning to inspire/heuristically guide our problem-solving, but we cannot rely on it, as its fundamental nature is a lack of constraints regarding objects and relations in the objective world.
English
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!!
Singapore 🇸🇬 English
Some news: I have joined Google Deepmind as a research scientist. I will be leading a team that does research on making Gemini more secure and private. LLMs are now highly capable, and security has become a barrier to realizing their full potential as agents. I am excited to join the team and help make Gemini the most secure frontier model.
English
Ed H. Chi retweetledi
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!
English
@Azaliamirh Very relevant to the research I'm doing at GDM. Maybe I should go to ICLR this year!
English
Looking forward to giving a talk at the ICLR'2026 workshop on Lifelong Agents: Learning, Aligning, Evolving on:
"Self-Improvement Through Test-Time Scaling of Verifiers"
April 26, 16:00 – 16:30
lifelongagent.github.io
English
Ed H. Chi retweetledi
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: summit.newturing.ai
Be part of the conversation!
English
Related: github.com/juliusbrussee/…
"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.
English
@datawarmup being able to invent new rewards that are interesting might be the real definition of AGI. :D
English
Highly likely this will happen if motivation in place. Not sure if they are already capable of inventing new rewards.
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.
English
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.
English