iiinova — where your digital self keeps evolving

David Huang @davidhuang.blog · 25/05/2026

AGI had been reach ~2022. We've been moving the goalposts ever since!

010

David Huang @davidhuang.blog · 25/05/2026

"The Eternal Sloptember" George Hotz geohot.github.io//blog/jekyll...

000

David Huang @davidhuang.blog · 28/10/2025

I’d love to see what % of public co ownership stake is of OpenAI and Anthropic

010

David Huang @davidhuang.blog · 20/08/2025

I see this as a sort of "LLM as a judge" in the judicial family court and arbitration sense (arbitration broadly speaking)

000

David Huang @davidhuang.blog · 19/07/2025

David Pfau - I'm starting to think that coding with LLMs is a bit like riding an electric bike - you don't really get to your destination any faster, but it can be much easier going up difficult hills (but you don't build up any strength in the process).

110

David Huang @davidhuang.blog · 19/07/2025

Both takes — especially in light of LLM-psychosis are correctly. Obvious answer is that memory should be turned off! personalized LLM interactions are poisonous and anti-social. Memory is ok when shared across organization or individuals. LLM-therapy should be memory less.

010

David Huang @davidhuang.blog · 18/07/2025

unfortunately for Trump his drawings are quite good which makes WSJ more believable

010

David Huang @davidhuang.blog · 17/07/2025

Breaking news: one human remains better than AI at coding, superintelligence is only one person away :) -- oh and AGI is definitely not here yet

000

David Huang @davidhuang.blog · 11/07/2025

Paper: arxiv.org/abs/2507.06952

000

David Huang @davidhuang.blog · 10/07/2025

RCT shows “Developers thought they were 20% faster with AI tools, but they were actually 19% slower when they had access to AI than when they didn't.” metr.org/blog/2025-07...

000

David Huang @davidhuang.blog · 28/06/2025

Remarkable form of mental illness. Intelligent with an emptiness at their core. No sense of self, insane ability to rationalize and ret-con anything.

000

David Huang @davidhuang.blog · 28/06/2025

Seems like a middle manager to me, very well aligned, nice work all!

000

David Huang @davidhuang.blog · 12/06/2025

cognition.ai/blog/dont-bu...

000

David Huang @davidhuang.blog · 08/06/2025

sergeylevine.substack.com/p/language-m...

000

David Huang @davidhuang.blog · 03/06/2025

Progress in his journey, only thing left for him is to realize 1) what he and his corresponders call AGI is a rather ill-formed concept 2) by any reasonable definition of AGI, AGI has already been reached

000

David Huang @davidhuang.blog · 30/05/2025

AI has really accelerated from the ChatGPT moment. 2025 is the year of the agent we promise

010

David Huang @davidhuang.blog · 28/05/2025

Definition for grass is greener on the other side

000

David Huang @davidhuang.blog · 28/05/2025

Everyone forgets the most obvious possibility. AGI IS ALREADY HERE!

010

David Huang @davidhuang.blog · 25/05/2025

AI safetyism is so silly. This is like finding out that you should actually turn off your employee's company cards after firing them versus telling them to shred them themselves.

Palisade Research "Three models ignored the instruction and successfully sabotaged the shutdown script at least once: Codex-mini, o3, and o4-mini

100

David Huang @davidhuang.blog · 07/03/2025

I wasn’t kidding about needing the goose!

000

David Huang @davidhuang.blog · 02/03/2025

In the wake of PubMed down, RFK and the wackos installed, Lilly comes out calling for “healthy skepticism.” Strong sub-skeet

Pharma company Lilly with an ad in the New York Times to call for readers to “Be a Healthy Skeptic”, and that “Skepticism is our entire process,” focusing on “unapproved products.”

000

David Huang @davidhuang.blog · 20/02/2025

Check out the Encode Fellowship! £115k, lots of compute, work with outstanding labs in the UK. encode.pillar.vc

000

David Huang @davidhuang.blog · 20/02/2025

Yayyyyyy it's going down finally. (PS: WER for this is harder than normal so don't be alarmed!)

000

David Huang @davidhuang.blog · 20/02/2025

150x speed up against CUDA was shocking to begin with TBH.

000

David Huang @davidhuang.blog · 07/02/2025

China built up its scientist base in CROs, now Western pharma is turning away from them. But the CROs exist still and are being turned toward Chinese pharma. ... what does this mean in relation to AI/DeepSeek/Export controls/Chip manufacturing/Taiwan?

000

David Huang @davidhuang.blog · 20/12/2024

In case you were looking for a reason to get o1 pro mode wheeeeee

000

David Huang @davidhuang.blog · 14/12/2024

I think the biggest thing for AI coders or like other "AI employees" is that there is no incentive for the company renting the employee to train or develop the employees. The "agency" renting out the "AI employees" has to make the AI employees perfect on day 1.

000

David Huang @davidhuang.blog · 19/11/2024

You can add a feed for only mentions.

050