Damien Henry

4.5K posts

Damien Henry banner
Damien Henry

Damien Henry

@dh7net

SVP Image research @heyjasperai Cofounder @ClipdropApp, YC W21, acquired by @stabilityai AI x Images @googlearts Created GoogleCardboard @Google

Paris Katılım Eylül 2009
970 Takip Edilen6.5K Takipçiler
Corey S. Powell
Corey S. Powell@coreyspowell·
I also have to point out that the most expensive space telescope (JWST) cost about $500 million/year. We spent 1000x that much on AI development in 2025. Data collection is essential for discovery...and it's remarkably cheap compared to other things we do routinely.
Corey S. Powell tweet media
English
2
4
19
687
Corey S. Powell
Corey S. Powell@coreyspowell·
I keep seeing versions of this post, which imply a bizarre misunderstanding of how we know the world. Do people imagine that if we'd never seen galaxies or exoplanets or the cosmic microwave background, we could have just *imagined* these things & that would be just as real?
Corey S. Powell tweet media
English
10
9
44
1.9K
Damien Henry
Damien Henry@dh7net·
@mikiane @ClementDelangue IMO, ça n'a pas de sens d'opposer fine-tuning et orchestration. C'est évident qu'il faut des tools, skills et data (RAG ou pas). Mais utiliser le best frontier model pour faire tout et n'importe quoi, c'est vraiment pas optimal (coût, qualité et latence). => Fine-tuning
Français
0
0
0
27
Michel Levy Provençal
Franchement tous les projets que j'ai vu qui ont tenté de faire du fine tuning (meme dans des grands groupes) ces dernières années sont rarement sortis du POC. En revanche, les apps qui orchtestrent des modèles fondation autour d'outils (tools), de process (skills) et de datas (RAG) s'en sont plutot pas mal sorti. Preneur de cas d'usage qui contredisent mon REX.
Français
3
0
3
654
Michel Levy Provençal
Mistral lance Forge pour fine-tuner. 95% des boîtes n'en ont pas besoin. Un frontier + RAG + tools bat un fine-tuning custom. À chaque fois. Fine-tuning = 2023 Orchestration = 2026 Non ? mistral.ai/news/forge
Français
9
1
28
12.3K
memo akten
memo akten@memoakten·
A description of a thing is not the same thing as the thing.
English
9
1
23
1.8K
Damien Henry retweetledi
François Fleuret
François Fleuret@francoisfleuret·
There should be a "human safe space" universal marker to indicate chunks of text / source code that should never be removed / modified / bypassed by an AI.
English
2
2
29
4.2K
Damien Henry
Damien Henry@dh7net·
@michael_nielsen TBH, I'm definitely an optimist, and for me, it has nothing to do with "hope." It's more about the realization that the present is much better than what most people think it is.
English
0
0
0
18
Michael Nielsen
Michael Nielsen@michael_nielsen·
A thoughtful set of clarifications: x.com/steve47285/sta…
Steven Byrnes@steve47285

I don’t like that “optimistic” conflates an epistemic / forecasting thing with a decision-making thing. A superforecaster will be simultaneously “optimistic” about things that will probably work out well and “pessimistic” about things that will probably not. And good for them, that’s exactly how they should be. Of course, that’s assuming they’re in the role of a passive observer. Separately, active participants in events should energetically try to make things go well. I think good leaders have both wolves inside them: given a plan, they are brutally honest and clear-eyed about how likely it is for the plan to work vs not work. But they also don’t give up searching for better plans prematurely. Hence CEOs are praised for “pivoting” away from unpromising plans etc. …Of course, I have an axe to grind here. I think we should be “pessimistic” about the AI risk situation, in the epistemic / forecasting sense, i.e. in the sense that I see a lot of problems that seem fatal and (based on historical analogies and other arguments) seem very likely to bite us. Thus, I usually describe myself as a “pessimist” about future AI, and I don’t like when people talk as if that’s my character flaw rather than my considered opinion. (Not you in particular, I see it a lot.) Of course, I am energetically trying to make things better. By analogy, if I’m captain of a high school football team, playing an exhibition game against top professionals, then I can and should energetically try to win, and brainstorm out-of-the-box strategies, etc. But after doing that, if I “optimistically” expect to win, then that would be moronic not praiseworthy.

English
2
0
3
2.1K
Michael Nielsen
Michael Nielsen@michael_nielsen·
Who is the most optimistic person you know? (DMs welcome)
English
28
0
64
14.6K
Damien Henry
Damien Henry@dh7net·
My agents will request recommendations from my human friends' s agents before consulting Google for sure. This is how this peer-to-peer search system will emerge @Plinz.
Damien Henry tweet media
English
0
0
0
169
Damien Henry
Damien Henry@dh7net·
translate The tragedy of too much clout is that you become surrounded by constant blandishment, and if you lack judgment, you eventually start to believe it. in french
English
0
0
0
85
Damien Henry
Damien Henry@dh7net·
. @huggingface storage is amazing! #trusted-by-launch-partners" target="_blank" rel="nofollow noopener">huggingface.co/blog/storage-b…
English
0
1
7
1.2K
Damien Henry
Damien Henry@dh7net·
Of course, you want to cultivate your driver skills. The jobs where AI guides you are the ones that will be replaced first. But meanwhile, be driven by an AI can get things done fast. Use this if you don't allready!
English
0
0
0
68
Damien Henry
Damien Henry@dh7net·
For some stuff, you want to be the driver, and AI is your workforce. But for some other tasks, when you are not the person that could make the difference, you want to be driven by AI. Just dump all the context, and ask, "Now ask me relevant questions one by one." It's magical.
English
1
0
1
154
Damien Henry
Damien Henry@dh7net·
@opencode is underrated. 1) Install opencode on your Linux server. 2) Use it through SSH. 3) You can almost do everything that openclaws does.
Damien Henry tweet media
English
0
0
3
198