Head of AI at https://nitter.cf/t.co/CuMA7vSuqV AI for conflict resolution @HumanAssisted

SF
Joined March 2008
One year from now AI companies will evolve OpenAI is Microsoft Claude is Apple Meta is Google SpaceXAI is AWS/ASML Apple is the best fashion brand Nvidia is TSMC Google is the US gov
36
ASI will need clothing.
15
Jonathan Hendler 🪰 retweeted
In CSI it was "enhance!" In ASI it will be "simplify!"
1
27
In CSI it was "enhance!" In ASI it will be "simplify!"
1
27
It has been a crazy year this week.
23
Hope @cerebras brings this back.
I am using @cerebras Gemma 4 31B. *So* fast and pretty stable model for simple things. Great combination if you have a lot of routine work you want to get done quickly.
92
I first wrote logical at an emergency hackathon on March 5, 2023. Logical to can compress documents into typed claims with checkable proofs using Prolog. Now, Astra translates the text; a logic engine preserves citations, tracks identity, and flags stale evidence. github.com/Hendler/logica
35
Replying to @X
@x there's a disincentive to follow people if your own follower count less than followers. I wonder if these game mechanics of growing accounts can change.
1
30
@grok how would you change the core algorithm to help accounts get followers and more high quality engagement? Perhaps create small communities with smaller networks of voices?
1
12
If I was 8 years old, watching grownups lose it everyday, I might start to not trust adults. What an amazing time for grownups.
this is insane - steve says he can make the sketch 100x better - tibo says do it by tomorrow - steve runs a walkthough of a chatgpt app they built with the product in it and 100x better (the products they built i in their replies - its this tldraw-chat-app-example.tldr…)
53
Jonathan Hendler 🪰 retweeted
What is an alignment startup? Harnesses, judges, measurements, safeguards. In short, trusted determinism.
1
1
23
Selling data to foundation model companies is lucrative. We need more alignment startups.
This may be the correct approach. However, what do we pace? Data centers? Chip speed? Algorithm improvement? What measures the measurement? It is an uncomfortable moment and an opportunity to steer. I don't trust that governments can do it.
1
17
What is an alignment startup? Harnesses, judges, measurements, safeguards. In short, trusted determinism.
1
1
23
This may be the correct approach. However, what do we pace? Data centers? Chip speed? Algorithm improvement? What measures the measurement? It is an uncomfortable moment and an opportunity to steer. I don't trust that governments can do it.
Replying to @janleike
We still have time to change to rules of the scaling race, but maybe not a lot. AI will keep improving rapidly and the stakes will increase. Unless we have pacing mechanisms that apply to everyone, companies are incentivized to accelerate, or else risk failing commercially.
1
55
The dialog @BernieSanders creates is useful because it is honest. There is fear about data centers, AI safety, wealth disparity, etc. But complex challenges need thoughtful, scientific, educated, and accurate definitions. Otherwise they lead to regulatory solutions being proposed which are not useful, and potentially dangerous.
It is kind of amazing that the 85-year-old Bernie Sanders is one of the few prominent American politicians (yes, I know there are some others, but none with his profile) with anything useful to say about AI regulation.
65
Fable or Astra *could* solve this, but the content filter.... only Mythos or Internal models allowed....
one time I met a guy who was doing a math PhD and I asked what the topic was and he said "Tits Rigidity". and that is how I learned that there was a guy named Jacques Tits who has a lot of things named after him
1
52
Instead of planning, playing.
19