I hope you are paying attention that the future belongs to efficient hardware.
Introducing Z1T:
Our first family of transformer-like models
made for sparse probabilistic hardware like Z1
Achieving up to 140x energy efficiency gains over GPUs
and revealing a new scaling law for sparse transformers
Read the blog:
extropic.ai/writing/z1t
2
92
This is an IQ test.
just went on stage with @bcherny for startup school
and he had a special announcement for all 6k+ attendees: free claude max 20x accounts for 6 months!
thank you!!
2
136
Apple is more likely to sign the open letter in support of open weight models than Anthropic.
I can’t name a single open weight model that Anthropic has ever released at all.
Because there are none.
Will Anthropic sign?
If you work at Anthropic worth asking your leadership to sign or make their position clear: Are they for banning open weight models?
78
Can’t argue against this.
Otherwise, someone name just one open weight LLM model that Anthropic ever released since their founding.
Just one. I’ll wait.
Anthropic is NERVOUS because
they know what’s coming:
- opensource AI will win
- AGI will run local, not on their servers
calling it now
> be early
> Buy a GPU
> get ur hands dirty
> learn how it works
> you’ll thank yourself later
bookmark this for later, it’s gonna be great
6
2,278
This is why AI safety is slowly looking like a scam unfortunately.
head of anthropic’s safeguards research just quit and said “the world is in peril” and that he’s moving to the UK to write poetry and “become invisible”. other safety researchers and senior staff left over the last 2 weeks as well... probably nothing.
1
219
I can’t imagine Tesla firmware being generated by LLMs directly into a binary just to replace the function of a compiler.
This doesn’t make any sense whatsoever.
Code itself will go away in favor of just making the binary directly.
The next step after that is direct, real-time pixel generation by the neural net.
1
151
It’s useless for remote hiring, but it will instead be used in-person as a final stage on-site pair programming session.
AFAIK, Anthropic does not allow AI assistance in their interviews but I can imagine them doing an in-person on-site final stage interview using Leetcode hards with the candidate pairing with the interviewer.
1
181
It took some time for everyone to realize that safely running AI agents locally on desktops is the future.
The only missing piece is performance of local inference.
There’s one account that predicted this with perfect timing and once again it is @iamgingertrash who told you so.
4
2,517
Will not be surprised to see @sama join Disney’s board of directors.
Disney Invests $1 Billion In OpenAI, Strikes Landmark Deal To Bring Beloved Characters To Sora zerohedge.com/ai/disney-inve…
2
176
This is one of the reasons why the US remains the worlds superpower.
They set the prices, add tariffs on everyone else and sets the terms of the trade deals.
If this isn’t already obvious with these ticket prices, the US is farming the world.
1
534
We will all come back to this post which is the system that @ilyasut was describing in his latest interview.
But @iamgingertrash already told you so earlier.
The secret is as follows;
An Edge network of 5B devices
is an MoE with numExperts = 5*10e9
Activated by capital incentives (micropayments)
Nothing on earth will scale better
Not even Elon’s kardeshev
Space Computer
Ethereum is the Router
Truffle, one such edge node type
2
151
In AI and Mathematics, this changes everything.
We are on the cusp of a profound change in the field of mathematics. Vibe proving is here.
Aristotle from @HarmonicMath just proved Erdos Problem #124 in @leanprover, all by itself. This problem has been open for nearly 30 years since conjectured in the paper “Complete sequences of sets of integer powers” in the journal Acta Arithmetica.
Boris Alexeev ran this problem using a beta version of Aristotle, recently updated to have stronger reasoning ability and a natural language interface.
Mathematical superintelligence is getting closer by the minute, and I’m confident it will change and dramatically accelerate progress in mathematics and all dependent fields.
1
177
Here’s another startup that should not be faded: Mira.
The video alone shows multiple killer apps for the product. The question is which one of them is the home run or will a new one emerge?
AR Glasses + LLMS (Search) + ??? + ??? + JEPA (Vision) = Ultimate Killer App.
I think this one has the right timing, and its design looks very natural to wear.
We just raised a $6.6M seed round led by General Catalyst, with participation from Pillar VC, Naval Ravikant, and others to build glasses that remember everything.
Same vision, new name: Halo is now Mira.
We’re building the first glasses to be your second brain. Instantly know the answer to any question, recall any conversation, understand any language, solve any problem, and more--all in a secure, private display.
Mira glasses help you remember everything in critical moments. Become superintelligent now at trymira dot com.
154
