Gonna post a banger soon, stay tuned!

Grok is so over
https://fixupx.com/vvsotnikov2/status/1946899789670466046

ChatGPT agent reasoning about clicking “I’m not a robot”

Intel’s new US fab seems to be cancelled as well

https://www.nbc4i.com/intel-in-ohio/intel-ohio-plant-likely-canceled-if-company-cant-get-new-manufacturing-customers/

I decided to post this shortly after GPT-5 comes out (which comes out in a few hours/days) to take it into consider when writing. Stay tuned

It was obvious something like this is coming but I’m still extremely impressed

Makes you think what “intelligence” and “reality” really are if matrix multiplications can achieve this:

https://deepmind.google/discover/blog/genie-3-a-new-frontier-for-world-models/

Keep thinking about this

GPT5 stream

They proudly show how ChatGPT:
1) Doesn't answer fireworks-related question
2) Decides how to do a cancer therapy

AI will decide for you how to live but fireworks is too much chud

A random thought: now when more and more tools are AI-powered, security by obscurity becomes [more] important

It’s harder for AI to hack you if its pretraining data had nothing similar to your system

Do you agree?

Thinking of starting a list of (mostly esoteric) public goods for AI dev/research that I personally would like to work on if/when I have more time:

- a toolkit to benchmark models of different quantizations: usually authors only check for perplexity which IMO is not enough
- open-domain datasets of various extremely niche knowledge, ideally requiring an internal world model, for example: Stalker games knowledge; lifehacks and howtos of my local neighborhood; etc. I would've also kept the validation part private just like the ARC dataset
- "questions requiring reasoning about multiple needles in multiple haystacks" datasets - this one is obvious. Would be interesting to see how all those sparse attention approaches handle them hehe

why do I think those are important?

1) I'm simply curious tbh
2) to turn the Goodhart's law into a tool: if everyone is benchmaxxing, the best way to make sure new models are aligned with your needs is to make benchmarks that measure things you want/need, especially if we're entering the era of "universal verifiers"

Y’all probably heard it already but after gamersnexus made a video report on Nvidia turning a blind eye on China GPU smuggling, some unhappy people tried to bring their channel down:

https://youtu.be/tUnRWh4xOCY

Here’s the reupload of the deleted video hehe

Working towards real-time reinforcement learning

1) New Prompt Enhancer is a new kind of an extremely smart AI that learns from AIKEK holders in real time

2) Under the hood it works as a scaffold for the online RL which is in turn a revolutionary thing that will bring us to ASI that is aligned with AIKEK holders

3) It leverages new tokenomics: bigger holders get more value

4) More RL widgets are in development: for data, for memes, and for completely novel use cases

—
https://x.com/alphakek_ai/status/1970511071074017292

AI Wars update: Dutch government has taken control of a Chinese-owned (and Netherlands-based) chip company "to prevent critical goods (chips) become unavailable

https://fixupx.com/onechancefreedm/status/1977550998253060508