aerodexis

super neat! Would be curious to read more.

It doesn't. This is extending an earlier PoC where someone created a tensor representation of a CPU and you prompted it with the code you wanted to run.

This will be interesting when something similar to a CPU can be triggered off of the LLM's own internal state representation - but imo we're far away from that.

LLM != CPU hardcoded into the same architecture used for an LLM. The non-determinism did not go away - all this is doing is intercepting tool-calls produced by qwen and executing that in different layers.

It's truly not that interesting.

depends on what being good at grammar entails. If it's just following rules, then you're not wrong. If it involves adapting concepts to novel circumstances, we're doing well.

Your idealism manifested as practical flourishing when you started a family.

Imo, the right is experiencing an intellectual renaissance rn - but this is much more aligned w/ recovering normative morality, re-enchantment and multi-polarity than it is with the libertarian, anti-democracy strain of thought that you describe. I was stunned to find out that a friend was using the word "monarchy" in it's premodern sense, denoting distributism and strong practical limits to the enactment of power, which is totally opposite to how libertarians used the word. There's a lot going on that is interesting, useful and hopeful.

Thiel is very interesting, but not for good reasons. It's amazing to see someone who is simultaneously very proficient at what they do, yet also incredibly shallow, decorating himself w/ whatever bits of pieces of cultural material that allows him and others to confuse "good investment" with "good for people". Is it just me, or does this guy flake out entirely under pointed questioning?

The differences b/w knowing, understanding and interpreting are suddenly very real and highly relevant.

I, for one, am profoundly thankful that the ~300 year mania around innovation and newness is beginning to wane. You don't need to read that many history books to appreciate that the values that we have today are profoundly different than what they were previously. It takes a few more books, and a pinch of humility, to appreciate that today's values are neither universal nor the necessary outcome of a teleological evolutionary process. The reworking that Tao suggests here reaches far beyond math. The fact that mathematicians are starting to wrestle with these questions is cause for great optimism. Once VCs do the same thing, we'll be on the other side of the storm.

"you will reach El Dorado" is profoundly different than "I will reach El Dorado"

Don't bother with this.

I'm reading Information and Meaning in Evolutionary Processes by Harms and it goes off on how weak Dawkins is.

I'm hyped for you. This one is a massive banger.

I wouldn't be surprised if it's marketing all the way down.

Personally - for banking - I'd be plenty happy dealing w/ 90% uptime if it gave me an order of magnitude more security against getting robbed. This tradeoff may not have been realistic a decade ago, but it seems to be where things are trending.

It also is de-facto the case today when logging in as per usual sometimes requires a captcha or additional verification.

You don't think frontier mathematics is already secret?

I could imagine it going so far as to have an LLM re-impl core libs every time a request is processed. Lightning never strikes the same place twice.