paufernandez

Oh! Nice surprise, switched instantly! Love it!

I see it like growing a plant. You can prune here and there, change the soil, but the plant does what it wants.

Yeah, Mojo.

It could be that those bugs are Claude or Codex works of art.

I think the real bottleneck is one layer down: ASML.

My thesis advisor and team doing interesting work.

Nerd sniping at its best... ;)

I'm still rooting for Modular, though.

I had thought of this, so nice to see it done!

We assume sometimes that everybody experiences this in the same way, but a lot of people might be super-sensitive to it, and others completely immune. It is quite possible that the ones obsessing about it are the sensitive ones, because they feel it much more.

A friend of mine actually did it, and he is quite happy.

I also believe this. And I think very few people can pull it off, and Chris Latt ner is precisely one of them. I hope that the agreement includes open sourcing Mojo at some point. Chris mentioned fall 2026, iirc.

The most beautiful form of power.

It is increasingly important to be able to see that many things are true. There is no single "truth". Many things are true at the same time, and in all aspects of life. Each brain is like a band pass filter, and the effort we should make is to try to imagine the points of view of others, which are just different slices of the same world. Then embrace the slices we like, and just ignore the ones we don't, but don't argue or fight for our slice as it if was the only one.

"They're made out of neurons"

"Neurons?"

"Neurons. Cells that fire impulses. We checked the whole thing through. It's nothing but neurons."

"Neurons doing what? Where do the words come from?"

"The neurons make the words. Are you understanding me? We opened it up. There's no dictionary in there, no grammar rules, no little man. Just neurons. A whole cortex of neurons sending each other impulses."

...

People don't understand emergence.

I've been taking a look at the source and it's a work of art :O

Upvoted just for the name.

This is solved by Mojo already, they must be rushing something to compete, since Mojo is in version 1.0beta1

That some people play survival all the time because "that's what life is" does not mean that humanity doesn't have pockets of more quiet, non-competitive environments where blue pressers thrive. Humans have had a lot of periods where they were not being "just animals".

But I guess everyone thinks the world is like he wants it to be in this respect.

The thing is, all those red button pressers can't see that, for all their exact and flawless reasoning, they have a bias towards individuality, which they see as the only possibility, and we blue button pressers perfectly understand the math and yet we have a bias towards others, and would sacrifice more readily at many situations like this one. This sacrifice is what reds see as "dumb", but natural selection has chosen this because it probably works.

Both sides have a mental bias, and just can't see each other's "reasoning" because of it.

We blues think of reds as selfish, because we can't conceive of anyone not thinking of the worst outcome for others, and being empathetic about it, making it one's own. And they see us as "virtue signalling", or getting some external value of some kind (recognition from peers) because they can't think of any other explanation to justify that behavior, when it is just pure bias towards sacrifice. Sacrifice is just that, giving something without asking nothing, which does not make sense for a red. Reds think we are dumb but society needs a little more blues than reds. Otherwise it probably collapses.

I'm such a proud blue. in fact... ;)

To me, the whole point of the riddle is that it reveals the most internal bias towards either yourself or others, meaning that you do things for society or for yourself. Blues don't understand reds, reds don't understand blues. The bias is invisible to the self but it is clearly there given the huge contrast in the opinions of people.

I could put my signature on your comment as if it was mine, wouldn't change even a comma.

Apart from the exhaustion of context switching, I believe there is a internal signal that gauges how "fast" things are happening in your life. Stress responses are triggered whenever things are going too fast (as if you were driving in a narrow road at too much speed) and it feels like there is danger since you intuit that a small mistake is gonna have big consequences.

Some people thrive in more stressful situations, because they don't get as aroused in calmness, but everybody has a threshold velocity at which discomfort starts, higher or lower. AI puts us closer to that threshold, for sure.

The funny thing is that upon registration of a .cat domain you are required to acknowledge that your website is not related to cats at all. So those domains are, in theory, not in compliance.

Agree. But sometimes there is no "root cause", the brain is still a mystery. If you had been depressed even when you knew there was nothing to worry about, you would see it differently, because then you deduce that the black cloud is produced within.

Chemistry trumps psychology. Good enough chemistry enables cognitive treatments. But to fix the wrong chemistry you need chemistry.

Yeah, but the xz algorithm is also not counted in the bytes... Here the "program" is the LLM, much like your brain remembers things by coding them compressed and then reconstructs them. It is a different type of compression: compression by "understanding", which requires the whole corpus of possible inputs in some representation. The comparison is not fair to classical algorithms yet that's how you can compress a lot more (given a particular language): by having a model of it.

Same here, I flew head first down a ramp because I slammed the breakes too hard and the next thing I remember is people asking if anybody had a handkerchief or something, since my head was bleeding. A solid five minute blackout.

I wonder what kind of gun uses books as munition.

In my case I fully grasp what such a future could be, but I don't think we are on the path to that, I believe people are too optimistic, i.e. they just believe instead of being truly skeptical.

From where I look at it, LLMs are flawed in many ways, and people who see progress as inevitable do not have a mental model of the foundation of those systems to be able to extrapolate. Also, people do not know any other forms of AI or have though hard about this stuff on their own.

The most problematic things are:

1) LLMs are probabilistic and a continuous function, forced by gradient descent. (Just having a "temperature" seems so crazy to me.) We need to merge symbolic and discrete forms of AI. Hallucinations are the elephant in the room. They should not be put under the rug. They should just not be there in the first place! If we try to cover them with a layer of varnish, the cost will be very large in the long run (it already is: step-by-step reasoning, mixture of experts, RAG, etc. are all varnish, in my opinion)

2) Even if generalization seems ok, I think it is still really far from where it should be, since humans need exponentially less data and generalize to concepts way more abstract than AI systems. This is related to HASA and ISA relations. Current AI systems do not have any of that. Hierarchy is supposed to be the depth of the network, but it is a guess at best.

3) We are just putting layer upon layer of complexity instead of simplifying. It is the victory of the complexifiers and it is motivated by the rush to win the race. However, I am not so sure that, even if the goal seems so close now, we are going to reach it. What are we gonna do? Keep adding another order of magnitude of compute on top of the last one to move forward? That's the bubble that I see. I think that that is not solving AI at all. And I'm almost sure that a much better way of doing AI is possible, but we have fallen into a bad attractor just because Ilya was very determined.

We need new models, way simpler, symbolic and continuous at the same time (i.e. symbolic that simulate continuous), non-gradient descent learning (just store stuff like a database), HAS-A hierarchies to attend to different levels of structure, IS-A taxonomies as a way to generalize deeply, etc, etc, etc.

Even if we make progress by brute forcing it with resources, there is so much work to simplify and find new ideas that I still don't understand why people are so optimistic.

Yes, and each person has a different perception of what is "good enough". Perfectionists don't like AI code.