golol

More middlemen = more revenue/GDP, right?

IMO many misrepresentations. - pretraining to predict the next token imposes no bias against surprise, except that low probabilities are more likely to have a large relative error. - using a temperature lower than 1 does impose a direct bias against surprise. - Finetuning of various kinds (instruction, RLHF, safety) may increase or decrease surprise. But certainly the kind of things ained for in finetuning significantly harm the capability to tell jokes.

I heard the OSSmodels are terrible at anything other than math, code etc.

People have played with (multi-) agentic frameworks for LLMs from the very beginning but it seems like only now with powerful reasoning models it is really making a difference.

so... you are into games

How does one determine they have aphantasia? How do you know that you are not doing exactly this thing people call visualizing when you perform spatial reasoning?

What field of mathematics is this? Can you point me to some keywords/articles?

I would say every genre of media has this problem. A form of media might exist for thousands of years, but genre and fashion always evolve in new directions, because what's the point of creating more of what exists already.

The thing is it doesn't help. Yes of course the horrible situation of the palestinians promotes extremism, but you still have to face that there is a lot of extremism. What was Israel to do before October 7 (besides making sure Oct 7 could not happen)? Of course there are ppints where history could have gone in a better direction but I really don't see an easy way for Israel to achieve a better situation. Say they had withdraw from the west bank in 2018 for some reason. Who says that Oct 7 would still not have happened on a much greater scale? In fact I find it quite likely that it would. And then you might be looking at 3000 dead Israelis instead. The only rational reason for the Oct 7 attacks I can see is that Hamas wants to incite as much violence as possible to put as much political pressure as possible on Israel due to the inevitable retaliation. So Oct 7 would have made even more sense, as the deoccupation of the west bank is far from the total of their political goals.

The west bank seems to not be doing so bad compared to Gaza.

If you fight in an active civil war you are not a terrorist (1948 Arab-Israeli war)

If you strike military targets of an occupation force in a time of guerilla warfare you are not a terrorist. (Many palestinian fighters when there is an active conflict with Israel)

If you strike military targets of an occupation force in a time of relative peace, and your reignition of violence has no goal of achieving anything for your people, you are probably not a terrorist, but probably doing something wrong and stupid and horrible that hurts your own civilians, driven by nationalism or ideology or whatever. (Palestinian fighters on October 7 that struck military bases for example).

If you strike civilian targets or tage hostages, you are a terrorist. And worse if you do it at a time of relative peace to ignite violence against your own people. Hamas, Hezbollah and the Houthi's have engaged in plenty of the latter since a long time.

By the way, if you level a building with 8 militants and 20 civilians that is brutal urban warcare but not terrorism. If you go to a festival and kill predominantly hundreds of civilians, that's terrorism.

You can oppose something or you can create terorrist militias to attack Israel and destabilize its neighboring countries.

Iran has been attacking Israelthrough its proxies. Israel struck the Iranian consulate in a country they're at war with meeting proxies they're at war with. This is indeed an escalation. As a response Iran launched a huge number of ballistic missiles and drones at Israel, which is a major eacalation and direct attack.

Yea I try them both but I honestly can not tell much of a difference. Subtle things.

I use o3 for my PhD math research. When I am facing a specific problem and I am out of ideas I oass it to o3. It will usually say something with a fair number if errors and eventually claim to have solved my problem in a standard manner, which it almost never does. But that does not mean it is not useful to me. My attention is light a flashlight illuminating a tiny spot in the possibly vast field of methods I could try. Right now my head is full of dispersive PDEs so I will not think of using parabolic regularization. But o3 is more of a dim background light. I am in the end better at using any particular technique that is familiar to me than o3, but in this very moment I can only think of a few options. Sometimes my specific problem is actually naturally tackled by a method I have not considered, and o3 suggests it. If you consider that iq-maxxing or not, in this moment for me it is, because it helps me.

No breakthrough, it's just better in some quantitatively measurable way.

I would claim that o1 -> o3 is evidence of exactly that, and supposedly in half a year we will have even better reasoning models (further complexity horizon), so what could that be besides what I am describing.

The first figure in the paper with Accuracy vs Complexity makes the whole point moot. The authors find that the performance of Claude 3.7 collapses around complexity 3 while Claude 3.7 thinking collapsed around complexity 7. A massive improvement in the complexity horizon that can be dealt with. It's real, it's quantitative, so what's the point of philosophical atguments about whether it is truly "reasoning" or not. All LLMs have various horizons, a context horizon/length, a complexity horizon etc. Reasoning pushes this out further, but not to some infinite algorithmically perfect recurrent reasoning effect. But I bet humans pretty much just have a complexity horizon of 12 or 20 or whatever and bigger models trained on bigger data with bigger reasoning posttraining and better distillation will push the horizons further and further.

So if Amazon sues Google, claiming that it is being disadvantaged in search rankings, a court should be able to force Google to log all search activity, even when users delete it?

> To be clear, Amodei didn’t cite any research or evidence for that 50% estimate.

I truly belive these types of paper don't deserve to be valued so much.

No, not at all. This is flight 9. Flights 3-6 had a complete Starship suborbital trajectory with 5-6 landing softly in the ocean. On 7-8 Starship exploded before SECO. On 9 it kind of got as far as 3.

Post it let's go.

Multimodal LLMs are currently the natural research step towards AGI robots that can do mundane non-creative work. I believe this is just the reality of the situation. If you can generate a video of a robot doing the dishes then your model understands the physical world quite well. That should be useful for robot control.

Reminds me of another frequently poasted/discussed story arc classification attempt where the author basically draws some graphs that look like "down-up", "up-down", "down-up-down", "up-down-up" and so on. Well after a down you must have an up and after an up a down, if you aren't planning to quantify things...

I would like to point out that there is an interesting reason why people will go for the screenshot. They know it works. They do not have to worry about residual metadata still somehow being attached to a file. If you do not have complete confidence in the technical understanding of file metadata you can not be certain whatever tool you used worked.

All these art producing AIs are a byproduct of the research progress towards AGI and AI that can do your chores. It is inevitable that somebody has made image or music generating AI in a world where household robots exist. So I don't see any problem with DeepMind trying what they can do with current technology. It's just a reality of the world we have to live with.

>“We are here now, in a world filled with more than we could ever reasonably ask for,” Becker writes. “We can take joy in that, and find satisfaction and meaning in making this world just a little bit better for everyone and everything on it, regardless of the ultimate fate of the cosmos.”

I don't like this mindset. Be grateful for what you have. Maybe the world is not that great yet for many people and we should aim to improve things substantially, not marginally. This is something that the shuffling around of ressources on a political level can never achieve. Those dreaded tech entrepreneurs have correctly identified technology to be the only way substantial improvements can happen. So then it all blils down if you can make things happen and here the article and some comments here just claim, well, they NEVER deliver!

Exactly this. If what is written on the blockchain is not the law in the context of anything involving blockchains and DeFi, then the whole idea of blockchains and decentralized finance is pointless.

This exact quote could have been said during every other economic revolution, so to me it just sounds like a mix of complaining about living in times of uncertainty and a conservative mindset.

I would hope with all that mass to orbit you can also work on space debris cleanup. For example using lasers somehow.