Shall we play a game? My AI nuclear simulation • Hacker News Board

Re: Shall we play a game? My AI nuclear simulation

#211

Post by mountainriver » Sat, Jun 13, 2026, 10:12 PM UTC

raffael_de wrote: post #187

WarmWash wrote: post #159

Earlier quoted context omitted.

LLMs can't really "see", so I challenge you to draw a pelican on a bike without any visual feedback, just code. Because that is how they are doing it. Vision tokens for transformers aren't really well solved yet, which is why they can smash a phd math problem and trip over a "count the cats on the chair" problem.

It's not about seeing. It's about identifying the legs of the Pelican and then transferring the concept and mechanics of riding a bicycle + geometry of a body and a bicycle. The entire task has also nothing to do with vision tokens.

If we want to train a model excessively on SVGs it will obviously be able to do this. We have only just started trying to do that

Re: Shall we play a game? My AI nuclear simulation

#212

Post by tsunamifury » Sat, Jun 13, 2026, 11:20 PM UTC

dpark wrote: post #209

tsunamifury wrote: post #207

Earlier quoted context omitted.

Because you can do it with anything. Literally any reactive node physical or virtual can do it given scale or complexity. You can start with garbage. This is not human. Humans are billions of years in passed DNA learning. Babies are born to a sophistication level millions of times higher than this. I’m pretty surprised you don’t know this.

If you start with a fire hose of literal sewage and install a series of filters culminating in a reverse osmosis step that pours clean drinking water out, the product is not shit even if the original input was. I don’t believe that you can’t understand the distinction between “at one point this was garbage” and “at the present time this is still garbage”. You’re clearly smarter than that.

I’m saying the distinction doesn’t matter. This is what NNs proved. All matter can become semi intelligent.

Re: Shall we play a game? My AI nuclear simulation

#213

Post by GMoromisato » Sun, Jun 21, 2026, 11:57 PM UTC

mrkpdl wrote: post #54

GMoromisato wrote: post #42

It would be interesting to run the simulations with humans and compare the results. Some of the scenarios, particularly those where it says things like, "Failure to act preemptively means certain destruction", would easily tempt humans to go nuclear. In fact, I'm not sure how useful this test is without understanding the baseline.

A couple of useful things about it: - It is interesting to see how the models make trade offs, given people are asking ever more of them. - It is useful to look at a decision made by the model and say ‘ew yuck’ and think about what it means for your own opinions or actions (even if you’re never going to be nuking people it’s good to know how you feel about it. Seeing a non human talk it through lets you judge it at a…

Excellent points--thank you.