That is a very specific and extreme example. Yes, you can not use magic there. At the end of the day, even the exhaust gases themselves emit IR. But you can make sure there is no direct line of sight onto the hot parts

If you train an AI to do it, I am sure it would completely destroy nature. It would find “bugs” like getting on an aircraft to reach new areas, teamwork to destroy any enemy etc.

How can you tell they lack all of these things when people testing it shows these have it? So it appears as if, but you say that is not actually the case. How do you tell?

It is not alive, why do you suspect I think that way?

And again, if those LLM are not intelligent we set an extremely high bar that even some humans will not reach. And it only gets higher.

2 minutes?

You and others here act as if this is some clear cut thing, while actually defining it is extremely hard and everything but clear cut.

I find it scary how many people here are blindly against LMMs directly (not the big companies behind them or whatever) with about as much irrational hate as the people they despise, like Trump. They act the same way. Saying LMMs are not intelligent is so absurd I am not sure how to go forward, such a statement is oozing with bias.

Define comprehension then.

How do you define intelligence? If LLMs are not intelligent NO other animal could ever be considered intelligent and many humans in the lower percentiles too. And that percentile is only going up.

You don’t need to absorb it, just emit less. Lower the emissivity and the at the same temperature it will appear colder to a thermal camera etc.

Thinking about it, if the clock runs backwards at the same rate, so -1 hour per hour, the rate of randomly being right is 2x the stand-still rate. This is also a linear function, at -1 its 4x per day, -0.5 is 3/d, 0 is 2, 0.5 is 1. When it runs true it is however very different, as it is either always exactly right (no offset) or never (with offset).

That is how heat works in general, but that has nothing directly to do with reflecting or absorbing IR.

How to recreate it? There are SO many variables. I am not going to set up a new system just too see if I can replicate some bug… unless it is a really big issue with something very important, at which point someone already did that.

Indeed, last time I paid for windows was like windows 7. Only free upgrades since then.

But fixing a bug is not as simple as “fork it or help fix it”, those are by far the most complicated options that only a tiny faction of users (can) take.

So if we focus on thermal IR: shiny metal surfaces hardly emit or absorb it. A black surface emits and absorbs as much as possible. We quantify this via the emissivity (and also reflectivity, shiny metal is a mirror). Emissivity of 1 = theoretical maximum of a prefect absorber/emitter (both are directly linked), everyday stuff like skin or wood is about 0.95, shiny metal surfaces roughly 0.1, a rusty or painted surface again about 0.95.

You only emit and absorb the same in thermal equilibrium. If the surface is hotter, it emits more. If it is colder, it absorbs more.

Sadly millions absolutely do.

So a hobby as someone else pointed out?

You are mixing different types of IR and on top are wrong for both cases too. We have materials for both NIR and thermal IR that absorb or reflect them.

Not everything is abusive.

I wish it did. Getting seconds is not even easy. Like pressing the clock on the lock screen sometimes works (for the few seconds it shows the screen) but absolutely not consistently…

When the stakes to this high and it happens so often: you don’t.