What AI has done for my productivity

[Dadtallica]

What happens if AI writes a few hundred lines of code that breaks the game.  Will you be able to find the problem assuming ai is happy with the outcome?  Just curious.

Could AI do things like adding the end of WW2 Pershing tank?  We so need a tiger killer.

Probably yes to all your questions, although it would be hard to break the production copy of the game unless you were using the production code which I’m sure nobody is.

[CptTrips]

What happens if AI writes a few hundred lines of code that breaks the game.  Will you be able to find the problem assuming ai is happy with the outcome?  Just curious.

Could AI do things like adding the end of WW2 Pershing tank?  We so need a tiger killer.

AI turned me into a newt!

The odds are far less than a new hire programmer doing the same and AI would make it a lot less difficult to track down the problem.

HT is not blindly letting it code unattended... yet.  It is a force multiplier. 


The biggest hole in AI right now is 3D modelling.  Not image generation but 3d modelling.  If they ever get that nailed, Katy bar the door.   

[Dadtallica]

AI turned me into a newt!

The odds are far less than a new hire programmer doing the same and AI would make it a lot less difficult to track down the problem.

HT is not blindly letting it code unattended... yet.  It is a force multiplier. 


The biggest hole in AI right now is 3D modelling.  Not image generation but 3d modelling.  If they ever get that nailed, Katy bar the door.

Pretty sure the horses are already out of the barn by now.

[AKIron]

Of course there will be backups before any major changes.

[AKIron]

It's been a few years now since I've seen it but ransomware was a real problem for many for a while. The designers were even getting into backups and encrypting those.

[CptTrips]

Getting so close.

3D modelling is a difficult problem, but it will be solved.

[AKIron]

Imagine what happens when all of the US AIs collaborate freely and the Chinese AI joins the fray. Joy and happiness everywhere! Or death and despair. It's a tossup.

[CptTrips]

Imagine what happens when all of the US AIs collaborate freely and the Chinese AI joins the fray. Joy and happiness everywhere! Or death and despair. It's a tossup.

Much less than the risk of humans interacting.

[AKIron]

Much less than the risk of humans interacting.

Touche

[CptTrips]

So the bad news is, that mathematically there is probably an upper limit to useful AI scaling.
The good news is, we are still probably an order of magnitude away from that limit.

If you look at the insane power of the leading edge frontier models today, and project that an order of magnitude forward, I'm pretty sure that is beyond what most people can conceive of anyway. 

In reality, what the limit means it that is the limit of what all the human language and concepts could encapsulate and has symbols to describe.  Including our mathematical language.

Is there anything beyond that?  Maybe, but it would be beyond our ability to grasp at a biological level.

And why shouldn't it?  A lowly John Deere dozer can certainly move more dirt than any of us.  Why shouldn't our AI tools be capable of more intellectual work any of us are capable of?  That is why we have tools.  To extend beyond what mere hairless monkeys can do themselves.

[AKIron]

Assuming a consistent scaling based on past and current progress. Something smarter with the ability to make itself smarter is uncharted territory.

[CptTrips]

I have to laugh at how so many people respond to all this.

I call them AI-Karens.

[Brooke]

HiTech, which AI and tools do you use?

[CptTrips]

This guys used to annoy me with his sneering anti-AI Karenism.

We broke him.

[AKIron]

Learn to mine coal.