AI 2023. Meet ChatGPT. - page 201

 
Volodymyr Zubov #:

That's what I mean. There on the board to draw new developments, based on the old) And the production of elements including transistors, it is a whole science of semiconductors, then combine them into logic and registers, it is not so simple.

Where there are formulas and an algorithm, it is very possible that he will do it.

 
Renat Akhtyamov #:

where there are formulas and an algorithm, there's a very good chance it will.

In addition to formulas and an algorithm, you need know-how.

 
JRandomTrader #:

In addition to formulas and algorithm, know-how is required.

It's already been done before us.

If I can do it, I can do it without any know-how.

However, recently I saw a super-chip on the internet, in the hands of the photo.

The dimensions are about 40x40 cm.

Let's wait, I wonder where it will be used.

but preliminary - just for AI.

 
Renat Akhtyamov #:

It's all been invented before us.

if I can do it, I can do it without the know-how.

However, I recently saw a super-chip on the internet, in the hands of a photo.

The dimensions are about 40x40 cm

We'll wait, I wonder where they'll use it.

but tentatively, it's just right for AI.

Try to reproduce at least a steam locomotive without the know-how.

 
JRandomTrader #:

Try to reproduce at least a steam locomotive without the know-how.

It's there.

when you can resort to the saying "why reinvent the wheel?", it's no longer know-how.

 
Renat Akhtyamov #:

he is

when you can resort to the saying "why reinvent the wheel?", it is no longer know-how.

It is there, but can any AI practically reproduce it (at least create technical documentation sufficient to reproduce the locomotive by stupid executors)? With knowledge of the laws of physics, formulas and algorithms.

 
JRandomTrader #:

It exists, but will any AI be able to practically reproduce it (at least create technical documentation sufficient to reproduce the locomotive by stupid executors)? With knowledge of the laws of physics, formulas and algorithms.

you could try asking him.

like

- drawings

- technology.

I think he'll answer

 
Renat Akhtyamov #:
I think he will.

But he'll screw up so badly that it's easier to do it from scratch than to fix it.

 
There won't be any locomotive. There will be some left drawings of unsuitable assemblies, and not all of them.
 

AI lacks empirical experience of interacting with the world.

For example:

  • AI can state the laws of mechanics in formulas, but can it intuitively understand that the rounder the wheels and the smoother the road, the easier the cart rolls?
  • AI can know physics in scientific terms - speed multiplied by time equals distance, but can it feel that the faster you go, the further away you will be?

And how can it understand a statement that contradicts the laws of physics -"the quieter you go the further you'll go"?


A human starts learning physics from empirical interaction with the world and only then learns formulas. AI has only data/formulas at the beginningand only then interacts with the world. Does it impose limitations on its capabilities? This is difficult for a human to understand.

Let's try to understand it further.

Reason: