[fotodoug2]

The car was shot on a white background with no reflection...AI added the background and reflection.

[fotodoug2]

...and on a rooftop in Paris.

[KensRedZed]

Doug,

I prefer your rooftop pic. It's original.

AI can't come close to your glass reflection tricks, and backgrounds. YOU can beat the pants off AI with both eyes tied behind your back. Or something like that... 

[Henk]

AI is not that I yet, isn't it?

There's something wrong with both pictures, especially the rooftop one. A sinking sun and no shadow casted on the car?

Also in both pictures the lower front of the car looks wrong, melted with the floor.

Like you non AI pictures much better 

[fotodoug2]

Henk, your right...AI isn't really there yet, but it's close.  If I hadn't said that about using AI, would you have thought there was anything weird about them?

[JasonB]

I think I might have noticed something was wrong with the first photo, even if you hadn't mentioned AI.

The reflection cuts off the bottom of the tyres, and the wheel arches are way out, particularly the rear arch.

[fotodoug2]

Thanks Jason!  You are right, and the reason might be that I have yet to master how AI works in Photoshop.  So far AI seems to manipulate everything as a pattern, and although it reasonably represents the car in the reflection, it can't get the printing right...it just turns it to blobs of the correct colour.  It's early days...remember when we used to have to connect our telephone cord to the computer to get the internet?  

[fotodoug2]

Here's a non-AI shot...with a nice, crisp reflection!

[Henk]

That last photo looks fantastic, with your skills you don't need AI.

A long time ago, 1983, I've been into AI during my study Informatica. The subject was chess, a great game to think about what intelligence is. Most, if not all, chess engines work brute force, they just calculate all the possible moves and evaluate every possible position on the board. With enough computing power the engine is strong enough to beat most humans. 

But it's not the way a human thinks, there's is a very interesting book by A.D. de Groot who tried to understand how a grandmaster thinks. He did a lot of research and in the end concluded that a chess grandmaster thinks in patterns, not single moves: https://www.degroot.ai/a-d-de-groot/

AI systems nowadays need to be fed with a huge amount of information, and then the neural network computer will be able to spit out stuff in such a way that one could think it has a certain intelligence. But it is all still based on the input, the AI is still not self learning like a human. Maybe in the (not so) far future the machine will be really intelligent but until now it cannot produce anything that was not fed to him in one way or another.

Still AI can not come up with Pythagoras rule let alone Einsteins relativity theory. So until now I wouldn't call it very intelligent.

[fotodoug2]

Thanks Henk...I think my AI skills are so limited at this point that the results are as you would expect.  In my own defense, I don't think AI bots are looking at slot car photography for information!