Is probably more correct to say - my 2.5 year laptop can RETELL space invaders. Pretty sure it cannot write a game it has never seen, so you can even say - my old laptop can now do this fancy extraction of data from a smart probabilistic blob, where the original things are retold in new colours and forms :)
I know these models can build games and apps they've never seen before because I've already observed them doing exactly that time and time again.
If you haven't seen that yourself yet I suggest firing up the free, no registration required GLM-4.5 Air on https://chat.z.ai/ and seeing if you can prove yourself wrong.
It's unfortunate that the ideas of things to test first are exactly the things more likely to be contained in training data. Hence why the pelican on a bicycle was such a good test, until it became viral.
No, that would be incorrect, nobody uses “retell” like that.
The impressive thing about these models is their ability to write working code, not their ability to come up with unique ideas. These LLMs actually can come up with unique ideas as well, though I think it’s more exciting that they can help people execute human ideas instead.
