You've probably heard that AI chatbots can completely fabricate quotes. It happened yesterday in a DU thread.

[haele]

32. I miss the old Google. Hell, I miss the old Alta Vista.

When my oldest grandchild was around two and a half, we were building a toddler's Ikea furniture set, a step stool that could act as a small bench and table she could eat from when doing TV Dinner night, a toddler's chair, and that ubiquitous 28" square side table for a craft and learning table.

She strongly insisted on helping. Okay, learning experience.

The step stool was the only Item that survived her "helping", because she got bored and hungry, wandered off to get a snack - and we did all the difficult work and only let her tighten a few screws we had started for her.
She was really upset the chair wasn't working out and the table legs kept falling out.
So, we told her Grandma the engineer would look it over, see where the factory got it wrong, and fix the factory problems, then she could finish the furniture.
I got another chair and table (and swedish meatball dinner kit) after work the next day, put them mostly together in the family car before coming in, and Grandpa Lazy oversaw her "finish" putting her furniture together while Grandma put together the dinner.

AI bots remind me of the average young toddler putting together more complex Ikea furniture.
Physically and conceptually not quite there yet...

The question is whether or not they can actually grow, or if they remain LLMs with a veneer of autonomy.
Since the variety of functionality and complex operations that enable brain function and human thinking and intuitions (let alone those of less complex species) is still not understood at all, it's rather difficult for me to believe AI as it is now, based on 1's and 0's, can reach the actual self awareness and evolutionary learning drives that the biological based computer - a brain - can achieve.