> If humans can be fooled by artificial intelligence in pretty much all areas,
This is just preposterous. You can be fooled if you have no knowledge in the area but that's about it. With current tech there is, there can not be anything novel. Guernica was novel. No matter how you train any probabilistic model on every piece of art produced before Guernica it'll never ever create it.
There are novel novels (sorry for the pun) every few years. They delight us with genuinely new turns of prose, unexpected plot twists etc.
And yes we have cars which move faster than a human can but they don't compete in high jumps or climb rock walls. Despite we have a fairly good idea about the mechanical workings of the human body, muscles and joints and all that we can't make a "tin man", not by far. As impressive as Boston Dynamics demos are they are still very very far from this.
> With current tech there is, there can not be anything novel.
I wasn't talking about current tech, which is obviously not at human levels of intelligence yet. I would still say that our progress in the last 100 years, and the last 50 in particular, has been astonishing. What's preposterous is expecting that we can crack a problem we've been thinking about for millennia in just 100 years.
Do you honestly think that once we're able to build AI that _fully_ mimics humans by every measurement we have, that we'll care whether or not it's biological? That was my question, and "no" was my answer. Whether we can do this without understanding how biological intelligence works is another matter.
Also, AI doesn't even need to fully mimic our intelligence to be useful, as we've seen with the current tech. Dismissing it because of this is throwing the baby out with the bath water.
> What made you think that is measurable and if it is then we can build something like that ever?
What makes you think it isn't, and that we can't? The Turing test was proposed 75 years ago, and we have many cognitive tests today which current gen AI also passes. So we clearly have ways of measuring intelligence by whatever criteria we deem important. Even if those measurements are flawed, and we can agree that current AI systems don't truly understand anything but are just regurgitation machines, this doesn't matter for practical purposes. The appearance of intelligence can be as useful as actual intelligence in many situations. Humans know this well.
Yes, I read the article. There's nothing novel about saying that current ML tech is bad at outliers, and showcasing hallucinations. We can argue about whether the current approaches will lead to AGI or not, but that is beside the point I was making originally, which you keep ignoring.
Again, the point is: if we can build AI that mimics biological intelligence it won't matter that it's not biological. And a sidenote of: even if we're not 100% there, it can still be very useful.
Again, the point is: you can not build AI that mimics biological intelligence because you do not even have any idea what biological intelligence even is. Once again, what's Picasso's velocity of painting?
This is just preposterous. You can be fooled if you have no knowledge in the area but that's about it. With current tech there is, there can not be anything novel. Guernica was novel. No matter how you train any probabilistic model on every piece of art produced before Guernica it'll never ever create it.
There are novel novels (sorry for the pun) every few years. They delight us with genuinely new turns of prose, unexpected plot twists etc.
Also harken to https://garymarcus.substack.com/p/this-one-important-fact-ab... which also happens to include a verb made up on spot.
And yes we have cars which move faster than a human can but they don't compete in high jumps or climb rock walls. Despite we have a fairly good idea about the mechanical workings of the human body, muscles and joints and all that we can't make a "tin man", not by far. As impressive as Boston Dynamics demos are they are still very very far from this.