hi this is an interesting questiono
hi
hi...again I think
Yeah
well how do you define true AI? I guess you would have to define intelligence
True
I think there is a difference from the intelligence from machines and intelligence from animals
there's a lot of things we call artificial intelligence And they might not be
this is true I guess a machine is intelligent when it does something like thinking or problem solving similar to how a human might
I agree We're not there yet
Ditto. You could pick an example, like a human reading a book and look at different levels of thinking about, or analyzing the book there would be surface level things that a computer could pick up easily
But there are many others that are harder.
Right. It would be much harder for the computer to read between the lines and come up with the overarching theme or symbolism
And you can put many people in a room to analyze the same data and have different interpretations.
That's true! A computer's interpretation of a book would probably not change over time
If the computer start making interpretations or something like that it's because it's picking someone bias.
whereas a human's interpretation can change based on their memory, experience, and other books they've read
That's a great point.
I wonder if you could evaluate a computer's bias based on the person(s) who built it
I think so
Or some computers you feed them data to 'train' them to do certain processes which we might call a 'thought' process
I read something about some AI they were trying to train and became hateful in a few hours.
Yeah or it becomes biased against certain minority groups because the data they trained it with were biased
Yeah. I also saw something about an AI that was being trained to recognize pictures, and after a while it was categorizing pictures of black people as monkeys.
How would you prevent something like that? if the learning was unsupervised like you describe. That would be hard to do.
This kind of things can be very bad, especially if they are using these AIs to decide things like credit and other important decisions.
Yeah hey I also remember reading that! There were a few others too. I remember it couldn't distinguish between certain dogs and muffins
Hahah
I agree. Although the government and other large entities move slowly, they will still want to make ai's to make important decisions
Absolutely, and if you think about it, the bias would cause prejudice against minorities
Yup I remember reading about an ai they used to try and make court decisions and this was the problem
Yeah thats sad I gotta go nice talking to you