Presumably this all goes into a giant, up to the minute training data set.
I wonder what such an AI could be used for, any speculation?
> So the rate of biological evolution in humans, is about a bit a year. By contrast, there are about 50,000 new books published in the English language each year, containing of the order of 100 billion bits of information. Of course, the great majority of this information is garbage, and no use to any form of life. But, even so, the rate at which useful information can be added is millions, if not billions, higher than with DNA.
Source (https://qz.com/1229362/stephen-hawking-argued-that-humans-ev...)
A lot of AI behaviours today is just emergent. Studies show that the results improve with more data. It seemed unlikely, and that there would be a cap somewhere, but apparently we haven't hit it yet.
However Meta has some kind of AI and a ton of data and yet theirs is extremely bad. So bad that people quit FB over it. I would assume that Meta has a tech team as competent as "China" and more data than TikTok, but they're not faring well.
training set 1: "THEY TOOK OUR JOBS!" "THEY TOOK ER JERBS!" "TERK YER JERB!"