AI = What Would The Internet Do?
The most ignorant thing possible, of course.
This video/podcast from Adam Conover was interesting:
There’s been plenty of talk about shortages of chips and energy to run ever-more-powerful AI systems, but something new to me was the discussion of how pretty soon the LLMs will have ingested all of the available training data, after which it will be difficult to make them any “smarter.” And recent updates to Chat GPT that provided only marginal improvements do not give the impression that the technology is on some kind of Moore’s-law exponential curve of improvement.
Thanks for sharing! Both Adam and Ed helped to articulate a lot of the gut feeling I’ve had about all the AI hype that’s been going around the past few years.
It was very interesting. I’m also always happy to hear a prediction about FB’s death that has some reasoning about it. 
If it is useful then by definition it is not bad data. If it is actually bad data then there’s no way around the “Garbage In, Garbage Out” problem.
One of the great dangers of the AI hype is that it creates a superficial appearance of competence. It will spit out a web search summary or an article or a research paper or a business analysis that reads like something an intelligent human would write but is fundamentally wrong.
Adding to the problem are the tech geeks who think that AI will somehow eliminate foolish human biases even though the algorithm is just incorporating and even amplifying the biases of its creators or the people who produced the data it was trained on.
… there is another kind of exponential curve
Crossposting here, too
More invisible and ai reform uk candidates BS
This topic was automatically closed after 5 days. New replies are no longer allowed.
