AI has been unleashed on the world in a profoundly irresponsible way. It's not generally leading to better products and solutions. But it is leading to a lot of people with money-- and who want to make a lot MORE money-- making lots of wild claims and hoping that leads to profit for themselves.
All this greatly muddies the waters. I am a tester. I have done some deep dives testing ChatGPT. I can tell you that whatever claims are being made on behalf of AI products, they are LARGELY UNTESTED. AI nurses? BARELY TESTED. AI tech support? BARELY TESTED. When I scratch the surface of AI I see shocking failures that are being papered over by hucksters and fanboys.
Can ChatGPT really pass the human tests the author spoke of? Is it cheating on those tests? Are the tests part of its training data?
In my field, claim are being made that AI can replace human testers. When I try to make ChatGPT test it falls right on its face. But if you don't know anything about testing, the demo would look plausible.
I'm sure a lot of of jobs will be lost, but not because AI is making anything better. It's because people in power have discovered that the law doesn't prevent them from making life worse for most people, at little risk to themselves. Yes, you can use AI to make illustrations. They are terrible illustrations; and they achieve what they do achieve by ripping off original artists.
World governments much pass laws taxing these companies and forcing them to compensate humanity for the theft of our collective intellectual property.
I'm less worried about theft of IP, and more worried about the energy cost for current AI systems. We're burning massive amounts of power for the current AI wave, and it is not all renewable.
I do think AI has a reasonable future, but we are not being reasonable about it today. I wish we could just punt it downhill for a decade, focus on getting our energy production to 100% renewables worldwide, let AI advance at a research-only level during that time, and commercialize it in 2035+, using whatever spare energy we have at that time.
Then we'd skip the power and money-based drive to get AI in everything today despite its poor quality and efficiencies.
All this greatly muddies the waters. I am a tester. I have done some deep dives testing ChatGPT. I can tell you that whatever claims are being made on behalf of AI products, they are LARGELY UNTESTED. AI nurses? BARELY TESTED. AI tech support? BARELY TESTED. When I scratch the surface of AI I see shocking failures that are being papered over by hucksters and fanboys.
Can ChatGPT really pass the human tests the author spoke of? Is it cheating on those tests? Are the tests part of its training data?
In my field, claim are being made that AI can replace human testers. When I try to make ChatGPT test it falls right on its face. But if you don't know anything about testing, the demo would look plausible.
I'm sure a lot of of jobs will be lost, but not because AI is making anything better. It's because people in power have discovered that the law doesn't prevent them from making life worse for most people, at little risk to themselves. Yes, you can use AI to make illustrations. They are terrible illustrations; and they achieve what they do achieve by ripping off original artists.
World governments much pass laws taxing these companies and forcing them to compensate humanity for the theft of our collective intellectual property.