AI is software, so it's under the umbrella of robotics anyway.
You joke, but that's how it worked for mankind. We, as a species (*), learned from ourselves. Strangely, it worked. Sadly, it had a few detours along the way, which cost "a few people" their livelihood and lives. I'd point out Galileo and the 30-year war as prime examples of bad detours.
That is not a joke, not really. It only worked with mankind because mankind did adjust the mental model of the world against reality (or, reality adjusted that model by adjusting mankind). That feedback is missing for AI which only learns from AI. It worked for Alpha Go, because it was in a simulated reality with correct feedback. For AI writing news articles, Wikipedia, Blog Posts, all learning from each other, there is no such corrective. The danger is that this leaks back into mankind's perceiving of reality, because we are Pan Narrantis, story telling apes.
www.linkedin.com
Actually, that is spot on. Creating content for datasets, that get fed to ChatGPT, that has been outsourced to India. We have enough morons in US to feed the model of recursive stupidity, but they got expensive as a workforce. So now the content farming task gets moved to a place where time zone and cultural differences (like deciding what's polite, what's not, and what the appropriate reaction would be) only compound the the problem of stupidity on ChatGPT and make it even worse. And of course, that is embarrassing for Amazon to admit. Did they use ChatGPT to decide that this is a Good Idea, or what?Amazon disputes the report that 1000s of Indians actually powered the AI-Computer Vision check-out free shopping tech in Amazon GO and Amazon
(Business Insider) Amazon's Just Walk Out technology had a secret ingredient: Roughly 1,000 workers in India who review what you pick up, set down, and walk out of its stores with. The company touted the technology, which allowed customers to bypass traditional checkouts, as an achievement powered e
Maybe ChatGPT also has "support" from humans.
That doesn't make any sense. Amazon has/had a fairly narrow scope use for their AI and it didn't work well so they used the cheapest labor they could find that can do the job. ChatGPT on the other hand has a much much wider scope and was trained on a massive corpus of data -- at least 570GB of data from the internet (webpages, articles, wikipedia, books, blogs, etc. etc.). Newer versions of GPT and similar tech will be slurp in even more data. Even a few thousand engineers anywhere in the world simply can't provide that variety and bulk of data.
Y'know, I would not put it past Amazon to rely on ChatGPT internally to try and create an automation of data analysis for a small scope. Even that required some cheap human labor to massage and farm the data so that later analysis can be automated.
