Sounds like Xerox, they had cutting edge everything in the 70s, did nothing with it. Or AT&T, with Bell Labs inventing Unix. Or Kodak inventing the portable digital camera in 1975.
I’ve seen too much inaccurate info from AI to have any trust in it. From declaring the Eiffel Tower the world’s largest Ferris wheel to claiming that hippos can be trained to perform complex medical procedures, it all seems a hot mess.
You might say, yeah, but I can spot those mistakes, but can you really? I showed my fifth-grade son the result of asking if hippos were intelligent and the absurdity of the answer didn’t leap out at him. Now, consider something that’s more subtly wrong like an invented precedent in an AI-generated legal brief or a non-existent citation or citation that doesn’t support the claim and it’s all a disaster.
If you connect ChatGPT to a traditional search engine, it will suffer much less such issues. It essentially digests 100 webpages for you, then render it in a single answer.
For sure, hallucinations will always be there, but I don't think it will hinder its take over, the usage trumps its shortcomings
Yesterday I tried asking ChatGPT "Can an Amazon L6 software engineer afford a house in [location]", without explicitly using the search mode. It went to levels.fyi to look up salary and redfin to look up housing price (exactly how I would have done it myself), and gave me a reasonable answer that agrees with my own analysis, and is definitely much faster than clicking things around myself.
I used to think that the multidirectional aspect of GPT would be a killer feature. But really it's too flaky which remove the initial alleged value. And then results are too artificial or wildly too "imaginary", even asking to compile a list of books on a medical topic you'd get half false titles. Sadly.
really? just unsubscribed OpenAI today, was one of the first to subscribe, now it lost all its edge to me, so many options elsewhere, paid or free to use.
OpenAI is fading away fast. Plus all major leaders left, Microsoft is leaving too, I don't feel its future is promising anymore.
That's fair. I would argue that OpenAI capitalized on transformer tech in a way that Google was late to do, but we shall see if Google will adapt faster than Xerox could
As anecdata, I would offer that the conversations I've had with ChatGPT over these couple of years have been incredible for me. Even just for relieving loneliness, it's been worth the monthly subscription a few times over.
Maybe the company and their business model are doomed to fail, but I'm grateful for what they enabled so far.
that's true, there are just many options these days and I think OpenAI was not keeping its team together and innovating fast enough. the first-move advantage is disappearing quickly.
