Welcome, new readers! It seems like a lot of people hate running as much as I do, which is extremely validating. This is a safe space. You don’t have to pretend that horrible things are fun here. This is a place where I sometimes rant about things that suck, like
I'm so afraid you are correct that it is improving, if 'improving' is the accurate concept, faster than can be comprehended. The attempts to put guard rails on the development and use of AI are laughably far behind. In a few months, if not a few weeks, your prompt will give you a photorealistic photogenic cute as a button baby giraffe or my name's P.T. Barnum!
I dunno if you've ever tried working with an illustrator, but the parallels are instructive. A few years ago we were trying to get some pictures done using Fiverr and similar platforms, and the results were never what we expected, often rubbish, and sometimes usable.
Obviously, the ways AI messes up are different from the ways in which people "mess up", but it's instructive to compare and contrast, because I think that people are holding AI to impossible standards. When you ask a human/AI to do something, they will do it imperfectly, sometimes because they themselves are imperfect; mainly because the medium (language) through which you ask them is pathetically underspecified; and often because of faults in the way that you asked.
None of which is to excuse the faults of AI, of course. It usually doesn't do what it says on the tin. But why should this wild new white hole of stuff behave like we want it to?
I'm so afraid you are correct that it is improving, if 'improving' is the accurate concept, faster than can be comprehended. The attempts to put guard rails on the development and use of AI are laughably far behind. In a few months, if not a few weeks, your prompt will give you a photorealistic photogenic cute as a button baby giraffe or my name's P.T. Barnum!
This made me laugh a lot. Which I haven’t much about generative ai, honestly.
I dunno if you've ever tried working with an illustrator, but the parallels are instructive. A few years ago we were trying to get some pictures done using Fiverr and similar platforms, and the results were never what we expected, often rubbish, and sometimes usable.
Obviously, the ways AI messes up are different from the ways in which people "mess up", but it's instructive to compare and contrast, because I think that people are holding AI to impossible standards. When you ask a human/AI to do something, they will do it imperfectly, sometimes because they themselves are imperfect; mainly because the medium (language) through which you ask them is pathetically underspecified; and often because of faults in the way that you asked.
None of which is to excuse the faults of AI, of course. It usually doesn't do what it says on the tin. But why should this wild new white hole of stuff behave like we want it to?