I think it's still some ways away. Technologically a lot can be done, but don't forget that it's still cheaper to have a factory full of low wages workers make clothing in Bangladesh than have some hyper advanced sewing robot do it. Even in the western world it's probably still cheaper to have someone pick up a toiletbrush and clean the toilets than to have some kind of autonomous robot do it.Κracus wrote:lol that's pretty funny but all they're doing is teaching AI how to spot humans who are actively trying to disguise themselves. Machine learning is exploding in terms of functionality at the moment and we're just at the start. I imagine I'll be seeing some impressive stuff by the end of this year, let alone 10 years from now. I mean, even in that article you posted I could not have imagined tech being able to spot humans 10 years ago. The concept of teaching an AI how to see the world seemed like a near impossible task when you were watching the DARPA races and competitions but along came machine learning and well... We're watching robots do fucking backflips and shit. Meanwhile I'm using AI to help me with my job on difficult tasks and cars are driving themselves... At this point it seems almost realistic to assume a mass produced robot can and will take over menial jobs in the very near future, along with not so menial jobs even including artistic ones.
As for your coding problems, don't forget that you have ChatGPT generate relatively simple bits of code for an isolated, context free environment. You can ask it to implement a sorting routine or write code that implements some sort of widely used interface or API (will probably be increasingly hard as the API becomes more obscure), but it cannot build an entire software suite that implements features and functionality that matches exactly what your customers are looking for. I also highly doubt that (at this point) it can write entire full stack application that's easy to expand and maintain and sticks to the (often complex) requirements a full team of developers has for the way the software is set up and written. I don't see ChatGPT take over my job as software engineer any time soon, certainly not the part that requires me to take a customer's requests and translate them to a functionally and technically appropriate implementation.
Lastly, for anything that requires creativity (and one can argue that software development does require a bit of creativity as well), ChatGPT can only create things based on content it's been fed before. ChatGPT won't, in its current form, come to some sort of extremely insightful conclusion or creative expression that's unique, let alone revolutionary. If you've kept an eye on all the AI generated art posted in the Q3W Discord, then one thing that's very noticable is that while it all looks cool and, from a technological perspective, is impressive, it's never something that makes you go "whoa, that's so fresh and unique and original". It always harkens back to some style or idea you've seen before (and probably quite a bit more than once).
