I use it daily as a glorified search engine. Ever since Google decided that showing ads is more important than showing search results ChatGPT is much better.

It will continue to improve function wise. Its innevitable. Its cannot currently and may never be able to replace humans but can act as one heck of an assitant. Its scary as heck as already the training is not being done in an optimized, deliberate way that makes it as high quality as possible and those who control it when doing deliberate things are likely to do it in a way that is good for them but not for all. Im worried we won’t get the very good training in the same way software is often done badly because once it can do good enough, even inefficiently, then it will ship. Our best options will be free/libre.

eh, we’ve been here before.

• https://en.wikipedia.org/wiki/ELIZA_effect
• https://en.wikipedia.org/wiki/AI_winter

It’s what they’ll use to decrease the population when workers aren’t needed anymore

The technology is remarkable, the implementation is lame, the impact is happening too fast for us to adapt, the damage to artists and creatives is to cry for.

Someone posted a meme yesterday comparing it to The One Ring from LOTR and I think it’s spot on.

What AI is not: Human connection. A creative.

What AI is: A very fancy search engine that gives human-like answers. Which includes the fallibility. People keep saying “you’re better off reading the page”… are you tho? Why are we suddenly defending the academy of google and “doing your own research”? People who are hell bent on being idiots will continue to be idiots, AI or no. Though AI has some chance to exposing logical fallacies in one’s arguments. If fucking anything, AI can help smart people with the exhausting task of dealing with gishgallop disinformation tactics in online discourse. And I mean SMART people, those who know better than to just have AI print a wall of text on their behalf, and who know the subject matter but don’t have every piece of proof at their fingertips at any given time (of course an argument can be made that a smart person wouldn’t get into this kind of argument in the first place - but ya’ll know how you are).

Also, as much as I dislike Google, there’s no viable alternative to YouTube. And Gemini is way better with YouTube searches.

And I’m happy to not be forced to go onto the garbage sites of Wikia etc. whenever I need to look something up in a videogame.

And it’s an utility that can help with the more tedious tasks of projects. Which isn’t an universally good thing since that usually implies more productivity, which implies more consumption. And we’re already over-consuming. But AI can also help production of something actually life-enhancing, from people who actually care and aren’t just looking to produce more to generate more profits.

It’s a tool, and a toy. And unfortunately a lot of people aren’t equipped to use it.

This is the largest technology con job right now. NFTs failed, the “metaverse” failed, now a badly trained AI is the “solution to humanity’s problems”. This is just as stupid as religion.

Potentially useful for lots of thing, unfortunately, everyone seems fixated on stuff that is not ready for and probably wont be, or on hating everything involved.

Me? Mostly on the fence, hating most of what I see and hopeful for those useful applications.

It’s simultaneously awesome and overhyped.

I like how Yaron Minsky from Jane Street characterized LLM: “It is smarter than we expected it to be, but dumber than we needed it to be… It feels like something really dumb, but somehow memorized the entire internet.”

This is kind of what I feel: despite all these impressive BAR and IMO achievements, in my work, I feel they do a great job at parapherasing the internet, but fails when you need it to do something mildly intelligent.

Does it improve my efficiency? yes, but only at some very tedious and specific taskes, once I go slightly out of scope, it comes up with inelegant solution that I will need to rewrite from scratch.

deleted by creator

Slop

Machine learning in general is pretty awesome. It solves many problems behind the scenes, but even that side is overhyped.

We hoped it would solve hard problems, but it can’t. It solves boring problems. We hoped you could implement it easily, but it isn’t that straightforward either.

Generative AI for text, audio, images, and video is here, but the same problems persist. It doesn’t solve hard problems, no matter how hard we want it to. Also, implementation is harder than expected.

Then there’s the misuse of LLMs. Oh boy what a dumpster fire.

Tldr: I think current AI hype is trying to go to get to the moon by building a better ladder. It’s useful but for a few select things.

I’ve managed to automate some boring tedious task (get measurements from a dozen different wikipedia article, do simple maths with them), something that would take about an hour to do manually took 15 minutes to argue with an AI and fact check after.

Creative stuff (stories, pictures, music) is amusing for a while but I suspect they are stuck in the same kind of uncanny valley robots have been in for a long time.