what will we actually use ai for?

we're currently in the phase of ai where we just HAVE to implement it in everything. who cares if it makes sense, let's add a chatbot to google maps. most of these features will probably be gone within 5 years. but this has got me thinking, what features could we actually use (that would be useful to society), and which can we throw away?

image generation

ok, im just going to get this out of the way, we can throw out text to image. however, i think the technology for them could still be used. image upscalers are cool and useful, as long as you aren't trying to rewrite history. also, being able to erase objects without the ugliness of content aware scale is nice.

text to video

we can get rid of this one too. i can't think of a single benefit that it has, like at all. the only thing that it can really do is make videos with the imperfections of ai, which i understand styilistically (im not seeing a red line under that, so as far as im concerned i spelled it right), but i think all the negatives of the tech justify getting rid of it completely.

text to music

this can also go away. ai companies are trying to sell you a future where instead of asking for a playlist full of a type of music, you just ask for a type of music and the computer endlessly spews it out, tailored just for you. this is stupid. the cool thing about actual songs is that you can listen to it, save it to your playlist, and talk about it with friends. if you have an endless stream of music, what are you going to do, screen record it?
also, real songs have thought and emotion put into them, something that a computer cannot do. it can only make something sad completely by luck

voices

contrary to popular belief, i don't hate ai voices. it's basically the same as text to speech. the only difference is that i think you shouldn't be able to just take someone's voice and clone it with no effort.

text summarization

this one is good. summarizing things or having it be rewritten in fortnite terms is really nice, especially if you don't really understand a subject.

natural language processing

i don't really see any negatives to this one, except for biases (which is just a large part of ai in general). you can search "why does my tv look like that" and google will tell you to disable motion smoothing. hooray!

object recognition

okay yeah, this one can be used for advertising, but the benefits heavily outweigh that. first off, you can write alt text automatically for images that don't have them. this already is a game changer for people with vision issues (that doesn't mean you should stop writing alt text). you can also just search for an object in your photos, and find it. with that being said, for things like foraging, this technology is not very reliable.

chatbots

chatbots can remain, as long as they stay as their own thing, instead of being integrated in everything. they have some practical uses, but they can get things wrong too much. gpt-4 still things you can take a screenshot on an apple tv by pressing home and menu at the same time.

deepfaking

no.