Gm AI is super overhyped right now
Discussion
Why you think itโs overhyped ? ๐
Itโs been overhyped for decades, I first noticed this when I was in uni in the early 2010s.
But recently, with all the chat GPT stuff, every one and their mother is trying integrate AI/ML into their business and most are going to be disappointed with the result
Thatโs an interesting take, I do agree that some social media influencers might be overhyping it.. itโs growth in the last 3 years is inarguably something to be proud of and excited about, only those that know how fast things move from now on forward have a real sense of what could be accomplished by letโs say 2030.. itโs only gonna get much better at a more significant pace. Academics wonโt always see this growth as fast as enterprises do simply bc of the capital difference between both parties, but it doesnโt mean the growth isnโt there.
For sure, progress is always cool and exciting.
Thought experiment: what do you see AI being used for in 2033?
I see it being implemented in nearly every farm around the country in order to better produce crops in the most efficient n sustainable manner possible, this includes but isnโt limited to marijuana. Also see it being used for logistics, in commerce as well as other areas like military. The propaganda machine will def use it and exploit it. Egalitarians like Europe will also benefit but but wonโt dominate like U.S and China which have easier ways to mobilize a ton of money. I see every industry adapting and using it in small ways like weโve already seen, GPTs might be the craze right now but this will come in waves.. next one is more interesting than this one for me. How about you ? What do you think itโll be used for in 2033 ?
ASI is what Iโm most excited for ngl
With chat gpt as a model, I definitely think weโll have really solid AI assistants in a decade. Siri and Alexa but with greatly expanded use cases. But these are essentially just language models: understanding some inputs, searching for info, and then giving it back in an understandable format.
But I donโt see what you described for farming, for example. Thatโs essentially an engineering/optimization problem. Same for logistics. Capturing the right data is the hard part. Once you have those, weโre already able to find the optimal solution
I canโt go too into detail on how it works since itโs part of my work but itโs essentially self optimizing and regulating itself, shared data is a big part of it. I agree that capturing it is the hard part but distribution and implementation of it with the use of Ai is what I was talking about, no need for humans to find the optimal solution, thatโs the computers task.
Need systems that can do logical reasoning and planning with gaurauntees, otherwise you wonโt really be able to count on the assistant to take care of things you actually care about.
Thereโs work in inductive logic programming (ML on symbolic data) and neuro symbolic approaches that are very interesting. I think weโll see new classes of systems in the next 5 years or so that combine much of the reliable decision making work thatโs been going on since the beginning of AI with NN learning progress from the last decade
you talking from an investment perspective?
Hmmm idk, but probably
The reason I said it is because at work senior people are trying to shoehorn AI/ML where it doesnโt fit.
And looking at stats/data science job postings, 95% of them mention AI. I would bet most of these companies donโt have the data to support such a use case