Demand for AI "Surging"

[willendure]

But, yes - it's very very difficult work. My nose is constantly in academic papers, reading newsletters/tweets by PhDs, trying to stay abreast of the latest developments (last year is was LoRA, QLoRA, multi-modal, mixtures of experts, more advances in prompt engineering, etc) to a point where I took a 2 month holiday to completely unplug.

Lol. Current gig, we tend to all take 1 months holiday in the summer. I used my holiday to learn a lot about the current gen AI tech, on both a theoretical and practical level. Could have done with 2 months, or even more, but also nice to be able to be both relaxed and learning at the same time.

[willendure]

https://www.bbc.co.uk/news/articles/cx7dx48ev91o

"Salaries for those who can tackle these challenges have hit “ludicrous” levels, he adds, because they are so important."

Apparently. I don't see this being reflected in the contract market though? But perhaps for reasons discussed already above, or just general contract market levels right now.

[dsc]

https://www.bbc.co.uk/news/articles/cx7dx48ev91o

"Salaries for those who can tackle these challenges have hit “ludicrous” levels, he adds, because they are so important."

Apparently. I don't see this being reflected in the contract market though? But perhaps for reasons discussed already above, or just general contract market levels right now.

They are so important for whom? cause I read yesterday that all this AI malarkey isn't really driven by users, we are just getting it cause the creators decided so.

[Snooky]

They are so important for whom? cause I read yesterday that all this AI malarkey isn't really driven by users, we are just getting it cause the creators decided so.

Where I work we're using it to speed up dev and also starting to include it in the end user products so they can ask "what if" type questions and it'll work out what data to present and how. It's pretty impressive stuff. I should add that I'm not involved in that side of it!

[dsc]

Where I work we're using it to speed up dev and also starting to include it in the end user products so they can ask "what if" type questions and it'll work out what data to present and how. It's pretty impressive stuff. I should add that I'm not involved in that side of it!

I'm not saying it's not impressive and / or speeding up loads of tasks, it's just where the demand for this lies. A lot of development in this direction (AI) seems to be coming from the makers themselves and it's possible that this is simply because they are the ones that truly understand how powerful it can be and your average user is still on the level of "chatgpt tell me a joke".

[ladymuck]

If we build it, they will come...

The user doesn't know they need it yet. I certainly have no immediate applications for it, but then I'm still working on my one-line macro.

[edison]

Elon Musk has just raised $6B funding for his AI startup. All the usual other big tech players are committing to spending billions on R&D, more data centres, more GPUs etc.

At some point this investment will have to start showing signs of feeding into revenue. Otherwise the share prices that have zoomed up in the last 18 months will start falling as investors lose faith in the investment thesis that "AI will change everything (no, this time it's really different, honestly.)"

[mgrover]

Elon Musk has just raised $6B funding for his AI startup. All the usual other big tech players are committing to spending billions on R&D, more data centres, more GPUs etc.

At some point this investment will have to start showing signs of feeding into revenue. Otherwise the share prices that have zoomed up in the last 18 months will start falling as investors lose faith in the investment thesis that "AI will change everything (no, this time it's really different, honestly.)"

i mean we saw it a lot with blockchain and apps made with cheap money.

eventually when they don't turn a profit as the interest rates rose they got killed.

for me everytime I try to get AI to do my work for me it just leads to me doing more work.

Like how hard is it to generate all the possible permutations for a pure function.

[Gibbon]

If we build it, they will come...

The user doesn't know they need it yet. I certainly have no immediate applications for it, but then I'm still working on my one-line macro.

When you release that it will kill present AIs in an instant. Shirley I'm in line for some shares given all the promotion work I've done for free

[willendure]

Its like any major innovation - electricity, cars, flying. Before a certain time its just too soon to do those things and anyone trying will fail. But then some technological milestone is passed and suddenly the time to be in on that thing is NOW! AI is like that too I think.

I got my Masters in AI back in 1998. AI covered a lot of different areas back then. One of them was machine learning, but it was just a small part of the whole domain. But everyone knew even back then that it was the one that was really going to change things. The problem was computers were just not powerful enough, training data plentiful enough - although the techniques and algorithms needed to develop too, the basic "back propagation" algorithm is still the same as it was back then. Once computers and data got over a certain level, whoosh the whole thing can take off and NOW is the time for AI.

You are right though, its not driven by user demand. The users don't even know what is possible, so how can they know what they want? Its driven by the technology and what new things are becoming possible. A Cambrian explosion of new possiblities. The users will decide though, what is useful enough to pay for and therefore what selection function will be applied to the Cambrian explosion as it evolves. And by users I do not just mean end-users, the owners are users too - Facebook will pay for AI that is useful to their business, whether the end users like or want it or not, for example.