Hacker News Reader

Comment by dpflan | Hacker News Reader

It’s the best demo-able software today that lets the customers’ minds go wild with the possibilities.

But then the nuances of human abilities in commercial settings, like call center, become realities and obstacles that the software can’t surmount. Costs rise, productivity not so much, and then legal gets involved and it’s even more costly.

View on HN

This is the best description of the current state of affairs. But who is going to risk not invest on what could be another possible major breakthrough in 3 to 5 years? Either from a GPT5 or another source? It's not like all the brainpower that left OpenAI recently, suddenly converted to farming...

At least load on NVIDIA options....

Banks that didn't have FOMO in the run up to 2008 didn't end up in bankrupt from it. Investing from a place of fomo is a bad sign.

A [most] banks that did got bailed out

> But who is going to risk not invest on what could be another possible major breakthrough in 3 to 5 years?

3 to 5 years is an awful lot of quarters . . .

I like to measure my time in PhD's ...3 to 5 years, means the major breakthrough could be just at a distance of 0.5 PhDs

It feels a bit like the 1999 dot com bubble. The internet was real but valuations got ahead of the reality.

There is a lot of money to be invested these days, 2017 tax breaks and covid stimulus...

The question is, on call options or put options?

Plus there's the reality that all of this mediocre capability requires VAST inputs of energy that are totally unsustainable.

I've said it here before and will again, AI is just the pivot from crypto that props up NVIDIA prices and lets tech bros speculate on a market that simply will never materialize.

As somebody who recently used Claude Opus to help quickly and affordably solve complex problems in 3D design work (OpenSCAD) while designing a marketable product, I have my doubts about your analogy. Prior to the availability of affordable and accessible AI, I could have taken some Bitcoin and paid somebody smarter than me to help me with this work, but I guarantee it would have been less economically viable, both in terms of financial cost to me and in terms of energy consumption costs.

If you need it to do small standard tasks that normally requires an expert, then these AI can often solve them for you.

If however you need an expert to do a longer term project, then these AI can not solve them for you. It does replace some jobs as you say, but most people don't work on short term project consultation so overall it doesn't matter much.

> and in terms of energy consumption costs.

Did you include the cost of training the model? I'm pretty sure the total investments that goes into AI today way outspend the cost of such small consultation gigs that AI can replace currently. Customer support is the biggest moneymaker so far and even that is hard to do right with customers suing over lies from the AI bots.

Wait until you hear how much energy a human worker consumes...

AI is legit, but you are right that currently it's too expensive and being propped up by investments paid for in compute from cloud providers. We're making rapid progress in bringing the compute requirements of high level models down though, and eventually we'll get over this need to make supermodels and start making distilled models for specific tasks, which will provide the benefits without the drawbacks of all knowing models.

I agree, AI/ML can be very useful.

As expected, the replies to this comment are about the alleged speculated or anecdotal utility of "crypto" or "AI" to solve problems. However the replies fail to address the substance of the comment: money, the personal enrichment of a selected group of individuals. The motivation behind these "innovations" is money, not problem-solving. As the comment indicates, "AI" actually _creates_ (not solves) _real_ (not speculative) problems for which its proponents have no solutions. Both "crypto" and "AI" seem to evade cost versus benefit analysis because the costs are real while benefits are pure speculation. As their proponents enrich themselves on what amount to pre-payments for something that may never be delivered, the public is expected to believe that the benefits will arrive "real soon" and will convincingly exceed the enormous costs society is now paying and will continue to pay.

Compounding matters is the fact that it has become really easy to roll out half baked shit at ridiculous scales.

The marketing mega machine then uses the Scale to pretend its not half baked shit - "hey man just look at how many people are using it".

Reply on news.ycombinator.com