AI successes

[FishFright]

It's been posted elsewhere that AI will soon get a lot more expensive. I don't know why.

Is this true? Will it affect the casual user of Ch*tG*t, or the AI on Search Engines?

Maybe because they are all running at a huge loss?

 

[Rusty Nails]

Maybe because they are all running at a huge loss?

It's like a loss leader and giant user test base at the same time.

 

[Pblakeney]

It's like a loss leader and giant user test base at the same time.

…as well as data scraping.

 

[stowie]

It's been posted elsewhere that AI will soon get a lot more expensive. I don't know why.

Is this true? Will it affect the casual user of Ch*tG*t, or the AI on Search Engines?

It is not a certainty that AI usage will get more expensive, but as others have said, AI companies are burning through cash like Imelda Marcos in a shoe shop and at some stage the investors providing said cash might want to have a glimmer of hope at some kind of return.

Between 95-97% of AI usage is via the free tiers currently offered. That is one hell of a customer base currently not paying for the service. Optimists would point to a huge latent pool of users that could be monetised but pessimists might wonder how many of that 95% would actually be prepared to pay for it or simply go back to google search etc. In fact, established companies like Google could be in the best position with these customers since their AI search is integrated into their already successfully monetised platform so could continue with a free user tier. I think the bet of Anthropic/OpenAI/etc. is that casual users will be prepared to pay something for the service, they will also be able to monetise the enquiries somehow and that existing subscribers will willingly pay more for current service levels. I could be being cynical, but I do think the model may be getting companies hooked on AI for development and when they have lost their human knowledge through redundancy to start whacking up the prices. It could be said that any company executive willing to hand over core company IP development to a third party whilst getting rid of the human store of knowledge might deserve such a business model. But if this is the case, it relies on these companies getting "hooked" on the AI model before the AI company burns through the whole cash pile and that is very far from certain.


The AI world is moving very fast, and I won't be surprised if the big winners in 10 years are not the companies we know now. For example, people using AI for specific knowledge areas are starting to look at smaller, more directed models which may be able to be run on premise equipment to ensure security. As it was put to me, why would an AI model which is being used by a law firm need to be able to answer questions around Harry Potter films? And, more importantly, a directed model could be more accurate and less likely to base a law argument on, for example, something it learned from Reddit...

And then there is the thorny issue of customer lock-in. In the technology / software world, people will jump from one AI vendor to another pretty quickly based on model accuracy for the task and price. It could quickly become a race to the bottom with thin margins and then paying huge cash for gigantic datacentres would appear far riskier than now.

In summary of a long, rambling post, expect AI companies to try to charge more people more money for their services. And don't bet on the people to deciding to pay it.

 

[Psamathe]

It is not a certainty that AI usage will get more expensive, but as others have said, AI companies are burning through cash like Imelda Marcos in a shoe shop and at some stage the investors providing said cash might want to have a glimmer of hope at some kind of return.

Additionally there seems an increasing push back with increasing numbers of younger people deciding to not use it.
eg from Gallup https://news.gallup.com/poll/708224/gen-adoption-steady-skepticism-climbs.aspx

While a steady 51% of Gen Zers in the U.S. report using generative artificial intelligence at least weekly, negative emotions toward it have intensified over the past year. Anger about the technology has risen, while excitement and hopefulness have dropped. Even daily AI users, who generally hold more favorable views of AI, have not become more positive, according to a new survey by the Walton Family Foundation, GSV Ventures and Gallup.

Generation Z — those born between 1997 and 2012 and who are currently aged 14 to 29 — is not convinced that AI enhances creativity or critical thinking, and the majority believe it may come at a cost, particularly to learning.
...
Gen Z’s AI Adoption Essentially Unchanged From 2025
Gen Z’s use of generative AI in everyday life has been largely stable since March 2025.

 

[briantrumpet]

Additionally there seems an increasing push back with increasing numbers of younger people deciding to not use it.
eg from Gallup https://news.gallup.com/poll/708224/gen-adoption-steady-skepticism-climbs.aspx

Interesting to see Alastair Campbell say he's getting increasingly sceptical about it: I started vaguely sceptical, and from reading some coolly scary stuff from academic types (as well as all the financial/political aspects), my scepticism is increasingly turning fairly radical. It needs to be treated like smoking, thalidomide or asbestos, and the serious dangers taken into consideration when looking at the things it does do well.

 

[briantrumpet]

OK, let's recognise what it can do well... solve problems I've never even heard of.

 

[AuroraSaab]

It's going to be useful in science and maths, likely in medicine if controlled properly. It's going to ruin art, literature, and music. In fact it already is ruining them with AI generated music and imagery becoming the norm.

 

[Pblakeney]

OK, let's recognise what it can do well... solve problems I've never even heard of.

View attachment 17141

My cynical self is wondering, how do we know the result is correct?
Not saying it's not, but wondering.

 

[briantrumpet]

My cynical self is wondering, how do we know the result is correct?
Not saying it's not, but wondering.

Someone has pointed out that to qualify it now needs to be peer reviewed and wait two years (so that people have time to test it) before it can actually claim to have been solved.

 

[farfromtheland]

I'll bet it won't touch the Riemann Hypothesis.

 

[stowie]

Additionally there seems an increasing push back with increasing numbers of younger people deciding to not use it.
eg from Gallup https://news.gallup.com/poll/708224/gen-adoption-steady-skepticism-climbs.aspx

I'm not particularly surprised. Gen Z gets a lot of criticism (and quite a lot is probably well deserved) but they have had quite a lot going on. Firstly social media psychopaths pushing addictive apps with the sole overriding aim of selling more ad space and keeping them glued to an everscrolling screen, and now the same lunatics are busy happily pronouncing the end of work for humans, mostly in the lines of work that they were told was exactly what the country needed and worth getting in huge debt for the qualifications. No wonder they aren't particularly enthusiastic to embrace it.

I think the reality is much more nuanced. Any company outsourcing nearly all work to AI is basically handing over control of their IP development (whatever that may be) and therefore in the hands of third party companies who at some stage will be looking to make a profit by hiking AI use costs. There will be some execs so mind blowingly stupid and short sighted to do so, but the hope is that most will understand the need for humans and balanced development. Of course this may well not help gen-Z where the jobs market will want more experience than they can possibly offer in many cases.

 

[briantrumpet]

More about the Navier Stokes thing from a Professor Chris Combs

Some thoughts on the Navier-Stokes news from a professor of aerospace engineering: 1) take a deep breath 2) posts indicating we "solved Navier-Stokes" are incorrect and overblown. A very specific mathematical proof involving a niche scenario for Navier-Stokes has *potentially* been shown (that a perfectly smooth incompressible initial condition with finite energy could produce a singularity--note the assumptions here piled on to an equation set that already involves some assumptions). 3) there is no general closed form solution for N-S and that's not what this news is about 4) this is more of a mathematical curiosity than anything else and could matter a lot to mathematicians but basically changes nothing in the way N-S are used in practice 5) there are already very useful exact solutions to Navier-Stokes that greatly simplify the equation set with the right boundary conditions/geometry like laminar flow through a pipe or a 2D Taylor-Green vortex

6) in aerospace, we already treat N-S as an approximation in many instances and are well aware of its limitations. we don't outright solve these equations anyway and regularly chop off terms or simplify parts (at the expense of accuracy) to make them easier to deal with. 7) to be honest I have never found this specific problem to be particularly interesting because we know a core assumption of N-S is a "continuum" fluid where you ignore molecules and assume hydrodynamic scales >>> molecular scales. We know a real fluid cannot have infinite velocity or energy. But clearly you can push equations outside of their bound of validity and make them produce funky results and singularities.

 

[farfromtheland]

More about the Navier Stokes thing from a Professor Chris Combs

The n-devil is in the deatil innit.

 

[C R]

More about the Navier Stokes thing from a Professor Chris Combs

Those were my thoughts when I saw the post yesterday. Nice to see them validated.