My take is that if AI really improved productivity, then that makes labour effectively cheaper. Usually, when labour is cheap, then you buy lots and lots of it so that you can get a lot of things done.
I think it's called the Law of Demand?
Companies over-hired during COVID because the money was free thus making the labour cheap. Why are they not over-hiring now, during a time when workforce productivity is supposed to be at an all-time high?
Instead we are seeing layoffs, blaming AI, because the free-money chickens have come home to roost. These CEOs want to save face and not admit they were short-sighted during COVID.
But it isn't cheaper yet, right? Your current salary was defined pre-AI.
Now, though, maybe you're only worth a third of what you were but you can't live on a third, and your employer can't just start giving you less money, so... they lay you off then in a year or three when the tailspin starts to hit bottom, they hire Younger You for ~$X/3.
If AI makes people more productive then labor is cheaper than it was pre-AI, even at pre-AI salaries, because you're getting more done at the same cost.
3. Your profit from this particular employee is z = y - x.
So far so good. Let's assume z > 0, though of course it's not an easy statement to make with many roles, that are more like investments. But let's assume they're good investments and you're confident z is positive.
Now, if the same employee produces 2y, but doesn't receive a raise, z just improved significantly, it more than doubled. So effectively, labour just became cheaper in relation to the value of its output.
If it was that simple, layoffs would hurt profitability significantly.
Now what if you can't translate improved productivity to additional value? Simple example would be an agency with a fixed contract volume. If increases in y can't be realised, e.g. by finding more business, then the only way for the company to realise the gains is to reduce x, i.e. layoffs. z goes up right away, no business development required.
I think it's a defensive stance companies are taking. The economy is not great, they're hitting the breaks on investments, increasing their runway, shrinking to force the organisation to become more efficient. Once they're ready to invest again, they can always hire again. But I read layoffs-because-AI as "we don't know what to invest in right now, so we'll buy some time to figure that out".
> make so much more money doing the same job in the US than anywhere else.
Unless you are comparing with top-paying faang employers, the difference is not as stark I think.
Once you deduct the higher cost of every day things, medical expenses, education fees, and other social safety net stuff that you typically get for free in EU, and compound that with a weak dollar, you'll see that the typical EU salary is probably not all that bad. Otherwise, as you say, companies would have outsourced to EU rather than India for example.
Regardless of whether there is value in chat history or not, for some people it is important.
Back in the day during the music streaming wars there were tons of "move your playlists from A to B" services. Streaming services could not hold on to customers because all their playlists were on there.
I'm sure that similar services will pop up for chatbots.
Also, you can always just ask your chatbot to generate a file with your chat history, given that it's all part of the context anyway.
The moment openai starts charging for their service properly, people will start shopping around.
See power users such as devs with coding assistants that have model selection dropdowns allowing you to switch on a whim. There is zero loyalty or stickiness in the paying user crowd.
Ads are a little more insidious, and normies aren't nearly as allergic to them as they should be. But whether openAI can achieve their revenue targets by ads alone is a different question.
I am starting to believe that OAI might actually succeed at getting per token inference cost to where it needs to be. Or that it's already there in principle.
Wafer scale compute is a very big deal. Most of HN is probably still unaware that you can get tokens out of one of these devices right now via public API offerings.
During my Hyundai Kona's service last year the tech kicked the tyres, shone a flashlight underneath and checked all the fluid levels. Then they washed it and told me to come back next year.
I paid like €120 for it.
Not sure where your "ev services are expensive" rhetoric comes from.
Hand to heart honest, I had a listen to this, and this is the only version I know. This was a very popular song when I was in university in 15 years ago.
Hah, I wonder how thick a German, Dutch or Afrikaans dictionary would be if it included all possible spaceless compound words. Literally any concept can be compounded together to make a new word.
Roovleisslaghuisinspekteur =
Rooi = red
Vleis = meat
Slag = butcher
Huis = house
Inspekteur = inspector
"Inspector who controls the quality of red meat in butcheries"
reply