Matisyahu's Blog

Oh the hysteria from the usual suspects in politics and the media – the never ending avalanche of hyperventilation about how there is a race for AI between the United States and how the United States cannot afford to ‘fall behind China’ because of…well, visions of what they believing AI is capable doing and by AI we’re actually talking about LLMs. Behind all this nonsense I think it is important to understand the difference between China and the United States – the United States is trying to make a business off providing AI where as the Chinese want to integrate AI into their business to make their products better.

On one side you have people convinced that if they throw enough money at a problem then they can corner the market and ‘name their price’ – remember Sam Altman talking about AI being like a utility? yeah, that is what I think they’re hoping to be – OpenAI and Anthropic provide the foundation models then charge businesses who will then build products on top of those models. Part of that cornering of the market involves both throwing huge amounts of money at a problem in the belief that they’ll have an unsurmountable lead and the other part is the hysteria about safety, the need for regulation, claims that the next model is ‘super dangerous’ etc. (I’ve compared the nonsense to a chicken wing restaurant boasting about how they have the hottest wings in town along with the theatrics of going as so far as getting customers to sign a legal waver because their wings are so hot and dangerous).

The second part isn’t just about pumping up excitement and drumming the investment community, most of who are as thick as two short planks (even that is an insult to planks because at least you can turn planks of wood into something useful), but the other part is about creating an environment where politicians are compelled to keep out of the United States the free and open source models that are getting closer and closer to the capabilities of the proprietary models made by OpenAI and Anthropic. The best description of such behaviour would be a non-tariff barrier – creating hurdles so that for most businesses they throw in the towel and not bother entering the United States market and instead focus their energy somewhere else.

Getting back to the the difference, the comparison I made in the first paragraph only really applies to OpenAI and Anthropic because if you look at the big players – Google, Microsoft and Apple, the value isn’t in the AI but how the AI integrates into their products. The value of Gemini is how it integrates into the Google Workspace, with Microsoft they’re working on their own in house models (Microsoft have five of them) then there is Apple that worked in collaboration with Google (model distillation) to improve their own models. I assume that as Google’s models keep moving forward that Apple will keep doing model distillation on the newer models then deploys those models when they push out updates for their platform every 2-3 months.

I think with he recent information we’ve had leaked about OpenAI’s alleged financials along with rumours of Meta pulling back on capex spending along with leasing capacity of the unused capacity to the market (the comparison I made on BlueSky is comparing the move to that of a fast food store using up the 5″ burger patties for 4″ burgers rather than cooking more 4″ burger patties – it is better to do that than lose a whole lot more money by having to throw out 5″ burger patties) and at least get some money back from their unwise investment.

Long term think there will be a correction, the AI companies will scale back, no more subscriptions in favour of pay as you go API access along with ending the free consumer version of their services as investors start demanding they make themselves profitable. It’ll be interesting to see if, once they remove subscriptions and free consumer access, whether there is a viable business model and if there is a viable business model I have a strong inkling that the business isn’t going to be worth $1 trillion let along half of that amount. Keeping in mind that essentially what these AI companies are offering are development tools with an AI service slapped onto the side of it – not particularly all that exciting in the grand scheme of things.

Regarding the AI washing, we’ve already seen some businesses come out and admit that their layoffs had nothing to do with AI and everything to do with the fact that hey had been over hiring for over a decade simply to keep talent away from competitors but now shareholders are demanding a better return on their investment. Maybe we’ll finally see more businesses be honest instead of hiding behind AI. when thy restructure because it is one thing to be laid off from your job but it is another thing to have your employer lie to your face about the reasons behind the decision.