AI efficiency, supply, and demand

Without getting into a detailed discussion of Jevons Paradox, let's look instead at the reasons for the renewed interest in the so called "paradox."

The reason the Jevons narrative has been deployed is to counter the rather negative sentiment caused by investors thinking that they were lied to about the power and semiconductor needs of AI. By investing massive amounts of money into Microsoft and OpenAI and the energy plants that are supposedly needed to power them, investors have put a lot of money on the line.

I find it very convenient to talk about Jevons in this way, since that is not how it really applies.

What we are actually seeing in the AI space is competition, not some process of paradox that makes money invested in OpenAI "safe".

Jevons might apply generally to AI (though, there are reasons to think it does not), but that doesn't mean that OpenAI and American tech firms will benefit from the expanded use.

As we have outlined previously, there are regulating capitals in competitive capitalism:

1. The regulating capital is the capital that can produce a portion of the product at the lowest price and highest profit margin.
2. The regulating capital is not always the same firm over time.
3. The regulating capital sets the price for the rest of the market.
4. The older firms cannot out-compete the new entry to the market (using new technology) because older firms have sunk costs such as debt owed to creditors/investors that they must pay-off before investing in buying the newer technology.

All this becomes more complex when the government steps-in and saves domestic firms (Trump's $400B gift to OpenAI/Microsoft), but this is the general process.

Now, just because there is a new entry into the market place using newer and better tech does not mean that all the other firms disappear. It is not one monopoly displacing another.

What it does mean is that there is going to be a loss of interest in investing in the older firms, all else being equal.

Of course, not everything is equal in Trumplandia. The subsidies will come to attempt to save USA-based capital investments in OpenAI/Microsoft and other US tech firms. AI is about more than Chatbots and is increasingly about war and nationalism.

OpenAI and other firms will also fallback onto the recent American past time of complaining that China steals American Intellectual Property whenever Chinese firms best American firms at their own game.

So, what does this all mean for AI and adoption?

1. AI is a productivity tool for capital. The production of AI is also a capitalist form and follows the same development process.
2. It is important to remember that just because it is complex and expensive does not mean that it is a "monopoly" or that it is impossible to have real competition.
3. Cheaper AI means more adoption ("demand"), but that does not mean the majority of the adoption will belong to current American household names like OpenAI.
4. Since application of AI in business processes is where the money is made, efficiency in AI product may over supply the demand for AI in the short term and efficiency gains may outstrip supply long-term.
5. Remember, these public LLMs are really just ads. The money that will eventually pay for these things is productivity at the firm level, advertisement development, and the state military application. Public consumer models are just test beds for development and improvement.

The issue for workers, as stated before, is unchanged. Application of cheap automation of their jobs where efficiency gains result in job loss, not augmentation of work.

As the tech becomes cheaper the threat of displacement of jobs becomes ever present.

But, the investment economics of that implementation, however may have just shifted.

Firms are now paying most for development/application and not perhaps not running the AI system itself. This is a change in the investment regime that the AI firms like OpenAI were selling investors on. And, that is the interesting part.

Do not be side-tracked by the discussion of Jevons paradox, along with complaining that Chinese firms trained their data on OpenAI systems (and somehow "stole" IP from OpenAI) are just ways for these firms to beg for government profit subsidies to compete.

AI has landed in a sort of standard economics of automation of business processes, and one that classical theory describes well. It also means we still have the tools to understand the way this is all unfolding.