Good article, which I think could be summarized as "let the competition between many different alternatives and free markets sort things out", which I very much agree with.

As for the future of AI, I have read that every hype bubble is based on one or more lies. The lie behind this 3rd iteration of the AI hype bubble (60's mainframes, 80's PC neural nets, and today) is that adding more billions of parameters to the LLMs will eventually eliminate hallucinations and deliver 100% accuracy sometime in the future. All the promises made by the AI-bro's about taking over the world are only possible if LLMs can achieve 100% accuracy... which they cannot do.

On the Fundamental Limits of LLMs at Scale (research) https://arxiv.org/pdf/2511.12869

An interesting question from the research paper: "Having established that LLM failures stem from five fundamental theoretical limits, namely hallucination, context compression, reasoning degradation, retrieval fragility, and multimodal misalignment, a natural question emerges—why do current benchmarks suggest continuous progress despite these intrinsic ceilings?"

The false suggestion of "continuous progress despite the intrinsic ceilings" is the 100% accuracy lie of this 3rd AI hype bubble.

At the moment AI costs more than the human it replaces.