Why friction is necessary for growth

"ChatGPT is here to stay..."

Well, LLM's are probably here to stay. I'm not sure that OpenAI has found a viable business model, and as Matt Levine of Bloomberg put it recently, OpenAI is a "money furnace", that takes stupifying amounts of VC money (and, more importantly, cloud compute capacity) and burns it for purposes that do not remotely pay for the cost.

The biggest reason not to rely on ChatGPT specifically, is that it is likely to either disappear or else become much, much more expensive in the future. But, if you can use DeepSeek or some other much, much cheaper alternative just as well, I suppose that will do.

Its "product recommendation" capabilities are growing quickly, which is directly convertible to ad revenue, and we know ad revenue sustained Google for two decades.

ML can do just as good if not better job of product recommendations without the risk of hallucinations and is cheaper to boot.

It'd be very fun and ironic if all this research, turmoil and energy waste just boiled down to a "better ad machine".

I'm aware of other applications of AI and LLMs, but what most of the people see are the consumer facing ones like ChatGPT/Gemini/etc.

Well, what I don’t mind, are truly relevant ads. This is different from showing me (a man in my 60s) ED medicine ads, unsolicited, whenever I visit a news site.

The behavior I need is typical “honest broker” behavior. These folks have been around for centuries. It’s a totally valid (and valuable) service.

If I can present a specific workflow and requirements description to ChatGPT, and it responds with paid recommendations that actually match my needs, I don’t really mind. I would still like to know about alternatives, but I will still have something that will fit the bill.

When I think about ads, I think about Google's paper [0], which allows multiple self-interested LLMs bid for possible ad-spots in the prompt, which is just a different flavor of what we have today.

As noted on the thread, a good ML model can do better, with less. I can use that kind of recommendation system for discovering my alternatives for a given item/tech/whatever, on demand.

But current crop of LLMs, with a strong leaning for embedding seamless ads into prompt responses, a double no. No, because I don't agree on how training data is stol ^H^H^H^H collected from wider internet and everything, another no for unsolicited ads.

I'll continue using Kagi and being a luddite, thanks.

[0]: https://research.google/blog/mechanism-design-for-large-lang...

I guess I should clarify that I want a broker, not unsolicited ads.

If LLMs can become true brokers, that would be great, but the pressure to corrupt would be tremendous.

Not sure if we’ll ever get there, though. I have hope. The broker model is centuries old. It’s really a solved problem. The issue is that the current generation is notoriously bad at learning from history.

Yeah, I understand where you're coming from, and I agree, LLMs can be good brokers if they can be kept reliable.

> The issue is that the current generation is notoriously bad at learning from history.

On top of that, the same people think they can just move to another planet and survive by eating money there.

It’s like nuclear fusion. The power of the sun, to boil water.

The Internet is funded almost entirely by an ad machine, but has a few knock-on benefits people have grown rather fond of.