> Nevermind that under the stewardship of the central banks, the financial system has lurched from one crisis to the next for decades on end, every one an order of magnitude worse than the prior.

Nice opening because it lets you know right up front that the author has no touch with reality.

Following this line, are we going to see an many ChatGPT charlatans as we see them in Bitcoin ?

That number is climbing fast already.

So… yes.

It's a lot harder to build a pyramid scheme because running the models requires money instead of printing money.

I doubt it will look much like the Blockchain Hypetrain.

It will still be difficult to connect the ROI of these chatbots to the bottom line. For most applications, they are an expensive UX upgrade. I suspect a few will do well through direct payment, for users who want a productivity upgrade.

It requires minimal money to put a shim between a language model run by someone else and a customer with an api key.

How many shims do you use and pay for when the primary application already has this feature? I personally can't think of any I use for anything today...

Also against OpenAI ToS to create BYOK services

I don’t think most users of this site are marks for this sort of scam. That doesn’t mean it doesn’t exist. Have you not seen BYOK apps and services? Several have been posted here over the last few weeks.

ToS doesn’t as matter as much to those that are already aiming to relieve the gullible of their money.

I don't disagree that there will be some, but back to the original point, it will not be anything like cryptos, where one could copy and paste the code for a new token and then build a marketing hype to scam the gullible. It's going to be a lot harder to build something people will pay for, because you don't expect to make millions by putting money in as the gullible person being scammed.

It won’t be harder to make something people will pay for. Just harder to get them to start making outright deposits.

But that’s crude. A shim could capture credit card information, bank information, or other PII of value.

Why take the money when you can take the whole person?

I understand the supply side and agree with you there. The gap is on the demand side, it will be nothing like it was in cryptos, for shims, when the main applications people already use have this embedded.

What I actually expect is that people will only want one or a few of these to interact with. Recall how we had to learn Google Fu to get better search results? A person has to learn how to interact with the chatbot to get the output they are after. No one wants to learn 100s of these, for each application. I fully expect a few dominate players to capture the vast majority of the market. No sense in paying for a wrapper around them.