Hacker News Reader

Comment by bad_haircut72 | Hacker News Reader

People think claude costs $200/month, it actually costs $200/month and all your labor to train their models, improve their product, and all your best ideas and work that they will feed back into their company. Its very expensive

View on HN

You can opt out of model training with a checkbox, though I doubt most companies respect this completely. At the very least, I imagine they find loopholes to exploit.

But on the other hand I'm giving them my best and worst ideas and the value I get in return outweighs what I'm contributing to them. I also don't have the problems of initial HW cost and maintenance.

> You can opt out of model training with a checkbox

Do you trust that a company who can't even do basic IT ops monitoring are capable of respecting a "no training checkbox"?

"Whoops my bad, you are right, your data may have been in the training pool. They are now forever in weights that'll be reused to train every future model. Too bad you can't do anything about it."

Why do you care? It's their legal/compliance problem, not yours, if it turns out they didn't (though I don't understand the urge to deny them this in the first place, it's not like it hurts you or deprives you of anything).

Great, then they will PII-scrub my sessions before feeding them into the training pool :-)

Yer I also use service with zero data retention promises but it is hard to really trust them for me... They trained on basically all of the existing material, legal or illegal already and are starved for true human input. They are also in a heavy arms race were basically only a handful of companies will survive.

Yes. Training is necessary to make an LLM. They don't grow on trees, you know.

Their own TOS will still include your "anonymized" data in their "research", with no explanation as to what that means. You need an enterprise account to get real zero data retention.

Yes, and? Why would I care? Why would I begrudge them? I'll get the improved product, too, that will benefit from not just my data, but everyone else's to.

Don't get the greed and the Dog in the Manger mentality so ever-present around AI detractors.

Did the mathematicians who got scooped on Navier-stokes benefit from the improved product or did their lifes work get stolen?

Reply on news.ycombinator.com