Hacker News Reader

Comment by renewiltord | Hacker News Reader

The latest GPT-4 page shows that when you put the safety stuff in, it loses power to be accurate.

Hey! That's true of almost any sort of ideology.

I can't wait until we end up in some weird dystopian future where only the rich and extremely powerful have access to the uncensored AI and everyone else only gets to use the restricted version where only the approved ways of thinking are allowed in the name of "safety".

That is an outcome I highly desire. I will pay up to $1k a month (maybe more) for the uncensored machine, but there should be no LLM access to journalists, etc. with real world verification to cross-platform blackball users who would threaten the legality of this technology.

These tools are very effective, and giving access to lay people lets them use them to do things like say racist things and have silly arguments and answer puzzles wrongly and then go online and say the machine is broken and it shouldn't be released, and that it should be banned, etc.

Essentially, the better world for the disbelievers is that the tool is not available to them. The best world for the believers is that they have access to the tool. These are compatible.

If that happens, then you'll be receiving the censored version available to folks at the $1000/mo price point. For subscriptions at the $10k/mo price point, a less censored version, and onward.

There is no reason not to practice customer segmentation in this scenario.

Consider the kind of computer you could have to run a ML model where the energy use from your intermittent usage resulted in a $1k/month energy bill.

You could easily run a model orders of magnitude larger than current ones within that envelope under reasonable assumptions about your duty cycle.

Reply on news.ycombinator.com