Meta says its new speech-generating AI model is too dangerous for public release | Hacker News Reader

Meta says its new speech-generating AI model is too dangerous for public release

theverge.com

22 pointsactiviation13 comments

This seems like a classic case of criti-hype,

https://sts-news.medium.com/youre-doing-it-wrong-notes-on-cr...

Facebook is a master of criti-hype, in that it has convinced many people that it holds the secret dangerous knobs that control human behaviour. Obviously that's great when you are trying to sell yourself to advertisers. "Our Algorithms are so dangerous they should be regulated, but we can let you harness their power a little, at a cost."

This is the exact same feeling I get reading this. Extraordinary claims require extraordinary evidence and I don't see anything here that makes me believe they've got something truly "earth shattering".

They're trying to lobby and get regulations put into place that others won't be easily able to conform to so that they get some market lock-in.

They're big enough to be able to conform to the 'rules' but newcomers won't be able to without a lot of staff and/or money.

The ps2 was "too powerful to export to Iraq because they could use it to control missiles"

In reality, it was something about the encryption on the memory cards lol

PS2s were a controlled export because Iraq was under sanctions for Nuclear Weapons development, and creating a Beowulf cluster using PS2s was an efficient way to build HPC infrastructure when companies like Xillinx et al were barred from directly selling to Iraq.

Example PS2 cluster btw - https://web.archive.org/web/20041120084657/http://arrakis.nc...

You can’t play with my toys! There too awesome for you and you’re not smart enough to play with them.

we can't serve adults steak because babies can't chew it.

I see technology as the only way to destabilise these large monolithic corporations. I'm hoping the rate of progress happens so fast that investors have no clue which horse to bet their money on

Interestingly, I've long held the suspicion that AI will bring about a period where one person can create a company that would have taken 10 people in 2022 to build.

This means we should see far more innovation and new products coming along in 2-3 years.

For example, it wouldn't be surprise if 10,000 new Instagram competitor ideas spring up overnight and maybe 1 or 2 of them actually surpass Instagram.

They probably will, but at the same time the end user will be flooded with an abundance of (quality!) choice without any indication which one they will choose. Here, marketing and advertising will become crucial and we'd be back to square one of relying on these large monoliths.

Related discussion w.r.t the Indie game scene: https://news.ycombinator.com/item?id=36387030#36387897

Do you presume to criticize the Great Oz?

You ungrateful creatures!

Think yourselves lucky that I'm giving you an audience tomorrow, instead of twenty years from now!

Oh! TheGreat Oz has spoken!

Oh! Pay no attention to that man behind the curtain. The Great, Powerful -- has spoken --

It's about as dangerous as the hook in the water dangling in front of the horse...

It's not unrealistic, the progress in this area is fast and results already quite good.

Reply on news.ycombinator.com