3 ms·
Honest answer*: I think building super intelligence for the company that owns and will deploy the super intelligence in service of tech's original sin (the alg
by pj_mukh 1y ago
Honest answer*:
I think building super intelligence for the company that owns and will deploy the super intelligence in service of tech's original sin (the algorithmic feed) is a 100x worse than whatever OpenAI is doing, save maybe OpenAI's defense contract, which I have no details about.
Meta will try to buoy this by open-sourcing it, which, good for them, but I don't think it's enough. If Meta wants to save itself, it should re-align its business model away from the feeds.
In that way, as a missionary chasing super intelligence, I'd prefer OpenAI.
*because I don't have an emotional connection to OpenAI's changing corporate structure away from being a non-profit:
- eli_gottlieb 1y agoIf you have "superintelligence" and it's used to fine-tune a corporate product that preexisted it, you don't have superintelligence.
- svara 1y ago> I think building super intelligence for the company that owns and will deploy the super intelligence in service of tech's original sin (the algorithmic feed) is a 100x worse than whatever OpenAI is doing, OpenAI announced in April they'd build a social network. I think at this point it barely matters who does it, the ways in which you can make huge amounts of money from this are limited and all the major players after going to make a dash for it.
- pj_mukh 1y agoLike I told another commenter, "I am judging the two companies for what they are, not what they could be." I'm sure Sam Altman wants OpenAI to do everything, but I'm betting most of the projects will die on the vine. Social networks especially, and no one's better than Meta at manipulating feeds to juice their social networks.
- epolanski 1y ago> In that way, as a missionary chasing super intelligence, I'd prefer OpenAI. There ain't no missionary, they all doing it for the money and will apply it to anything that will turn dollars.
- makeitdouble 1y agoAs a thought exercise, OpenAI can partner to apply the technology to: - online gambling - kids gambling - algorithmic advertising Are these any better ? All of these are of course money wells and a logical move for a for-profit IMHO. And they can of course also integrate into a Meta competitor's algorithmic feeds as well, putting them at the same level as Meta in that regard. All in all, I'm not seeing them having any moral high ground, even purely hypotheticaly.
- pj_mukh 1y agoWait if an online gambling company uses OpenAI API then hosts it all on AWS, somehow OpenAI is more morally culpable than AWS? Why?
- makeitdouble 1y agoI saw the discussion as whether OpenAI is on a better moral ground than Meta, so this was my angle. On where the moral burden lies in your example, I'd argue we should follow the money and see what has the most impact on that online gambling company's bottom line. Inherently that could have the most impact on what happens when that company succeeds: if those become OpenAI's biggest clients, it wouldn't be surprising that they put more and more weight in being well suited for online gambling companies. Does AWS get specially impacted by hosting online gambling services ? I honestly don't expect them to, not more than community sites or concert ticket sellers.
- pj_mukh 1y agoThere is no world in which online gambling beats other back-office automation in pure revenue terms. I'm comfortable saying that OpenAI would probably have to spend more money policing to make sure their API's aren't used by gambling companies than they'd make off of them. Either way, these are all imagined horrors, so it is difficult to judge. I am judging the two companies for what they are, not what they could be. And as it is, there is no more damaging technology than Meta's various algorithmic feeds.