You can't be trusted
https://www.youtube.com/shorts/tG6ornTymSQ
Whatever happened to taking responsibility for our own actions? ChatGPT taking responsibility for any misuse on their platform isn't tip toeing some fine ethical line, it's totalitarian information control.
This opens the door for more decisions to be made on our behalf.
I understand the concern about the responsibility of platforms and their potential impact on free speech. However, taking responsibility for any misuse of a platform is a necessary step in promoting a safe and ethical online environment. It is important to find a balance between individual responsibility and platform accountability.
Agree, i'd be more comfortable if there was a discussion, or maybe options put forward by these mammoth companies for we the people to have a say on. OpenAI are now owned by Microsoft in a 11 bn deal...they will dictate information access moving forward until we have another Facebook or TikTok trial, which is rare.
Calling it totalitarian information control is over the top in my opinion. You still have the freedom and choice to use a different LLM service that doesn't do this. Maybe a moot point today, but within shortly there will be rival services that are more hands off in 'censorship' (incl. services that allow you to run locally).
I think these companies are also trying to figure out how Section 230 applies to them. You may be able to take responsibility for your own actions, but the law could still hold them accountable.
I genuinely hope you are right, my worry is it could be the first big ones get to set the trend and the legal bar (or perceived legal bar). People on both sides of the aisle once said 'you can always find another platform' on twitter, then found themselves on the other side (before/after Musk took over)...maybe its different with LLMs?
Let's play out a simplified version of an opposite scenario: There are no limitations to what questions the LLM services will answer, or to what it can say or suggest. My bet is that there would be massive public outcry over LLMs suggesting how to commit terrorism, murder people, slandering, and anything else that isn't generally acceptable in society. Politicians would legislate quickly, and harshly. The publicly available LLMs would end up more muzzled than they are today, permanently.
That's how I think they'd force the legislators to set the legal bar. I don't see why that would be a better outcome. If you think there's an alternative scenario or other assumptions that leads to a better outcome I'm happy to reconsider.
Sorry, I don't think I understand what you're trying to get at here. Competition is in fact always a click away. And with the AI-powered build craze where anyone can build anything in no time (exaggeration!), this is going to be more true than ever. The quality of the options may not be as good as what you're leaving behind, but that's largely downstream of demand for the competition.
Thanks for your long reply, though I think maybe you are conflating illegal content with 'harmful' content, which is part of the problem the policy makers and corporate puppet masters are all too happy to make.
CGP won't tell you lies that Biden has made in his career, but all too happy to list Trumps alphabetically. Maybe someone made a decision to 'prevent harmful dialogue', in this case it most likely didn't happen but you get my idea though.
Afraid competition won't be a click away if Microsoft and their lobbied buddies in office set the terms, which they will. Just like when Larry page and his friends got their windfall with google.