Unless superintelligence is developed under a global consensus, the risks will be shared by all, but the upside won't. This is why I signed the superintelligence statement. The superintelligence statement is the following: > We call for a prohibition on the development of superintelligence, not lifted before there is >...
Moloch, as articulated by Scott Alexander, represents the coordination problems that lead to outcomes that leave everyone worse off. While prediction markets explore what people think will happen, YouCongress aims to aggregate beliefs and desires regarding ideal outcomes. This open-source platform proposes a novel coordination mechanism, making use of public...
Unless superintelligence is developed under a global consensus, the risks will be shared by all, but the upside won't. This is why I signed the superintelligence statement.
The superintelligence statement is the following:
... (read 265 more words →)We call for a prohibition on the development of superintelligence, not lifted before there is
1. broad scientific consensus that it will be done safely and controllably, and
2. strong public buy-in.
Context: Innovative AI tools may bring unprecedented health and prosperity. However, alongside tools, many leading AI companies have the stated goal of building superintelligence in the coming decade that can significantly outperform all humans on essentially all cognitive tasks. This has raised concerns, ranging from human economic obsolescence and disempowerment, losses of
Users can register now with email/password.
Being able to vote for 'liking' or 'disliking' the quality of a Digital Twin response. Enough dislikes might trigger a model update and regeneration of the response.
@Nathan Helm-Burger As a first step, moderators can now regenerate AI responses. I have also added you as a moderator, so feel free to use this feature.
Once we have more users (and have more time/resources to do it properly), we may introduce a rating system as you suggest. For now, any user can report a comment, and a moderator can regenerate it if deemed reasonable. Is it sensible?
Thanks!
Thank you for sharing your thoughtful ideas and vision for YouCongress, Nathan!
Regarding auto-updates of current legislation, I've been considering building something akin to subreddits (let's call them "halls" here) on top of our current topics. Some would be automated by us (e.g. AI polls, climate polls, bills in US Congress), while others would be managed by users. The latter would allow the owner(s) of the hall to add new or existing polls manually or via the API. For example, someone could create a hall about their local region or topic of interest. This user-driven approach could offer a wider coverage.
The downside here is whether this may lead to duplicated polls.
What are your thoughts on this?
Does it? Sorry about that. The minimum set of permissions of twitter login seems to be "Read". I guess we'll need to give more priority to sign up with email and password.
Good catch Mako! There was an error in the priority list diagram. Yet, as I understand a priority list, it should choose the first active vote – including abstention. So if the first one in the priority list doesn't vote but the second votes abstention, the voter would abstain. Does it make sense? I've updated the diagram. Thanks for pointing out the error!
Done! The website and LessWrong post now use "public opinion poll" instead of "voting". Thanks again for the suggestion, @Nathan Helm-Burger!
Thanks for the feedback Nathan! We'll definitely consider framing it as public opinion polls—you're right that it could be more impactful and less confusing.
Regarding Twitter/X, we plan to add normal user/email login. We initially used Twitter login to simplify things and allow public figures to manage their content. But we'll definitely expand login options.
And allowing users to create digital twins from LessWrong posts and uploaded documents is a fantastic idea! We'll definitely explore that, thanks for the suggestion!
Moloch, as articulated by Scott Alexander, represents the coordination problems that lead to outcomes that leave everyone worse off. While prediction markets explore what people think will happen, YouCongress aims to aggregate beliefs and desires regarding ideal outcomes. This open-source platform proposes a novel coordination mechanism, making use of public opinion polls with delegation and AI to enable large-scale participation and pave the way for more rational and impactful policy decisions.
Numerous proposals have been put forward to improve democracy through concepts like liquid democracy—a hybrid of direct and representative democracy—and augmented democracy. Yet, the goal of YouCongress is not to make binding referendums but to assist in decision-making, whether these decisions are made... (read 1636 more words →)
That's a fair point, I should have been more explicit.
My post is examining the risk conditional on the labs solving alignment well enough to keep the ASI under their control.
So yes, I agree that the primary risk is uncontrolled alignment failure.
I'm just pointing out that even if labs develop aligned superintelligence, we face a second risk: a global, perpetual monopoly on power.