iiinova — where your digital self keeps evolving

gfodor @gfodor.id · 28/11/2024

It might be more efficient to go with a whitelisting approach on this hellsite.

160

gfodor @gfodor.id · 28/11/2024

Logged in here to just say this is retarded

170

gfodor @gfodor.id · 21/11/2024

Ah looks like you’re about to get mobbed here, good times

000

Not true, what Elon stopped doing which was the bad thing was the three strikes ratchet to being permabanned for violating trust and safety (aka ministry of truth) rules like saying “Covid vaccines aren’t safe”

000

(Sorry for all the q’s, if there is a faq or something pls link, or if busy no need to respond)

Ok thanks, so users can hide posts based on voting on stuff now? Is that where the “intolerant” thing I was seeing screenshots comes from? Are those posts nuked or can you still read them?

210

gfodor @gfodor.id · 19/11/2024

Can someone explain the hugbox rules here? Ie is this like twitter a few years ago where there are verboten things and I can get disappeared if citizens report me to the secret police, or is it more nuanced as this system is supposed to be one where users choose their filters?

290

He will be Community Noted

gfodor @gfodor.id · 19/11/2024

That’s fine Like I said, there is less censorship than there was because since people are not being permabanned for their opinions

gfodor @gfodor.id · 19/11/2024

Elon doesn’t permaban people for “misinformation”. QED

120

gfodor @gfodor.id · 18/11/2024

He has but he doesn’t permanently suspend people for “misinformation” the way twitter used to. That’s the main change

gfodor @gfodor.id · 18/11/2024

That’s more true now than when the exodus started, which was back when Elon unblocked a bunch of banned posters.

gfodor @gfodor.id · 18/11/2024

Bluesky is ultimately the path to the most free and unfettered internet speech - it is fantastic and fantastically ironic that its inflection point seems like it will be driven by a cohort largely angry about a reduction of censorship on twitter

8622

gfodor @gfodor.id · 14/11/2024

Great. Looks like they shipped it all since I last hung out here. Awesome

1100

gfodor @gfodor.id · 14/11/2024

I know but did they ship the bits for the reference servers?

000

gfodor @gfodor.id · 14/11/2024

Is this thing decentralized yet?

4200

gfodor @gfodor.id · 06/11/2024

Logged in to say this. It’ll be ok.

040

gfodor @gfodor.id · 02/07/2023

*crawls in through the front door, battered and beaten*

4110

gfodor @gfodor.id · 29/05/2023

I think this is true in the case of AI up until the Turing test passes universally at which point we have a deep epistemological crisis about personhood. But until then, yes :)

040

gfodor @gfodor.id · 29/05/2023

*Provably not probably

000

gfodor @gfodor.id · 29/05/2023

The idea, at least to me, is I would want labelers that probably label things in alignment with people I endorse. So it’s still people-oriented, the AI is just multiplying the impact of that person’s taste, principles, etc. it’s not an either/or given we can fine tune models.

2110

Reposted by gfodor

antpb @antpb.xyz · 14/05/2023

Made a simple moderation endpoint that uses gpt-3 turbo to filter out negative vibes from my skyline and it kinda works. Don’t trust it for an everyday prod experience but the idea of putting my own logic in the middle of calling my timeline is very interesting. Even if just string matching.

2114

It’s insane that our ability to do custom filtering of tweets is less capable than email spam filtering tools of the 1990s. All because Twitter isn’t a protocol.

0371

Yep - the high order bit here imo has always been the lack of a protocol that would let us build social User-Agents like the browser. If that existed for Twitter you’d have seen rudimentary versions of this a decade ago just doing simple custom naive Bayes filtering. It’s insane we can’t.

I don’t think AI can solve all content moderation, that would be ridiculous. I think AI can *significantly* impact inflow of harassment and user agency over it of the kind Knox mentions here which are just streams of harassers all saying the same thing in slightly different ways.

000

I would be really surprised if a fine tuned LLaMA model right now would not move the needle on highly replicated, pattern matched harassment Knox is getting of just filtering tweets accusing her of murder. The prompt could evolve easily as new attacks form. This isn’t some insane classifier problem.

The key thing to understand is I’m not talking about centralized moderation. I’m talking about a personal AI filter that you train and fine tune for your own personal blocking regime of extreme harassment cases. Seems very doable now other than the lack of a good Twitter API.

You already have a web browser that protects you from all kinds of attacks. This is just a natural extension to that, which would leverage the recent breakthroughs in LLMs being able to understand text. The error rate on something like “is this message blaming me for murder?” would be solid.

It seems obvious to me that a simple local LLM literally today could shield you from nearly all incoming death threats of the kind mentioned here. It’s not a silver bullet but would basically eliminate the case here - the problem is platforms like Twitter have no hooks to inject into.

I’m not sure how it decreases agency to add a AI intermediary that you have full control over. It would need to be open source and probably also easy to perform fine tuning on. People would use it wildly differently. Some would simply tell it to block death threats.

Build the social User-Agent - an AI that you control which sits between you and the wider social internet, pro-actively filtering and rewriting it for you - and you’ll not just become wealthy but you will cut down an immense amount of completely unnecessarily human suffering.

2180

Sridhar Ramesh @radishharmers.bsky.social · 11/05/2023

AI is the solution here. It needs to become possible for an AI to proxy all of your incoming tweets and you can delegate it to filter proactively by analyzing them. The fact we have lived without this for so long is insane. It’s because of the closed protocols, bluesky fixed it.

210

Reposted by gfodor

LeVar Burton put himself through hell hosting Reading Rainbow. The strain from reading all those books badly damaged his eyes, confining him to a prosthetic visor.

413717

gfodor @gfodor.id · 11/05/2023

bringing your A game to bluesky, strong endorse

I slerp up content like this

I too would like this feature, good idea

060

The only consistent topic on here seems to be WWND (what would Nazis do) and intellectual takes or irreverent shitposting about Nazis is not recommended

140