Artificial Intelligence (AI)

X/Twitter will have a way for devs to create AI bots that participate in Community Notes later this month (will be marked that it's an AI). goal is faster notes at the same helpfulness.

we already see users tag @grok to factcheck in real-time, but Verge reader comments are nottttt happy with this announcement.

building farcaster | surfing @ venice, ca | eating @ gjusta / gjelina | “lucky me, lucky mud” 🤠

teddit

we used to rely exclusively on newspapers, radio, or letters for information. so limited, so slow. no way to “fact check” unless you were there on the frontlines.

now we have endless resources to access information, the ability to fact check in minutes if not seconds, and… the majority of the world just doesn’t. this is egregiously bad on tiktok (more on this below). 

i know it is human nature to default to system 1 thinking (lol Daniel Kahnemann reference): fast, intuitive, emotional. i know that we tend to process information in ways that reinforce our group identity. and i know that emotion drives virality.

and so ofc i know that if we keep treating misinformation as an individual problem, this will only get worse.

platforms like groundnews are great, but that’s opt-in so it attracts users who already care about truth (unfortunately that’s magnitudes smaller TAM).

to me it is much more interesting to think of it as a platform problem. with AI, the design space could expand significantly (but obviously comes with its own risks).

two good built-in platform examples:

- twitter community notes on inaccurate tweets cut retweets by at least 50% and increased user-initiated deletions by 80% (but right now it is too slow, should be done within an hour not 12 hours!)

- twitter reminding users to read an article before they retweet actually reduced blind shares by ~33% (a little bit of friction can be a good thing!)

tiktok, however, hasn’t done anything really at all.

some labels for COVID, a politifact and snopes partnership i never saw, and a bullshit STEM feed for “reliable, vetted data” (hint: it was a paid partnership for two science brands to get guaranteed distribution).

the most dangerous feature of tiktok, imo, is its algo-driven echo chambers. it can ensure the content you see is what you want to see, that the comments are what you want to read. 

it pushes users deeper and deeper into self-reinforcing narratives. you can scroll for hours and never be exposed to an alternative perspective on a single topic.

when users are exposed to entirely different information ecosystems, there’s zero foundation for shared discourse. this has got to be the most powerful, least visible force driving division today.

SAD!!!

yes, my general thoughts on it here

https://farcaster.xyz/ted/0x1b91af29