Now Elsewhere

Elsewhere

[Now Elsewhere @cyberneticpond]

@waffelhauss @CabbyDM As for the original image, the source linking it to the brutality of the Belgian Congo is historian Paul Van Damme's 2020 book 'Wit-zwart in zwart-wit'.

[Now Elsewhere @cyberneticpond]

@waffelhauss @CabbyDM I can understand not finding OP's photo immediately - but are you really saying you didn't find a single thing corroborating the 1958 Worlds Fair 'Human Zoos'?

[Now Elsewhere @cyberneticpond]

@djwhatley @lastpositivist The original claim here is that specific problems occur when a small segment of a population has all the power/resources to run things. Your counterpoint is a scenario where the entire population (2 people) decide cooperatively how to run things.

[Now Elsewhere @cyberneticpond]

@SickoBoingo @goof @_alialkhatib @blprnt Not sure if we can say the AI didn't have access to any info about the support-recipients. Video says their initial message was sent to ChatGPT, which could include disclosure of some very personal details (eg. trauma history). This could be ingested back into the LLM.

[Now Elsewhere @cyberneticpond]

@SickoBoingo @goof @_alialkhatib @blprnt Yep, my point above is that there's an inconsistency between the project's narrative about why people eventually didn't like the assistive responses (they felt disappointed when they found out about the AI) and the idea that they were fully informed prior to its use on them.

[Now Elsewhere @cyberneticpond]

@_alialkhatib @goof @blprnt Handling that responsibly seems like it requires disenchanting support providers on the harmful potentials of LLMs from the get-go. Doing that effectively for the support providers and not for the support receivers seems pretty deceptive.

[Now Elsewhere @cyberneticpond]

@_alialkhatib @goof @blprnt I feel like the P2P ethics deserve a thread of their own. If KoKo's helping me help people, but ChatGPT suggests a message that's harmful (which I don't recognise), that message could go on to inform how I reach out to people in hard times in future.

[Now Elsewhere @cyberneticpond]

@goof @_alialkhatib @blprnt Stranger still, if they received informed consent from participants, why are they arguing to Business Insider that they're exempt from informed consent law? https://t.co/GjcaPm4uH6

[Now Elsewhere @cyberneticpond]

@goof @_alialkhatib @blprnt The satisfaction scores dropped when the participants learned messages were composed by AI. If they were fully informed before, they'd already know about the AI copilot, so why the score drop?

[Jason Goldman @goldman]

I guess they were banned for revealing the location of where all the Twitter users are going.

[Now Elsewhere @cyberneticpond]

@PatronSaint2 @Kylez1967 @lvgcommunities @UnitingChurchAu Hey @PatronSaint2 - just in case you were still looking for examples, here are 101 of them (the site just launched today). A national charter would mean every Australian could expect these kinds of protections:
https://t.co/8Yu6w5l72a

[Now Elsewhere @cyberneticpond]

@JonWolfMueller @Real_Boaz @DorsaAmir @Asher_Wolf @dblplusgoodful Trade-off is that it's easier for me to find related work in semschol, and their "highly influenced" score is really useful. My general workflow, for now, is elicit -> semschol (generate list to review) -> scite (qualify + contextualize list items).

[Now Elsewhere @cyberneticpond]

@JonWolfMueller @Real_Boaz @DorsaAmir @Asher_Wolf @dblplusgoodful @JonWolfMueller both are great, semschol is free while https://t.co/HMMZcG5h1C requires a monthly fee (reduced for academics). Key difference for me is that scite includes text extracts of the references, which helps me grok key areas of contribution/controversy before I read.

[Now Elsewhere @cyberneticpond]

@troutish @peacenicsta I don't have the capacity/skills to help on the socials or content side. But if someone's happy to brief me in on the rough shape of the campaign, I could definitely help towards getting a site set up.

[Now Elsewhere @cyberneticpond]

@Real_Boaz @DorsaAmir @Asher_Wolf @dblplusgoodful Also, the easy-to-use "export to .bib" function, means you can find the right question in elicit, export papers to .bib, map citation patterns using something like https://t.co/BRdGUtZpG9, and continue to add new nodes using https://t.co/y9PE1Ul6n0 or https://t.co/HMMZcG5h1C

[Now Elsewhere @cyberneticpond]

@Real_Boaz @DorsaAmir @Asher_Wolf @dblplusgoodful This is fantastic - especially love that the search uses a plain text query, so you can hook it up to an Alfred custom search to make accessing it even easier.

[Now Elsewhere @cyberneticpond]

@troutish @Graham_Caswell @ruxton I’d need to dig into Twitter’s API Terms of Service to see if they’d be cool with this, but the tech seems pretty do-able. Would be happy to contribute!

[Victor Galaz has moved to Blsky @vgalaz]

”The future is not yet written. We are writing it now.”