Re: SocialWeb CG Feb 7, 2025 call - possible agenda item

Sounds interesting, I'm thinking Task Force that starts with
informational/use-case FEPs to gather more buyin before doing anything
normative?

I would mention anecdotally that some amount of ML is already used in prod
by some bigger mastodon instances for the purposes of "pre-moderation",
i.e. flagging inauthentic behavior and/or spam, largely to reduce workload
on the moderation team and save bandwidth/storage/garbage-collection
surface.  I know i'm a broken record on this, but some form of uniform
metadata that can track a given Content URI across pre-moderation flagging,
moderation actions, etc would be great... See also:

https://fosdem.org/2025/schedule/event/fosdem-2025-5813-unlocking-transparency-in-platforms-content-moderation-activities-introducing-dsatdb-a-python-package-for-analyzing-the-digital-services-act-transparency-database/

I think the T&S task force is already very busy with a long to-do list, and
would probably appreciate my/our not adding coordination costs, but one
thing I think is worth thinking about early on is this metadata/tracking
question.  Happy to work on that specific little thing, as I think it could
be as simple as a sha-256 hash of the JCS-canonicalized "original activity"
before an `id` being added by the server that anchors all future metadata
(including revisions, new `id`s across migrations, etc).  Also helpful for
client/server cuz the client could send that hash before a server
broadcasts...

party on,
__bumblefudge


On Fri, Feb 7, 2025 at 12:58 PM Aaron Gray <aaronngray@gmail.com> wrote:

> Hi Group,
>
> I would like to briefly bring up in todays meeting an agenda item
> concerning the idea of possibly creating an AI, robots, and Agents CG
> possibly as a Task Force or longer lived "Agentic Social Group" to use
> Melvin's term. This was triggered by Bob's recent post on ActivityPub and
> AI.
>
> https://lists.w3.org/Archives/Public/public-swicg/2025Feb/0000.html
>
> This would be to look at the possible benefits and also the impacts of AI
> on ActivityPub and ActivityStreams ecosystems and into mitigation and usage
> in terms of moderation and also the wider implications socially and
> technically.
>
> I know there are cross cutting concerns with moderation and The
> ActivityPub Trust and Safety Taskforce. I do feel this is a separate
> parallel, much wider area that needs to be looked at in its own merit as
> well as in connection with the moderation, security and safety issues.
>
> We need to establish the knowledge, insight, and people in order to guide
> SWICG, ActivityPub and Activity Streams through these technological
> developments we find ourselves with, including possible effects on the
> protocols from issues like such as AI fuzzing attempts, to social
> engineering by AI's, through to how we deal with and approach and
> facilitate Avatars and Agents.
>
> I have to say unfortunately at this point I would not want or be able to
> head this group but really feel it needs to be put in place. This is due to
> my absence of experience in leading such a group and also due to other
> duties and current projects, such as studying AI. But, given my existing
> knowledge and insights into the impact of AI I would love to be a
> contributributing member and to see this happen.
> Kind regards,
>
> Aaron
> --
> Aaron Gray - @AaronNGray@fosstodon.org | @aaronngray@threads.net |
> @AaronNGray@Twitter.com
>
> Independent Open Source Software Engineer, Computer Language Researcher
> and Designer, Amateur Type Theorist, Amateur Computer Scientist,
> Environmentalist and Climate Science Researcher and Disseminator.
>
>