@funnymonkey If those kind of accounts fill up medium it will drown writers who work on genuine content (as in: well researched and new).

IMHO that will lead to people leaving the platform eventually.

@funnymonkey The mass following is the same tactic that catfishers have been using on Twitter/FB/Instagram forever, i.e. a half-naked woman with 5 followers who is following 20,000 people, etc.

I guess the difference here is the computer-generated images/videos and better organization across platforms

It’s an utter waste of energy, server space, etc. but I’m not convinced (yet) that we’ll be drowned in this shit

@funnymonkey

Having tracked fakes for yrs on the bird site, everything you said has been my concern about AI from the start.

- People are incredibly easy to dupe w/ fakes. Most don't even do bare minimum to check accts w/ which they engage

- The appearance of popularity is part of the dupe. Big numbers? MUST be popular &, thus, real

- Then there's the fact that most people insist they *can't* be duped--even as they're being duped

My guess: We're in for a really bad ride here.

@VirginiaMurr @funnymonkey I think a lot of it has to do with people's willingness to suspend disbelief when something fake supports a deeply held belief or bias that they would *insist* they don't have.

@peterbutler @funnymonkey We will be bailing as fast as possible even if it doesn't sink all of us.

@funnymonkey

The “search engines” are becoming bullshitting programs primed to generate pattern-matching “content” that seems to say things but isn’t saying saying anything bc they don’t actually understand what anything is, they’re remixing patterns

How do we find one another now?

Human-human connections and referrals matter more than ever

@funnymonkey I came across some so far. I just hope they are not getting boosted by some algo while the real writers disappear.

@funnymonkey I can't wait to see how this stuff gets weaponized in 2024 election cycle.

@funnymonkey Are people saying "ouroboros of shit" yet? Did I just coin that?

@funnymonkey

Will be co-facilitating a #MYFest23 session in July, and I think I'd better cite you properly:

"[T]he bulk of what we're seeing is mediocre, but even if it becomes amazing that doesn't make it good."

@mistakenotmy @funnymonkey

Kyle Hill recently covered this problem affecting #SciComm content creators on #YouTube

https://youtu.be/McM3CfDjGs0

The auto-generated bullshit is being boosted by the recommendation algorithm and drowning out legitimate info

@bornach @funnymonkey well yeah, youtube is becoming a total garbage bin with all the robovoice news, robo science videos and faked live streams.

I fail to understand why google doesn't intervene properly.

I do have to wonder how much time conversational AI's are spending simplifying deliberately overly wordy text generated by other language models. If you search for stuff like "when does the next episode of X?" come out you will get a lot of long-winded answers that look like they've been generated by AI. I believe the idea is to waste precious moments of our finite human lives so that we'll be in front of the ads for longer.

@funnymonkey Nice analogy, "micro plastics". I've referenced your piece in my "Chat AI Hype" https://wrily.foad.me.uk/chat-ai-hype

@VirginiaMurr @funnymonkey Yet another way of saying we need a privacy centric user reputation scoring system for the Fediverse.

@jonahstein @VirginiaMurr @funnymonkey Wouldn't a reputation system also be vulnerable to being gamed by AI spam though? (Like how Amazon reviews are) What would a spam-resistant reputation system look like?

@wh0sthatd0g @VirginiaMurr @funnymonkey Roughly speaking, imagine a digital identity management where users can control what accounts are connected to their identity and what identity attributes they are willing to share with the user table of each site they authenticate with. Essentially, users connect (PIN via privacy preserving method) all of the fediverse accounts they control to their digital identity and then choose what to share with each individual service or Fediverse instance.

Add a digital footprint trust system similar to Google's original Page Rank where an account reputation score is derived by analyzing the accounts you subscribe to/follow and those that follow you.

The next layer would be to allow the users to pin other accounts to their identity.

The trust score could then be derived from the user's social graph and activity across all of their pinned instances. The user could increase their digital trust by also pinning their account to trusted services and/or identifier documents.

@jonahstein Sounds a bit like the old ideas behind signing people's digital keys for public key cryptography, although what you're describing seems to be centralized - which prompts the question of who oversees it. Who becomes the incredibly powerful arbiter of who is genuine and who is not? @wh0sthatd0g @VirginiaMurr @funnymonkey

@wh0sthatd0g @jonahstein @funnymonkey

Oh, absolutely.

The gaming is deep already (with micro-job sites paying anything starting at pennies for likes, RTs, and--yes--reviews by the thousands. AI will make it cheaper and easier.

And, again, it's part of the dupe. Make something appear popular and it "must" be true. A horrible--and very modern--fallacy that works far too well.

@tokyo_0 @wh0sthatd0g @VirginiaMurr @funnymonkey Definitely related because the foundation is users pinning accounts to their identity by signing with digital key(s).

Having said that, the system could be federated instead of centralized as long as the nodes trust each other. Conceptually the same as current federation but with a stricter set of requirements to join. It would require something like a "centralized" directory of trusted node IP address and public keys.

Users would choose an identity guardian to represent one or more digital identities (personas) so they can pin them together. It would be easy to manage/build trust if their personas used the same guardian but that is not a requirement.

Users may also be choose a guardian based on legal/jurisdictional requirements.

Beyond that, users, instances and other services would need the ability to display and or filter messages using the trust score.

@VirginiaMurr @jonahstein @funnymonkey The digital equivalent of "Many people are saying..."

@funnymonkey If there was a choice of identity guardians, though, in a federated-like structure with easy portability, you'd think people might ultimately sort themselves into groups that share ideas about what trust means... similarly to how services like Mastodon allow people to self-sort into communities that agree on what norms they would like to maintain. @jonahstein @wh0sthatd0g @VirginiaMurr

@tokyo_0 @funnymonkey @wh0sthatd0g @VirginiaMurr Exactly.

To be fair, I anticipate users will select guardians for a variety of reason. While shared ideas about trust (or truth) will be important, I think jurisdiction and features will be big factors.

For many, a guardian based in Germany, for example, will offer full GDPR protection as well as Germany's even stricter privacy provisions.

Other users (and service) may chose guardian base in Switzerland or the Cayman Islands while (theoretical) users in China might be required by law to use a Guardian located in China (and register with an official state ID).

Still others may chose a guardian who offer useful features such as sharing a person with family members and/or setup legacy rules in the event of their death.

Long term, users who want a digital identity for commercial purposes may elect to use a guardian controlled by one of their financial institutions and/or a trusted service provider (Think Apple Pay).