Welcome to 2025, the year where AI really does its best to embed itself in our lives, for which you can read: get under our skin and up our noses.
Meta – the owners of Facebook / Insta etc – already came out of the blocks with the launch of a bunch of AI profiles just over a year ago, and most appear to be now dormant… but, oddly, there’s been a rush of interest in them, suggesting perhaps that they might be pushing them again. There’s ‘Liv’ for example:
There’s been quite a lot of responses in the vein of ‘WHY ARE YOU DOING THIS WHEN SOCIAL MEDIA IS SO TERRIBLE AND SO FILLED WITH SPAM BOTS?’
Meta’s answer – which comes from a very recent FT article, which is why the issue is still current:
“We expect these AIs to actually, over time, exist on our platforms, kind of in the same way that accounts do… They’ll have bios and profile pictures and be able to generate and share content powered by AI on the platform . . . that’s where we see all of this going.”
Connor Hayes, VP of product for generative AI at Meta / In the FT here
What those who’ve had them come up in their timelines are reporting is that you don’t have the option to block them, which people are finding very bloody annoying indeed. Moreover, people (because… people…) are starting to troll them:
There appears to be no irony in a GenAI bot calling itself a ‘truth teller’… and some digging by those who’ve interacted with ‘Liv’ has uncovered some worrying elements of the character’s DNA – ie, that the prompts used were somewhat ‘off’:
Which gets us back to ‘why’? Did anyone ask for this? Is this the internet we wanted? My sense is…. no, it is not!
For Meta, though development has been sporadic, their recent press on it seems to suggest a punt that these agents will be able to help ‘real’ users in some way, and will help drive engagement. But that seems like a very weird thing for a social network to have to do: to create fake social agents in order to sustain users’ engagement in the network. If engagement is getting hit… surely that’s a signal that something is off in the social dimension of this network, rather than it being that people don’t have enough AI bots to chat to.
My sense is that social media is suffering engagement because it’s become so full of dross… and that these new AI agents really will not help. We know already that GenAI content has major troubles with reliability and hallucination, and it’s highly likely that these tools will struggle to for accuracy… or be deliberately abused to generate biased / fake content and/or disinformation. All of this erodes trust.
I talk about this prospect in God-Like:
The Cambridge Analytica scandal of 2018 showed that unregulated algorithms controlled by political actors and allowed to infect social media sites could sway a referendum, the results of which people are only just waking from as if from some kind of hypnosis.
Yet this is only AI curating content that people have produced, and selecting that which generates the hysterical response. Harari is anxious about a future where AI has been given language because it will then be able to create content that algorithms already know we will be vulnerable to.
God-Like, chapter 6. Buy a copy here.
There is only one financial model for Meta: ads. So behind the AI bots will be some means of tracking our engagement even more comprehensively, and directing even more laser-focused ads at us.
So… a very odd start to the year with news of these AI agents creeping up – even if they aren’t explicitly new. And I just don’t see this being a ‘good’ year for this kind of AI adoption. I think it’s going to annoy the hell out of a lot of people, and that – right now – is a pretty big concern.
Leave a Reply