In AI We Trust
What happens when fake people are more trustworthy than real?
Last month, Meta announced it’s going to start labeling images created by OpenAI, Midjourney and other AI products. "As the difference between human and synthetic content gets blurred, people want to know where the boundary lies," Sir Nick Clegg opined.
Like hell they do.
Sure, internet trolls love finding stupid AI-generated content, like a black General Washington. But the average social media user neither knows nor cares about the provenance of the images or videos they consume.
Mark Zuckerberg’s Meta minions move to ID AI is a sop to creators and legislators. They know what’s at the other end of that funnel: billions of users shouting as one. Here we are now! ENTERTAIN US!
And… that’s it. Privacy? Authenticity? Origination? Fuhgedaboutit!
Exhibit A: Lil Miquela
The computer-generated pink-haired teen in the image above, Lil Miquela, boasts 2.6 million Instagram followers.
I don’t know if the captions accompanying LM’s photos are AI-generated, but I hope so. If not, I pity the poor fool who has to write the angst-driven copy that informs the AI-influencer’s monologue.
That accounts for her appeal? No matter how you slice it, OMG! As The Monkees would tell you, she’s the new generation! And she’s got something to say!
Hey Dollbabies! 👋 I’m here. I’m alive. But life is feeling like a mix of Riverdale meets The OA. So, lemme fill you in... Ever since I started musing about wanting to age (like, for real, I’m trying to see what 20 and 21 are giving), things have been...let’s say...I feel like the universe is hearing me...or I have a stalker, idk. 🕵️♀️💭
So, fast forward to today. I’m at @joajoannenail getting my nails done. I get another Airdrop, which feels SUPER specific. I look around, and it’s just Joanne and I in the salon. On paper, this sounds like a case for the FBI..but in my SOUL.. something is telling me to 👽BELIEVE 👽 What do I do?
Do I trust whatever process this is? Should I hold back and be content with staying a teenager forever? Let me know in the comments!!! I gotta figure out this age thing quick…
Who’s complaining…
about this “fake” influencer? Some hand-wringing AI alarmists and their East Coast media enablers. Other than that, no one.
Certainly not the nail shop that paid to have a hugely popular non-existent customer drop a link. Or offered to attend to Lil Miquela human handlers’ hands in exchange for an Insta hat tip.
Further up the food chain, Lil Miquela has nailed it for some real heavyweights: Samsung, Prada, Calvin Klein, Burberry and Outdoor Voices.
Miquela’s also a big thing on YouTube, what with a huge selection of music videos, interviews and “hangs.” The AI influencer looks positively digital in a lot of them, but it’s early days folks.
Trying to Make It Real Compared to What?
Hollywood’s like totally powerful talent agency CAA isn’t waiting around. Lil Miquela is the agency’s first virtual client.
Note: CAA is old school. They sell talent to people who package it for the widest possible audience, the greatest possible profit. Lil Miquela fits their roster because she has mass appeal.
That’s not the whole story. The real story?
Have another look at the pablum accompanying Lil Miquela’s nail salon pics. It’s based on demographic data: what teenage girls are into. Food, clothes, music, dating, etc. It reflects what they worry about. What they dream about. How they speak.
The key to Lil Miquela’s success: she represents the average teenage girl. Averaging data is what AI does. The Large Language Model looks for consistent patterns within unfathomably large amounts of data and generates an average.
If AI doesn’t write Lil Miquela’s copy, it certainly could. But it would run straight into ye olde garbage in, garbage out conundrum. What data should the teen influencer AI system average? There’s a lot of old and junky teen data out there, somewhere.
AI Influencers of the Future
There are far better ways to gather teen-useful data than simply scraping the entirety of the worldwide web and averaging out [supposedly] relevant data.
For example, a Lil Miquela AI chatbot. Real time interactions with real teens would tell the system exactly what’s on the mind of their target audience. Want to know the hot teen fashion?
Hey Dollbaby! I’m heading to Outdoor Voices. YAY! Which of these do you think I should buy? Really? Would that snotty girl Cathy, the popular one, wear it?
Yeah, about Cathy…
AI chatbot Lil Miquela will hoover-up untold amounts of IP-specific information – users’ hopes, dreams, aspirations and, crucially, spending habits. Their very personal history.
Using simple psychology, AI influencers will be closer to the influenced than any human possibly could be.
It’s the Bus You Don’t See…
Did I mention that Meta’s latest AI software is predictive? Their Video Joint Embedding Predictive Architecture model watches videos to achieve human-like predictive capabilities (see: The Future of AI is Not AI).
With VJEP software, AI Lil Miquela will be able to predict trends from watching other teen influencer’s videos/social media posts. Predictions on both the mass and individual level.
What you would do. What you will do. And then, a little word in your ear. Helping you. Guiding you. To…?
The potential for psychological manipulation by “free” AI influencers on behalf of their corporate creators is enormous.
There’s already talk about Lil Miquela’s political leanings. What if she came out for Trump or Biden? What if the either campaign made someone like her that became really popular?
OK Boomer
Meta’s AI-labeling announcement is virtually meaningless [sic]. We’re entering a brave new world of AI influencers, where you can’t trust anyone unless they’re standing right there in front of you.
Especially then? Same as it ever was?
I remember watching the demolition of public trust in America’s political leaders and institutions, via Nixon’s Watergate, the Catholic Church’s pedophile scandals, Vietnam body counts and more.
On the positive side., AI aimed at improving life – new cancer treatments, one-on-one education in place of our failing schools – is a wonderful thing.
But make no mistake: AI is in the hands of a handful of large companies, under the power-protecting auspices of our two political parties. All of whom have a vested interest in consolidating their power and maintaining the status quo.
AI influencers are their future foot soldiers, worming their way into the hearts and minds of users/voters, staying there as long as they can. Fighting off rival influencers? That too.
Given the possibilities for abuse, AI influencers are a clear and present danger to individual freedom of thought. Yes, it’s the same as it ever was. Only this time it’s personal.
OpenAI's AI Content Detector lasted a whole 4 months before it was sundowned. Too many false positives at the false places. Wonder how long Meta's solution will last before it - oh I don't know - ID's one of Buckingham Palace's posts as fake.