“Please roast my Insta in one paragraph.” With those fateful words, many a poster’s self-esteem has been shattered in the last week. One member of the Dazed team (who shall remain unnamed) has even been brought to the brink of tears by the savage comments of ChatGPT after granting it access to their feed.
In case you’re not caught up: a recent trend sees people offer up their Instagram (or X) profiles at the AI-powered altar, to receive merciless feedback on every facet of their feeds. With no more than a short prompt plus a social media handle, profile link, or screenshot of a profile’s last few posts, tools built to enhance human productivity and build self-confidence are turned into insult machines, coldly pulling apart the content we deem fit for public consumption – and by extension the innermost parts of our personalities.
“Seriously, half your posts look like someone accidentally stumbled into Photoshop and couldn’t find the exit,” ChatGPT tells Dazed. “Your content feels like it’s trapped in a never-ending cycle of forced edginess and obscure references that only a handful of art school dropouts would pretend to understand.” The GPT-4o-powered Monica adds that Dazed’s X profile is: “just a well-curated cry for help”. Duly noted.
Why are we so desperate to be subjected to AI’s brutal analysis, to the extent that websites are overloaded with roast requests? Maybe it’s the uncanny sense that machines know us better than ourselves. Maybe we’re instinctively preparing for the psychological punishment doled out by our future robot overlords. Or maybe it’s just funny.
No one cares what chatgpt said about your feed, we saw you willingly feed the machine yourself and that was enough of a self destruction any roast could be
— hells bells 🍒 (@sputnikvalntine) August 19, 2024
Whatever the reason, the “roast me” trend also makes clear just how comfortable we are with being surveilled by AI in 2024. By the thousands, users of these tools are willingly handing over their data to companies like OpenAI in exchange for a quick laugh and a chance at demolishing their sense of self. And yes, it’s pretty likely that these companies have scraped every last remnant of our public data already. It’s not like they’re waiting for our permission (404 Media recently reported that Nvidia scrapes “a human lifetime” of video content per day to train AI, despite concerns about legal issues). But acknowledging the widespread guzzling of our data by AI companies is a very different thing from actually embracing it and playing an active role in feeding them.
Jak Ritger, an artist and researcher at Do Not Research and New Models, talked to Dazed about the growing phenomenon of “surveillance-as-a-service” last year, in relation to the AI-powered face filters that transform our selfies on apps like TikTok. Then, he summed up the general outlook among users like this: “We are fully aware that we have no way to control how our likeness is used, so why not at least have fun while being surveilled?”
Now, people asking AI to roast their Instagram or X feels like more of the same. We’re fully aware that our data is being scraped en masse to train AI models, and that those models are capable of building a pretty accurate profile of us, to sell us things more effectively – or for even darker ends. So why not at least have fun interacting with them in the meantime?
you all switched pretty quickly from “AI is bad!!” to feeding your whole Twitter account into an AI for a funny quirky roast … amazing.
— Lyro ✨🔧 VTuber (@lyro_vt) July 31, 2024
Not everyone feels that way, of course. Many anti-AI spokespeople – like independent artists whose livelihoods are at risk, or people concerned about AI’s environmental impact – have criticised the trend for helping normalise the technology and its training methods. Others have pointed out that many adopters of the trend have previously opposed AI in its other forms, with one digital artist and X user writing: “AI doesn’t become good and ok to use when it’s funny.”
Will this backlash stop people asking AI to roast them, in the long run? Probably not, and no doubt we’ll find other ways to feed it our data as well. If an all-knowing AI does lord over our lives one day, it probably won’t have emerged from nowhere. More likely it will creep up on us, building on all these little compromises, conveniences, and comedy applications that arise along the way. Then it won’t even need to roll out the slaughterbots – it can just slide into our DMs and unleash painful psychological truths until all humanity is reduced to a quivering wreck.
