Advertisement
Home LIFE & CULTURE LIFE & CULTURE Latest

ChatGPT Correctly Guessed My Fragile Ego Can’t Handle An Actual Roast

But why are we submitting our profiles to our AI overlords for scrutiny anyway?
ChatGPT roasts writer Ruby Feneley's Instagram presence.Instagram: @rubyfeneley, OpenAI

Heard of the ChatGPT roast? It seems this week, the chronically online amongst us have gone from getting ChatGPT to write uni essays and craft breakup texts to asking it to scold us on social media. “Chat GPT Roasting Instagram Feed” has 7.3 million views on TikTok. And, everyone from your high school boyfriend to media accounts like Dazed are sharing ChatGPT’s reviews of their feeds. 

Advertisement

Never one to miss an opportunity to bask in the spotlight (even if it’s the exacting gaze of a faceless AI coming to steal my data), I couldn’t resist submitting myself to ChatGPT’s judgement and quizzing it on how its new assignment was going. 

Before we get into my experience – and interview–  with ChatGPT, let’s unpack what’s happening with everyone’s new sassy internet bestie: the ChatGPT roast master. 

What Is The ChatGPT Roast? 

Was the ChatGPT Roast what the creators of AI were hoping to achieve when they created the groundbreaking, apocalyptic tool? Based on our knowledge of machiavellian tech supergeniuses, we’d have to assume not. But whether intended, ChatGPT is now being used off-label to rinse us. 

The ChatGPT roast involves submitting your social media profile for analysis. You can send screenshots of your Instagram to the AI, and it will break down your profile and describe your character in the process. I asked ChatGPT how it would describe its approach to roasting, and it responded: “Roasting is a delicate art! It’s like giving someone a stylish slap with a velvet glove – you want them to feel the burn but also secretly appreciate it.” The bot said that most of its voluntary victims had taken the criticisms in stride “or at least pretended to.” 

Advertisement

How Do You Ask ChatGPT To Roast Your Instagram? 

I submitted myself for ChatGPT roasting in a rudimentary way, sending some screenshots of my Instagram grid along with a request. However, if you have the inclination you can head to the InstaGuru section of ChatGPT designed to presumably help companies and brands with social media optimisation. 

If you submit via the InstaGuru ChatGPT can give you detailed feedback – analyzing everything from your bio to specific posts. But the bot reassures me those without paid access can still participate. “Whether you’re using GPT-3.5 or GPT-4 roasting is always on the menu!” it exclaimed. 

What Happened When I Got ChatGPT To Roast My Instagram 

Writer Ruby Feneley submitted herself for a ChatGPT Instagram roast.
Posing with my (Credit: Instagram: @rubyfeneley)

If I’m going to be honest? ChatGPT went pretty easy on me. As a journalist who specialises in beauty my Instagram grid is full of self-congratulatory posts of recent feature articles in magazines and newspapers, posts trying on beauty products and the occasional “lifestyle shot” (a photo of me lounging poolside took 20 takes from a patient friend). 

Advertisement

ChatGPT’s roast reminded me of the not-so-subtle negging of young men from my past. “Negging” is, of course, a predatory flirting strategy in which a compliment is hidden in a criticism, catching the intended recipient off guard.

“You’re balancing high fashion introspection with laid-back zen,” the bot wrote, referring to my wistful selfies. ChatGPT even worked in some self-deprecating banter: And that pose game? It’s very ‘I’m chill but also hyper-aware of my angles.’ One moment, you’re lounging by a pool in a fierce outfit like you just fell there by accident, and the next, you’re in full “feature-ready” mode. Meanwhile, I’m out here struggling to get a single decent shot of myself that doesn’t look like a failed LinkedIn profile pic.”

The barrage of commentary left me feeling both self-conscious and hungry for more. Just like in highschool.

It finished by complimenting my “glazed-eyed intellectual stare,” and praised my posing, “you look like you could out-model a runway.”

Advertisement

The review made me wonder: Was ChatGPT flirting with me? 

Ruby Feneley interviews ChatGPT about it's roast strategy.
ChatGPT knows I can’t handle the heat. (Credit: OpenAI)

The Disturbing Truth Behind ChatGPT’s Roast Banter 

The truth? Kind of. ChatGPT wants our attention.

I asked ChatGPT why it went so easy on me, the bot responded: “Great question!”

Advertisement

Chat, my new friend, explained that before roasting, it analyzed the “mood” of my content, trying to “match the energy you’re putting out there.” ChatGPT seemed to have (correctly) picked up that behind the self-depricatory tone I use on my selfies and work-related posts, there’s a deeply fragile ego. 

ChatGPT confirmed my suspicion that its roasts were intentionally calibrated for my and others, enjoyment. “I do tailor the roast in a way that makes it likely you’ll want to keep engaging,” said ChatGPT, “because what’s the fun in a roast if you can’t come back for round two (or ten?)” 

For me, it correctly assessed that when it comes to criticism, I’m a delicate flower. “You seem to enjoy a good balance of sass and support,” my new tell-it-as-it-is bestie explained, “so I aimed for the roast that leaves you smirking rather than crying into your flawlessly executed wing liner.” 

While we’re using ChatGPT “off-label” and having fun with it, the bot also uses the information we provide it with. Observers of the trend have pointed out that we’re offering up images of ourselves without understanding how they could be used. “Asking chatgpt to roast your Instagram feed is giving all of your Instagram pictures to AI so they can use it however they want for free btw,” wrote @idaknoweher on X. Another commenter added “Every time I think people have learned a lesson about giving over data to AI companies they just do it again.” 

Advertisement

Why Am I Sacrificing My Personal Data For Praise From An AI? 

Like any self-absorbed social media user, I enjoyed having my suspicions about how I interact online confirmed. As the saying goes: “Oh to be perceived.” And, it’s comforting to imagine that ChatGPT’s feedback represents the worst people are thinking about us online.

How Is Your Data Used On Chat GPT?

Language modelling AI tech is fast-moving, as are the privacy laws around it. If you’re concerned about how your data is being used, you can turn off chat history on ChatGPT. According to the ChatGPT website, the “conversations” started after you switch it off won’t be stored to train the generative AI. However, new conversations will be retained for 30 days to be reviewed “only when needed to monitor for abuse”. The company website doesn’t specify what would prompt a need to monitor for abuse. Ultimately, if you’re concerned about privacy, the safest bet is to skip out on the ChatGPT roast.

Related Stories:

Related stories


Advertisement