You can't fake sincerity with AI

I received an unsolicited email the other day that caught my attention, but not for the reason they would have hoped.

The email was titled “Partnership Proposal”, with my Instagram handle. I wondered why this was coming to my email instead of to my DMs, as this kind of spam normally does, but I’m not foolish enough to believe that this is somehow “beyond creepy” on the intrusiveness scale; it’s just too easy to connect those dots and do this sort of thing.

Okay, stranger, why are you in my inbox?

It was a marketing email for a service to “organically grow Instagram accounts”. Apparently I don’t have as many followers or as much engagement as I “should.” Strange, considering they failed to notice I also don’t have the retirement savings, own home, cis-het White heteronormative marriage, 2.5 children, investment accounts, luxury cars, vacation home, highly lucrative and successful career, and expensive tight body that I also, presumably, “should” have. Clearly they didn’t look as closely as they pretended to.

Wait, I skipped that part. That’s the good part.

The intro

I recently came across your Instagram page and was immediately drawn to your passion for horse life. Your profile showcases the beauty and joy of horseback riding, and I love the way you incorporate it into your daily life. Your web development skills are also impressive, and I appreciate the balance between your professionsal and personal interests. Keep up the great work!

Someone had clearly advised this enterprising young person to start these cold emails with some sign that they had checked their candidate’s target’s profile personally and determined that they were a good fit for this “partnership”. But, as I mentioned before, I’m just not foolish enough to believe that happened. In fact, this fake sincerity was a giveaway. It might have fooled anyone, except, you know, anyone who’s actually looked at my Instagram feed.

I used to post a lot of equestrian content to my account, but I decided sometime around 2020 to split that off into its own account (because horsey people can be so annoying with all their horse stuff all the time, you know?) And, yeah, I had occasionally posted some web dev stuff, mostly in my stories, but definitely hadn’t done that in quite a long time. So what was the deal with this person picking up on stuff I’m really not posting about these days?

Busted

Then it hit me: my outdated profile bio mentions those things.

Toronto beach life 🌅, web dev 👩🏻‍💻, outdoor adventures ⛺, & general observations 😎. Horse life 🐎: @s_v_equestrian

That’s it. That’s where they learned so much about me. That’s where they had evidently let chatGPT (or one of its friends) loose to pick up on a couple of elements and wax poetic about how much they cared about all my incredibly passionate yet disappointingly unengaging content.

It really would have only taken 30 seconds to see that Ihaven’t posted much in the last couple of years, and it’s mostly my dog and some drawings. Mentioning those would have been believable.

Fakers gonna fake

I’m sure this approach of scraping profile bios and using “AI” to write a cold email seemed like a great idea to someone (lazy, of course). I’m sure it works very well for a lot of people who keep their profiles more up to date and post predictable content. But for me, it was just one more example that these LLMs are not even remotely intelligent, and can’t yet do anything that real people do. They can, however, excel at what fake people do [but I never had time for them either].