Lukianoff’s Presentation Harmfully Misrepresents AI Art

I took one of the above photos of a lakeside restaurant near Lake Eerie, and the other came out of Midjourney, an AI image generator. Can you tell which one’s which?

Probably. We’ve become really good at telling what is AI and what is not. So good, in fact, that the school community instantly flagged Greg Lukianoff’s school meeting presentation as entirely AI-generated. From ancient philosophers to aliens gathered around a UFO, we saw, time and again, the same distinctive artistic style screaming, “This is AI slop!” in our faces.

As the crowd hummed with whispers in response to his “slop slides,” a different question popped into my mind: How developed is AI image generation really? I refused to believe that Lukianoff’s glaringly AI-generated visuals were truly the final frontier. 

As it turns out, AI is getting really good.

Even through cursory internet browsing, I’ve realized that these models are, in truth, much further along than I ever expected. They were nothing like the ones on Lukianoff’s slide deck.

To give you an idea of what I mean, neither of the above images is real. 

Lukianoff’s presentation, though, is not problematic just because he used AI. The real issue is that his use of bad-quality images further desensitizes our community from the growth of AI image generation.

Moreover, these images are becoming harder and harder to flag, as evidenced by a study from Microsoft AI for Good. Researchers found that participants’ ability to distinguish between AI and real was only slightly better than random chance. This statistic was worse when participants were given landscapes instead, where the success rate was as low as 21.5%.

In my search for more of these photos online, I came across an article about AI art from Shalhevet High School’s newspaper, based out of Los Angeles, CA. In it, the writer includes artwork made by Midjourney, an AI image generator. Glancing through its visuals, my fingers stiffened on the trackpad. I was shocked: the art wasn’t just human-looking, it was moving. It elicited feelings of nostalgia and appreciation so profound that I had to snap myself back into reality, tell myself hey, this isn’t human made. But what does it matter what the person or thing behind this artwork wanted me to feel? That’s when it hit me. We’re past the point of being able to tell if art is human-made or not.

While this does raise fundamentally disruptive questions on what art actually is, I think the biggest issue isn’t addressing these new questions, but rather getting more people to recognize that the ability of this technology simply isn’t what it used to be. 

I think we are most definitely at the point of AI being good enough to replicate realistic-looking images, and we’re certainly way past the point of our art being replicated.

Despite the obvious philosophical complications that arise from this, I think it’s more important for us as a school community to focus on what this advancement actually means and what it looks like. It’d be impossible for us to consider the ramifications of a new technology if we don’t know the extent of its abilities.

The audience’s reaction to Lukianoff’s presentation, therefore, does not make us a school that is thriving without AI, but rather one that is oblivious to what it’s capable of.

The fact that most of my conversations with fellow students ended with jokes about how terrible the slides were should deeply concern us. By treating an emergent technology like a joke, we significantly diminish its implications, which are real and imminent. We cannot pretend for any longer that the growth of these influential technologies has halted, nor can we deny that we are barreling toward a future where we can’t control this scarily advanced technology.

Of course, this is not to say that we should shortcut all things creative, elbowing past each other to crank out Midjourney photos and never have to think for ourselves again. My goal here is not to question the Deerfield mission of fostering intellectual curiosity, but to raise the point that we simply cannot be intellectually curious about the things we do not know about. 

The moral, even existential questions that new technologies raise can only be asked if we understand what they are capable of. 

So what does this mean for our community? It means that Lukianoff’s slide deck should have been flagged for its misrepresentation of AI’s abilities, which can have deep-seeded, albeit unseen impacts on an already-insular community.

Lukianoff’s presentation, therefore, was a hidden disaster for the school community. While it might seem like our reaction is confirmation that AI is not taking jobs anytime soon, it’s just evidence that we as a school community are simply unaware of its terrifying new abilities.

The more we cherrypick examples of AI slop to inform what we believe about these technologies’ capabilities, the more we will perpetuate an echo chamber of denial. 

To that end, let’s take Lukianoff’s presentation as a warning for just how sheltered we are from AI’s growth. 

Disruptive technology isn’t waiting for graduation to rear its ugly head. It’s already posing new existential issues. It’s already disrupting realms of human creativity thought to be untouchable by machines. 

How can Deerfield students hope to answer these dire questions if we graduate without even knowing what AI is capable of?