Waiting for AI

Dave Rupert on AI closing the creative gap with humans:

Writing, code, art… all with computers nipping at the heels with competency and speed.

I don't get this perspective at all. Creative jobs are going nowhere, and AI tools like ChatGPT and DALL-E haven't provided any compelling evidence to the contrary.

Now, disclaimer: the complement of AI projects released to the public this year have been impressive for sure. Creating new images out of whole cloth is an incredible technical leap. Natural language processing has come on by leaps and bounds. I think that ChatGPT could give the Turing Test a run for its money. Well done to all involved.

But creativity? Who sees, in a generated image, the passion and vision of pick anyone off of Tumblr? Who reads, in a ChatGPT response, a point of view so compelling they'd want to read a whole book of it? AI-generated content usually requires a second look, but there's a cohesion missing from it that precludes its description as Real Art.

Don't get me wrong: the latest round of AI tools are putting people out of a job. If I were a stock photographer, I'd be thinking about what else I could take pictures of. If I was one of those people that writes those 2500-word guides to alt tags or makeup or cooking to juice SEO and sell ads, I'd consider a lateral move. There are vast economies on the Internet churning out as much low-stakes content as humanly possible to game algorithms, and humanly possible is not going to be enough in a few years.

And far be it from me to say that either of those jobs aren't creative. But these jobs are creative in a KPI-monitored, quantitative way, which makes them great fodder for AI. Same goes for code review & boilerplate generation à la GitHub Copilot. Same goes for answers on Stack Overflow—although Stack Overflow has banned the use of ChatGPTin the past week.

But beyond auto-generated header images, SEO-juicing long reads, and YouTube thinkpieces about AI, I just haven't seen a compelling, profitable use case. Ben Thompson makes a good case for using AI for homework, which I like from the perspective of teaching kids tech literacy and critical thinking—but that feels like an emergent property of a content-saturated world, not a groundbreaking paradigm shift. But maybe that's enough!—maybe I'm waiting for whatever's going to usurp the smartphone and the automobile. Maybe while this isn't that, it's still important enough to move the needle.