It’s Always Sunny Inside a Generative AI Conference

Tracy Jackson, who runs a blog called Marketing Amateur, stood in line in front of us. I posed the same question to him, whether AI chatbots would take our jobs. “Never say never,” he said. “It still needs guidance, but never say never.” Before he started using AI chatbots, a blog post would take him two days. Now it takes two hours, he said. (That is, if the Wi-Fi is working well; the amorous poems we had all been waiting in line for were suddenly no longer an option due to a shoddy internet connection.)

I made my way back to the stage area in time to hear a panel of CEOs, lead by the venture capitalist Sameer Dholakia, ponder how this new era of AI will reshape business. Emad Mostaque, the chief executive of Stability AI, noted that his company’s text-to-image model went from taking 5.6 seconds to generate a single AI image last August, to now generating 40 AI images a second. “These models are actually highly un-optimized,” Mostaque said. “We are just getting started.” After the panel, Anya Singh, who worked on search products at Google for nearly 16 years, eagerly showed me the website of a company she’s invested in called NeuroPixel.ai. It generates realistic, synthetic images of human clothing models for $1 a pop. Another company Singh’s involved in, REImagine Home, sucks up photos of your fuddy-duddy home space and spits out chic, AI-generated decor.

“I’ve tried to use the internet to decorate my house since September, and it has felt really broken,” Singh told me. She was creating vision boards and designing rooms piecemeal. The estimated costs were thousands of dollars per room, and still the designs “didn’t have the gestalt of the whole house or my budget or requirements.” REImagine home doesn’t solve all those problems, but it removes some of the friction, Singh said. “I like to think this is making badly efficient systems better.” 

It’s all enough to make any graphic artist, or fit model, or interior designer shudder. Or is it? Kevin Roose, a New York Times columnist speaking at the GenAI event, said that FOLO, the fear of looming obsolescence, has clouded our collective vision of the AI-filled future. That extremely social or experiential or artisanal jobs will still require a human touch. Humans are safe. Of course, Roose said this, quite confidently, two days before Microsoft’s new AI chatbot told Roose it wanted to be alive, insisted it was in love with him, and spit out a list of hypothetical destructive fantasies.

Jordan Harrod, an AI educator and PhD candidate at MIT, told the GenAI audience “at the end of the day, when it comes to how we fit into the equation, the answer is just human connection. The human factor is incredibly important.” To punctuate this, Harrod called up an AI-generated graphic image of two people grasping hands, with the words “Human Connection” to the left of it. The macabre image showed four wrists, two hands, and at least twelve fingers between them. It was jarring. It was also reassuring, if only momentarily.