I tried to reassure a friend about AI at a party, but it scared us both
AI is inescapable in a way that makes you wonder why you took a day off in the first place. It’s not just that there are countless chatbots and generative models to choose from, it’s that the arrival and prevalence of this rapidly growing and changing technology is an earworm that everyday consumers can’t shake.
The relentless pressure of artificial intelligence is as contagious as Hey YaOutkast’s propulsive melody of the early oughts that helped perpetuate the myth that you have to shake a Polaroid to develop it. And Polaroid photos are actually a lot like AI, needing no help going from blank and fuzzy to sharp, bright, contrasty and specific.
The AI beat follows me everywhere. As someone who writes about technology for a living, everyone wants to talk to me about it. At work, where I write about it and often experiment with different generative models to test their limits, this is understandable. But now the AI chatter follows me home.
AI appetizer
I was recently at a friend’s house with over half a dozen high school friends, people I’ve known for literally decades. As we stood around drinking beverages (me, water) and sucking down cheese and crunchy snacks, one friend casually started quizzing me about AI.
As someone who works in banking, she had some idea of how AI could be used in the workplace, but she worried about more personal and hard issues. When would AI take over? When would we get too attached? When would we start loving a robot?
We talked about all of these topics in depth for about 30 minutes. I explained that one of the things that made it difficult to assess the AI path forward was that most of the AI model development for, among other things, the best AI chatbots had broken through the traditional 18-month development cycle of the technology and even Moore’s Lawwhich assumed that the number of transistors on a chip would double every two years, effectively doubling computing power.
AI, however, is amorphous and vast. I challenge you to find the beginning, middle, or end. It is a vast corpus of possibilities that extends in all directions.
AI models, on the other hand, developed at a pace of 3 to 6 months, and the leaps in intelligence and capability often appear to more than double in these time frames.
The more we talked, the more uncomfortable I became, which was strange. I’m usually the rationalizer. I take difficult technical concepts and explain them to friends and family (and sometimes to an audience on TV). What seems scary or out of reach usually isn’t. Technology—or even most devices—can be confusing when viewed as a whole. But when you break it down into its components or functions, the picture becomes sharper, clearer, and more understandable. Plus, most technology has a beginning, middle, and end. It’s a standalone smartphone, your computer, or even a VR headset. Software extends the possibilities, but doesn’t necessarily change them in any fundamental way.
AI, however, is amorphous and vast. I challenge you to find the beginning, middle, or end. It is a vast corpus of possibilities that extends in all directions.
Unfair desserts
The other thing I was worried about in our conversation was that I couldn’t comfort her. I couldn’t convince her that AI wouldn’t eat up certain tasks (basic writing, managing money and accounts, customer service). I couldn’t find an argument that said we would never love an AI.
On that last point, I went down a rabbit hole where humans are fundamentally complex biological machines, and the only difference between us and a robot with an AI brain was the level of sophistication. While AI can only mimic human emotions now, who’s to say that in ten years (or less) it won’t have something even remotely resembling it?
I remember Colin Angle, CEO of iRobot, once telling me how C-3PO-style robots were decades or more away, and we shouldn’t expect something like that to happen before 2050. But I realized we hadn’t spoken in a while, and maybe with the advent of generative AI, he was thinking differently. I told her about Future.AI and his new Future 02 robot. Sure, it walked poorly, but its torso and hands could move expressively, and OpenAI’s large language models gave it a simulacrum of personality.
Conversations with AI chatbots are not uncommon these days, and in recent months they have shown alarming signs of humanity. There’s a back and forth, I told my friend, that was once possible only with another person. She seemed fascinated and more than a little alarmed.
We agreed that the care wasn’t necessarily for us. We would be wobbly and old (we hoped) by the time AI took full command. The care was for our children.
She remembered how she once told her child that they could love anyone they wanted, except a robot. “That was a joke,” she began, “…and now it’s not.” I added.
I walked away from the conversation somewhat shocked.
As the evening was winding down, my friend was telling another friend how we had talked about AI and “Lance made me feel better about it.” The other friend asked, “Really?”
“No,” she laughed, “no, he didn’t.”