Face it, AI will come up at Thanksgiving dinner

Just know this: There’s going to be a conversation about artificial intelligence at Thanksgiving this year. 

An AI superfan is going to gush about chatbots and go on, at length, about how “These things just seem to know everything.” The dinner table’s funnyman will play a highly cringe video they made with the technology. Someone else will either be flummoxed or horrified. A proud guest will declare a vow of abstinence—in fact, they’ve never even used ChatGPT, they will reveal. One self-important guest will feel very smart when recounting the time they caught an AI making a mistake, once. They’ll tell everyone about it. 

These conversations will be bad. There will be camps: the Luddites, the accelerationists, the skeptics, and the 85-year-old ChatGPT power users. There will be the extant Elon evangelists, the people who are very tuned in, and the people who have not been paying attention to any of this. Conversations will touch on both the anticipation and the terror of the tech. The economy. The tech oligarchy. The environment. The bubble. No one will really be talking to each other. Not in any meaningful sense.

Have the conversation anyway. Not because you’ll form some sort of consensus, but because these long human conversations—at their best—come with love and also tension. AI provides neither. 

Of course, the major hurdle to reaching any sort of common understanding is that AI is too ambiguous a term to serve as a stable jumping-off point for a coherent discourse. 

For some, the term references a capital-intensive recipe of hyperscaled data centers and transformer models. For others, AI means the consumer-facing, knowledge-loaded models like Sonnet 4.5 and Grok 3. For others, there are simply the characters of “Grok” and “ChatGPT” and “Claude.” 

To many, AI is simply synonymous with the current age, some loose sense that the internet is increasingly automated and agentic. A good number of people simply use AI, increasingly intermingled with social media and the internet, as a shorthand for all technology.

All of these definitions are, in their way, completely accurate, and too divergent for having any meaningful discussion about AI. Proceed, still. 

We do not choose our families—or at least our blood relatives. AI, meanwhile, promises potentially limitless self-selection. Eras past gave us the YouTube rabbit holes and personalized algorithism. Now chatbots promise private universes of confirmation bias, personalization, and sycophancy. Younger generations are growing up with an unprecedented level of intimacy, and confidentiality, with these bots. 

Worse, people are forming deeply psychological and romantic relationships with AI tools, plopping their deepest selves into a digital abyss—instead of their loved ones or human professionals. This siloing away of our intimacy leaves us with impossibly difficult-to-predict consequences for our social skills and human relationships.

For this reason, many people who have eschewed AI are protecting themselves, while the rest of us are still looking to define the relationship we want with the technology. The challenge, of course, is that AI companies already know what kind of relationship they want us to have with these chatbots: an all-encompassing one, an endless saccharine dialogue. We don’t yet know what lifelong microtargeted conversational partners might do to us, but it’s probably not wonderful.

This raises the stakes for our human interactions, including imperfect and ever-trying family holidays. AI firms want to woo us with frictionless interfaces for everything. The antidote is sitting across the table from the people we care about, and all the friction they come with, to discuss our interesting times and everything else.