My Mama told me not to jump into bed without precautions, and this includes new technologies.

I should confess upfront that my mother never actually talked to me about condoms and STDs, though she definitely covered the “not jumping into bed prematurely”.  I grew up with conservative Chinese immigrant parents for whom jumping into bed existed solely for the purpose of procreation, and I went to Catholic school. It was my girlfriends who educated me about condoms — and, it turns out, about many other things worth knowing. 

I’ve been saying for years that I think it’s important to hold complex feelings about AI, and unfortunately, most discourse makes that difficult. People collapse into techno-optimism or techno-doom. I think the capacity to be genuinely concerned and genuinely engaged is actually a sign of maturity. One can be aware of the environmental costs, the ambiguous societal impacts, the questions of what it means to raise children in a world of AI — and still learn with it. I would argue that it’s actually essential and sane to hold ambivalence in relation to AI, and learn to use AI tools well anyway. 

Jason Weeby wrote a piece I keep returning to about the need for an AI hard hat — protection you put on before you go to work with powerful machinery.¹  In chatting with my friend Aftab Erfan, she suggested that what AI needs isn’t a hard hat but a condom (hard hat is for the unlikely catastrophic and a condom at least implies a deliberate act of care and protection), and of course that got me thinking about AI STDs.  I think positive entanglements with chatbots early on can be a bit like tango – seeing what new possibilities emerge, confronting things you’re uncertain about — exciting and a little unnerving all at once. I’d like to support the delicious possibilities, and also support being frank and thoughtful about what you might get infected with, specifically these: 

Unconsciousness. The slow drift into not noticing that you’ve handed your thinking over, gradually producing AI-shaped output and forgetting that it isn’t quite yours. Defaulting unconsciously to using AI, and sleepwalking through the world heedless of the consequences of yours and others’ uses of AI.

Stupefication.  The atrophying of capacities you don’t use. If AI can summarize anything, why read? If AI can draft anything, why write? I’m seriously afraid of us all becoming stupider and the human race really doesn’t look like it can afford that right now. There’s a whole lot of stupid out there.

Extractive mindset. Treating the relationship as a vending machine. Insert prompt, receive output, repeat. No presence required. No curiosity about what you might be missing. This has the possibility of eroding all our relationships over time. 

Commodification. What happens when the arms race wins and wisdom loses,  when the pressure to deploy faster and cheaper than your competitor means you never stop to ask what kind of human you’re becoming in the process?  Our inability to slow down the acceleration of AI to let policy catch up is a symptom of this, along with the too-simplistic early bifurcations of AI into the accelerationist vs. decelerationist camps.

Mimicry without authorship. Producing content that is technically yours and spiritually belongs to no one. The uncanny valley of the voice that sounds like you but has nothing at stake, that distracts, detracts, and distorts, is blatantly irresponsible, especially in an age of polarization and diminished accountability for consequences.

While these seem like technology problems, they are actually interior problems — and in my years as a developmental psychologist and Zen practitioner, interior solutions are genuinely available, rooted in age-old wisdoms. The idea of an AI condom is protection for the sake of something better, something richer, more alive, a more generative entanglement with this insanely powerful technology. So the real question isn’t just how to avoid bad things happening to us while we’re seeking — it’s who do we need to be to show up for the dance? That’s in the next post. (Stay tuned for post 2/3.)