Entrepreneurs routinely email to tell me about a new app or tool they’ve designed. They promise it will reduce women’s cognitive workload or make it easier to share that load with a partner. In keeping with the “just-add-A.I.” frenzy that seems to be happening in every other sector, lately these emails also describe how the app has harnessed (or soon will) artificial intelligence.
I find most of these pitches underwhelming. It’s not that I’m anti-A.I. ChatGPT has helped me with both professional and personal tasks: it’s written quiz questions for my students, planned meals for crowds with diverse dietary needs, and generated travel itineraries.
Nor am I skeptical of A.I.’s potential to make household life easier. A customized tool powered by A.I. and designed specifically to help with household management tasks will undoubtedly do amazing things! Already, Outlook uses A.I. to remind me of promises I made via email and haven’t yet followed up on. It can, or soon will, sync with my husband’s and my calendars and alert us to potential scheduling conflicts or suggest we make time for events and chores we usually handle this time of year. I can only imagine what else is coming down the pike.
But cool as these features are, I doubt A.I. will meaningfully reduce our collective cognitive workloads. Instead, much like dishwashers and dryers, I suspect they’ll change the composition of our cognitive chore time and scale back the drudgery factor without notably reducing the overall footprint of housework on our lives. And that’s not necessarily a failure.
My skepticism is partly pragmatic. Outsourcing—whether to a human or a large language model—is rarely the unmitigated time- and effort-saver we expect it to be. Sometimes we outsource physical labor but create new cognitive tasks for ourselves: finding the cleaning service, scheduling and rescheduling the cleaners’ visits, remembering to tell them which spaces to avoid and which to give extra scrutiny.
When we outsource cognitive tasks, meanwhile, we typically take on new ones. I could teach an A.I. assistant rules for managing my schedule: no meetings before 11am, make sure there’s time for a workout most days, and so on. But many of my preferences are neither static nor ironclad. I sometimes make exceptions to meet early with collaborators abroad. If there’s a sudden break in the rain, that’s when I want to squeeze in my run.
A.I. can learn patterns, but it’s much harder for it to learn my system of values and ever-shifting priorities. Perhaps there’s a way for people more patient than I to convey and continually update this level of information for a chatbot. But in effect this means they are simply shifting from one kind of cognitive work to another, not eliminating cognitive effort altogether.
Let’s say some tech genius finds a way around these obstacles. Wonderful! But my skepticism is also philosophical. The idea that humans should, and one day will, outsource cognitive household labor to an A.I. assistant reflects a misguided understanding of what that labor is in the first place. It involves discreet tasks, of course, like researching flights or planning a birthday party, and many of those tasks can indeed be done as well or better by an algorithm.
More fundamentally, though, cognitive labor is relational work built on a deep understanding of who your family members are, what they need to develop and thrive, and what your individual and collective priorities are. It’s built on attunement to a child’s emotions, a spouse’s sensitivities, and a broader sense of what makes life worth living. It’s about care, which even the smartest of machines cannot replicate.
I don’t mean to romanticize cognitive labor. There’s a long history of that sort of thing, built on benevolently sexist notions of women as inherent nurturers who show their love by tending to others. Folks who look down on paid childcare and sneer at store-bought birthday cakes resist the idea that caregiving is work rather than high calling. By lumping those tasks in with the stuff we do for our bosses, they feel we cheapen and commodify (usually feminine) love and care.
The challenge is to find a way to hold onto two truths at once: care work, whether physical or mental, is both care and work. Labor and love. It’s irreducible to the kinds of prediction A.I. thrives on, and it’s also effortful and productive and deserving of recognition.
The way I’ve come to think about A.I. and cognitive labor, then, falls somewhere in between cynicism and blind optimism. It’s unlikely to change everything, and that’s probably for the best. But nor will it alter nothing. Perhaps you’ve heard of the Manager/Helper dynamic? In this arrangement, one partner/spouse is responsible for overseeing household affairs: deciding what’s worth doing, in what order, to what standards. The other assists with discreet tasks when called upon by the Manager.
A.I. has the potential to be the best Helper ever. It won’t passive-aggressively ignore the Manager, criticize her delivery, or forget a key instruction. It will do more thorough research, on a shorter timeline. But it cannot rise to the Manager level without killing some fundamental humanness at the core of cognitive care work.
Book updates
After chugging along for what feels like ages, I somehow have a complete manuscript of my book about cognitive labor in family life! There’s still plenty of revising to do, so don’t expect it in stores next month. But it’s been incredibly rewarding to have the space to tell stories and develop my ideas more fully, and I can’t wait to share it with you. More to come as I proceed slowly but steadily toward publication…