Caitlin Clark has taken thousands of jump shots. I doubt any of her coaches watch her practice and think of ways to outsource her next thousand. We understand that the repetitions have something to do with why she can make the shot when it matters, and that knowing how to shoot doesn’t excuse her from picking up the ball again tomorrow.
I keep thinking about those repetitions as Silicon Valley offers to relieve us of more of the thinking that fills an ordinary day, including the parts we have learned to resent.
On Sept. 8, Meta introduced Muse, a personal AI agent that can book travel, manage email and negotiate on a user’s behalf. According to the company, it can remember preferences, fill out forms and keep working after the app is closed, attending to the details while its owner gets on with something else. The appeal is easy enough to understand for anyone who has reached the end of a day and wondered how so much effort produced so little satisfaction.
Meta Chief AI Officer Alexandr Wang has been selling that promise in the most persuasive terms available. “Muse will make you money!” he wrote after a user reportedly found more than $1,500 in unclaimed funds, later promoting another user’s airline refund.
There is considerably less work involved in persuading someone to try the future when the future appears to owe them a check.
I understand the attraction, and I have no interest in pretending that an afternoon spent arguing with an airline is an essential part of a well-lived life. Some chores deserve to disappear, especially when getting rid of them gives someone the time or energy to do something they otherwise couldn’t. What unsettles me is how easily that sensible concession can expand to cover almost anything that asks something of us.
We have become generous with the word ‘busywork,’ applying it to tasks whose inconvenience is obvious and whose value is harder to name. Planning a trip means deciding what is worth the money and what you can live without, often in conversation with someone whose answer differs from yours. A calendar can force an uncomfortable reckoning with your priorities, while writing an email sometimes requires you to reconsider a sentence because you can imagine how it will sound to the person receiving it.
There is thinking inside those errands, even when we would rather be doing something else. An awkward first draft can reveal that you haven’t decided what you mean; a close look at your spending can expose a habit you had managed not to notice. I worry about losing those small encounters with our own decisions, particularly when the service taking them over is good enough that we have little reason to look behind the result.
Researchers use the term “cognitive offloading” for shifting some of a task’s mental demands onto an outside aid, something people have long done with notebooks, calendars and calculators. That history is a useful check on the temptation to treat every new convenience as a threat to civilization. The question becomes more complicated when the tool can also weigh the options, choose an approach and act, leaving us to decide how much of its reasoning we need to understand.
A 2025 study by Microsoft Research and collaborators surveyed 319 knowledge workers and found that greater confidence in generative AI was associated with less reported critical thinking during AI-assisted work.
The researchers also found that some thinking shifted toward checking and integrating the machine’s output, a reminder that using AI can require judgment of its own. These were workers’ accounts of their behavior, not evidence that AI causes lasting cognitive decline, but they suggest that how much we trust a tool can influence how carefully we engage with its work.
Earlier research on automation bias and complacency describes related risks, including accepting automated recommendations too readily or failing to monitor a system closely enough. None of this establishes what years of using Muse will do to a person’s abilities, and an honest argument has to leave that uncertainty intact. It does give us reason to be careful about assuming that a decision has become less deserving of our attention because a machine can make it.
Meta has addressed the more immediate risks with safeguards that, according to the company, include a dedicated virtual machine, a separate security agent and approval requirements for sensitive actions such as purchases or sending emails. Users can limit access and review what Muse has done, which matters when software is acting in their name.
Yet an approval button still depends on the person pressing it having enough understanding, and enough interest, to decide whether the proposed action makes sense.
I suspect the larger surrender, if it comes, will feel reasonable at every step. You hand over the refund request because you have already spent too long on it, then let the agent arrange a trip because it knows your budget and has done well for you before. After enough useful results, even reviewing the details can begin to feel like doing the work twice, and the habit of asking whether you agree gives way to the relief of seeing that something is finished.
I use AI every day, so this concern comes with no claim to purity. I want the hours back, too, and I can imagine putting them to better use than the tasks that currently consume them. What I cannot assume is that the saved time will automatically become time spent thinking more deeply, or that judgment will remain sharp because I once had plenty of practice using it.
For me, that means keeping some of the work close, particularly the work through which I discover what I think or understand another person. I can ask for help with a difficult email and still take responsibility for figuring out what I owe the person reading it, which often takes longer than finding the words.
Perhaps that is what keeps bringing me back to the practice court, to the athlete doing something familiar because she wants it to remain hers. I want AI in my life, and I want some of the difficulty left in it, too, if that is what it takes to stay capable of living it for myself. However much Muse learns to do for me, there are still parts of life where I want the ball in my hands.
AJ Fabino, Senior Editor, Stock Trader Network
Opinion Story
