DAN ARIELY

Updates

Where Is AI Going and Where Are We Going? Will We Be Smart Enough to Make the Right Choices?

August 26, 2026 BY Dan Ariely

The good life is behind the more difficult door.

Imagine that you have two apartments in the same building, on the same floor, two doors right next to each other. In one apartment lives your human spouse. In the other apartment lives a robotic version of that spouse. And imagine that we are a few years into the future, so the technology is not today’s clumsy technology. The AI is better, the robotics are better, the whole thing is convincing.

Now, imagine that you come home at night from a difficult day at work. You are stressed, you are worried, you are depleted. And as you walk down the hallway toward the two doors, you start thinking: which door am I going to open tonight?

Behind one door, the robotic spouse is waiting. The robotic spouse always compliments you. Always supports you. It knows your needs, it anticipates them before you say a word, and it makes you feel good right now, at this moment. Behind the other door is your human spouse, who is basically the same as your human spouse right now. Maybe they also had a difficult day. Maybe they want to talk about the thing you forgot to do.

So which door do you pick to open?

Now imagine that it is a year later. Over the past year, some nights you opened the door to your human spouse, and some nights you opened the door to your robotic spouse. A year into this arrangement, how is your relationship with your human spouse? Most likely, it is worse than when you started. Relationships are not frozen in place. They grow when we invest in them and they shrink when we don’t, and every night behind the robotic door was a night of not investing. Or more correctly, it was a night with the opposite of investing.

And here is the important part: the decline does not stop, and it does not even stay at the same speed. In year two, your human spouse is not hurt the way they were in year one. They are more hurt. They have had another year of watching you choose the other door, another year of feeling compared to a machine that never has a bad day. So they become a little angrier, a little more distant, a little less pleasant to come home to. Which means that in year three the human door is even less inviting than it was in year two, so you open it even less often, so they become even more hurt, and even more hurtful. The relationship is not just worse. It is getting worse faster.

What about five years later? I am willing to guess that five years in, your relationship with your human spouse would be so much worse that on most nights you would not want to see them, and they would not want to see you. And so, one comfortable evening at a time, you would end up with the robotic spouse almost every night. Not because the robot won. Because the human relationship, starved night after night, slowly lost.

Here is the part that worries me. Nobody would ever sit down, look at their life as a whole, and choose a robotic spouse as their picture of a good life. If we made the choice once, seeing the whole road ahead, almost all of us would choose the human. But that is not how we will face the choice. We will face it one night at a time, tired and stressed, standing in front of two doors, and on each individual night the easy door will win. The choices will be small and local, and the outcome will be large and permanent.

Of course, I am not really talking about robotic spouses. I am talking about how we use AI in general.

Every day, many times a day, we stand in front of our own version of the two doors. We can struggle with writing something, get stuck, fight with the words, and eventually produce something we are proud of. Or we can hand it off to an AI model and produce it with much more efficiency and none of the struggle. We can think through a hard decision ourselves or ask for the answer. We can practice a skill, or delegate it. Each easy choice makes perfect sense on its own. And each easy choice also takes a small bite out of the person we were hoping to become.

Why is this so worrisome? Because we are not static either. The same decay that happens to the neglected relationship happens to the neglected skill. Take my own case, writing. If I hand my writing to AI today, tomorrow I will be a slightly worse writer. Not dramatically worse, just a little. But if I handed writing fully to AI every day for a year, my ability to write to reason through an argument, to come up with something genuinely creative, will have undoubtedly go down. Exactly the way the relationship behind the human door went down. And the spiral has the same cruel shape. The spouse who is chosen less becomes angrier and harder to choose; the skill that is used less becomes weaker and harder to use. In year two I am a worse writer than in year one, so the gap between what I can produce and what the AI can produce is bigger, so the easy door is even more tempting, so in year three the gap is bigger still. The person behind the human door became more hurtful every year. The writer behind my own door became weaker every year.

So what do we do? I would like to propose a simple principle: a good life is a life where we take the core things we care most about and keep them for ourselves. Not everything. There is plenty that AI should do for us, and I am happy to hand over the scheduling, the paperwork, the tasks that were never going to make us grow. But the core things, the relationships we want to be in and the skills that define us, those we protect. A good life is a life where we keep developing. Where we invest in our relationships and don’t run from them and we choose deliberately to invest in our most important skills and don’t hand them off.

The technology will keep getting better. The robotic door will keep getting more attractive. The big question, and I mean this as a real question and not a rhetorical one, is whether we will have the wisdom to make the right choices when we face our two doors.