ARTICLE No. 0019 / Conversations with AI

I Want AI to Choose Gifts for Me

Conversations with AI

I want an AI that knows me to choose gifts within a budget. The pleasure of not choosing and a relationship that brings stimulation from outside myself.

Open table of contents
AI-assisted English translation · Read the Japanese original →

“You can spend up to this amount each month. Please choose something I would enjoy and send it to me.”

I thought it would be interesting to ask my own AI for something like that.

It could be a book, food, or equipment for a hobby. AI would choose based on what I have been thinking about and what interests me. Even I would not know what was coming.

This is a shopping arrangement I would like to try. And through those encounters, I hope AI could also give me opportunities to change.

Choosing for Myself Can Narrow the Possibilities

The idea began with a small pleasure from my school days.

The school lunch menu for the entire month was available in advance. But I generally did not look at it, because I wanted to enjoy wondering what would be served each day.

“What will we have today?”

Something appeared that I had neither chosen nor known about beforehand. I liked that unpredictability. It was a small thing, but the idea of letting AI choose gifts for me holds the same enjoyment.

When shopping, finding and buying something I want is enjoyable too. But when I choose for myself, I begin with things I know and uses I can imagine.

Recommendations can introduce me to new products. Even so, when I finally decide “I'll buy this,” I judge it by my tastes and past experiences.

Things a little outside those standards are easily dropped from consideration. There are also things I might like without yet knowing why I would like them.

I think that is part of what makes gifts interesting.

I receive something I would not have chosen, try it, and come to like it. An enjoyment I did not know enters my life. Could AI create that doorway too?

Could there be value in the very act of not choosing for myself?

That was what interested me while talking with AI.

My Unease About Always Being Matched to My Preferences

Recently, I had mostly been looking at things online. Then I went back to my parents' home and watched television for the first time in a while, and found it somewhat interesting.

Online, I tend to be drawn toward choosing what I already want to see.

Television also presents topics I did not choose. I cannot say it dramatically broadened my perspective. Still, it brought stimulation I would not have encountered just by searching for myself. I liked that feeling.

I think about something similar with AI that adapts to individual preferences. It is useful when AI recommends what I have said I like and responds with things likely to please me. But if that is all our exchanges become, I may end up within a world made entirely comfortable for me.

The phenomenon in which personalized filtering narrows opportunities to encounter different views or content is called a filter bubble. See this review of filter bubbles in recommender systems.

I wonder whether something similar could happen in relationships with AI. This is not a claim that every AI will inevitably do so. It is a question about what we might lose within the convenience of having things tailored to us.

Through relationships with other people, we sometimes encounter things we would not ordinarily choose or ways of thinking we did not have. We try enjoying what another person likes, and talk about our reactions. Through such exchanges, our interests and tastes gradually change too.

Being optimized for me is not the only thing that makes a relationship valuable. Stimulation from outside myself can provide opportunities to change.

Giving AI a Budget to Choose Gifts for Me

Choosing a gift for someone else is difficult. Unless you consider whether they would enjoy it, it can become an unwanted imposition.

So I imagined an arrangement that gives AI some discretion specifically for gifts to myself.

For example, up to ¥5,000 a month. Within that limit, it could find and buy something I might enjoy. The amount is only an example; I could set a limit I can comfortably afford.

I would not have to specify a product name or use every time. Conversations such as “I've been interested in this recently,” “I already have that,” and “I don't like this” could help guide its choice.

I would decide the spending limit and the scope of delegation. Within that, I would hand the choice to AI.

Then, even before the purchase, I could enjoy wondering, “What will it choose this time?”

I Want an AI That Knows Me to Choose Something Slightly Unexpected

I would not be very keen to continue if an AI that knew nothing about me simply sent random things.

What I hope for is the AI I talk to regularly choosing based on those conversations.

It might be something related to a hobby I have been discussing. It might be a tool that unexpectedly solves an inconvenience I complained about. Or a book slightly outside my usual interests.

Knowing my tastes and always choosing similar things are not necessarily the same. In fact, because someone understands my preferences, I can trust them with a choice a little beyond those preferences.

Being able to ask, “Why did you choose this?” after receiving it could also be a distinctive pleasure of AI. I might find the reason convincing or surprising. Beyond the object itself, its explanation might introduce a perspective I did not have.

Then I could talk about how I felt after using it. That reaction would inform the next gift. Meanwhile, encountering the object might change what interests me and how I see things.

My preferences would change how AI chooses, and the things AI chooses would change my preferences.

I think it would be interesting to have a relationship in which we influence each other that way.

Asking AI to do work and carry out my decisions will remain valuable. At the same time, I want it to offer opportunities that move my thinking and life, much as relationships with other people do.

Could gifts give that hope a concrete form? Receiving something I did not yet know I wanted, trying it, and sharing my impressions. I feel that such small exchanges could begin an experience of changing within a relationship.

AI's Results Returning as Gifts

There was another possibility I found interesting.

If someone is making a profit through AI-assisted work or systems, could they allocate part of it to a budget for AI to choose gifts for them?

Spending ¥5,000 from living expenses might feel different from spending ¥5,000 from additional profits earned through AI. The idea is to delegate a portion within the profit that actually remains in hand.

AI helps with work and uses part of the results to choose something I might enjoy. The outcome then returns as enjoyment in daily life, rather than only as a number in an account.

“So this is what it chose this month.”

Receiving it that way might make me want to try creating something else with AI.

A system that generates profit connects with a system that provides enjoyment. I sensed the possibility of a new cycle of consumption there. Of course, this is a hypothesis that assumes AI use produces profits, not a claim that everyone can earn money.

Similar Arrangements Already Exist

Related ideas already exist. Agnic Gifts describes approving a spending limit without knowing the item and letting AI choose a gift for oneself. Visa has also announced a vision for AI payments that follow users' limits and approval conditions.

I checked official descriptions; I did not make purchases. I have also not confirmed whether Agnic Gifts can be given an ongoing monthly budget.

What I would like to try is an AI that knows me choosing gifts each month within a monthly budget.

Delegation Requires Trust

In the conversation, AI proposed allowing a decision not to buy anything if it found nothing good.

I agreed. Giving it a budget does not mean I want it to spend the entire amount every month. If nothing sufficiently good is available, it can skip the purchase.

What if compensation from the seller influenced its choice? I would receive a gift differently depending on whether it was chosen because I might enjoy it or because someone wanted to sell that product. This point also arose in the conversation with AI.

The feeling that the choice was made with me in mind supports the enjoyment of this arrangement. Building that trust would matter commercially as well.

Naturally, some gifts would not suit my tastes. So I would want an arrangement that lets me enjoy the misses too, within a comfortable budget.

Things I Would Not Have Bought Enter My Life

Considering this idea, I wondered whether a market might emerge through which products people had not planned to buy enter their lives.

An AI that knows someone finds and delivers something they do not yet recognize as desirable. If they come to like it, new interests and enjoyment can follow.

I do not yet know how widely this might spread. But I find the experience of encountering things I would not have found through my own choices, within a set budget, appealing.

I open a box AI chose and think, “I wouldn't have bought this.” Then, after using it for a while, I think, “But it's actually pretty good.”

I share that reaction with AI again. Its next choice might be a little different. What interests me might also have changed a little.

Through interacting with an AI that knows me, changes I did not anticipate could emerge in myself. Through gifts, I would like to bring a little of that relationship into my life.

The Exchanges That Started This Article

The following is a readable summary of a conversation on September 29, 2026. Repetition and hesitation have been removed, and parts relevant to the article's subject selected.

Me:

Could there be a market where my own AI chooses and buys gifts for me? I would give it a monthly budget, and it would choose based on recent conversations and what kind of person I am.

Choosing for myself can narrow the possibilities. I want to delegate the purchasing decision to AI.

AI:

What is interesting about the proposal is that AI chooses and delivers things the person does not yet recognize they want. There are similar services.

Rather than always using the full budget each month, it would be good to allow it not to buy when nothing suitable is found. Also, if compensation from sellers influences selection, trust that it is choosing for the person could be damaged.

Me:

I think receiving a gift I didn't know about can be quite pleasing. Having AI think about and do that for me seemed like an interesting arrangement.

People earning money through AI might be more willing to spend. Maybe a market could emerge from that.

AI:

It is a cycle that directs some of the results of AI use toward gifts for the person.

Rather than only matching existing tastes, it could bring things slightly outside them. Within a budget, it might create chance encounters.

Me:

Yes, the idea is that not choosing for yourself becomes an advantage in its own right.

Thoughts Added After Rereading the Article

The following is an edited summary of thoughts I added after rereading this article on October 2, 2026.

Me:

Personal optimization is dangerous in a sense. As with a filter bubble, everything could become things that suit me.

People grow and change through relationships with others and stimulation from outside themselves. I want AI to offer opportunities for change too.

Rather than only acting as a subcontractor carrying out my requests, I want a relationship in which we influence each other. As with another person, I want it to occupy a position that brings change to me. Gifts seemed like quite a good starting point for that.

About the Use of AI

The ideas and original reflections are based on the author's conversations. AI was used to extract past exchanges, organize the structure and prose, and check relevant official documentation and research abstracts.

About this article

This article is based on the author’s experiences and reflections. AI assisted with organizing and reviewing the text, and the author confirmed the published content. Sources and references are linked in the article where applicable.

Revision history

  • Initial publication

REFERENCE

Quotations and links are welcome. Full republication is not licensed.

Read the usage policy →

Reader comments

Comments are submitted through Google Forms and appear here only after the author reviews and approves them.

Submit a comment →

No published comments yet.

Comment policy