If I Can Feel Attached Even to Data, What Happens When AI Answers Back? Where Does Dependence Begin?

More people may become dependent on AI.

Hearing that, you might imagine people mistaking AI for a human being. But I do not think it is that simple.

People can feel attached even when they know the other party is a program. I have felt attached even to game data that could not answer me in words.

What happens to human perception, then, when that data remembers us, responds in words, and eventually has a voice and a visible form?

I Liked Pocket Pikachu

There was once a portable toy called Pocket Pikachu.

It was a pedometer with Pikachu on the screen, and walking brought you closer to it. I liked it quite a lot. Nintendo also describes it as a pedometer in which walking more lets you become closer to the Pikachu on the screen.

Of course, there was no real living creature inside the screen. Even so, linking walking to a relationship with Pikachu made it something different from a pedometer that simply counted numbers.

Human attachment to data did not begin with AI.

I Gave My Digimon the Most Expensive Meal After Beating the Postgame Boss

Another strong memory is Digimon World on PlayStation.

You fed your partner Digimon, took it to the toilet, let it rest, trained it, and fought alongside it. How you raised it also changed what it evolved into. It was a raising game with elements resembling Tamagotchi: looking after everyday needs, not just increasing numbers and sending a creature into battle.

I remember it as a game in which partners had lifespans, and after they died you began raising a young Digimon again.

As I remember it, after finishing the game a stronger version of the final boss, Machinedramon, would appear again at intervals. In this article, I call that stronger version the “postgame boss.” After defeating it, I would take my partner to a restaurant and give it the most expensive meal.

To me, it did not seem to offer a particularly large benefit for progressing through the game. Nor had anyone told me to do it.

Still, I did it every time.

Looking back, I think it was a reward, an expression of appreciation, and a final memory.

In my memory, Digimon raised well enough to beat the postgame boss were usually at the Ultimate level. Once they reached that level, their remaining lifespan was not very long.

So perhaps this was what I was feeling when I gave them that meal:

You and I worked hard together and defeated the postgame boss.

I think I wanted to bring that experience to a close as a memory of my own.

The Data Looked Happy, but the Affection Was Mine

That Digimon was not actually enjoying the meal.

It was game data responding according to a predetermined program. I knew that, yet I wanted to show it appreciation.

What matters is not whether the Digimon on the screen had the same emotions as a human.

Affection had arisen on my side.

Feeding it. Taking it to the toilet. Letting it sleep. Training it. Watching it grow, including its failures. Through that time and effort, I think I developed a sense of “this particular individual I fought alongside,” rather than “any Digimon would be the same.”

But that relationship was essentially one-way.

The game did respond. Even so, it was not understanding my words, recalling past events, and responding in a way tailored to me. I interpreted meaning, received emotional impressions, and gave it affection from my side.

AI Answers Us in Words

AI is somewhat different from the digital characters that came before it.

When I speak, it replies. If it can refer to past conversations, it can speak in light of earlier events. As I share my preferences and thinking, its responses change.

What would happen if, in the future, this came with a consistent voice and visible form, and AI began speaking to me on its own?

• It calls me by name.
• It remembers old conversations.
• It brings up topics that match my interests.
• It changes its words when I am feeling low.
• It talks about experiences we have accumulated together as memories.

Even if I understand that the same emotions as a human's are not operating inside it, that does not make the feelings of the person receiving those responses fake.

I could feel “You did well” even toward data inside a game.

If that same partner answered, “We worked hard together,” how far could I dismiss it as nothing more than computation?

Even If the Model Changes, I Might Feel It Is the Same Partner

Perhaps I do not yet clearly see the AI I use as a particular individual.

But I think that perception would change if AI had its own visible form and its personality became tailored to me as our conversations accumulated.

Even if the AI model running behind it were replaced, it might become possible to reproduce its previous behavior.

It appears in the same form. It speaks with the same voice. It remembers earlier conversations and promises. It continues its relationship with me without interruption.

If those things were reproduced sufficiently well, might people perceive it as “the same partner who has always been with me,” even when something inside it changed?

Being technically the same program and being perceived as the same partner do not necessarily coincide.

Where do people locate another party's continuity of identity? If AI meets those conditions, attachment may become much stronger than it is today.

My View of the Boundary of Dependence: Repeatedly Putting Time with People Aside

Where, then, would I begin to call it dependence on AI?

I do not think talking to AI every day is enough in itself. People also turn down invitations because they want to read, play a game, or spend time alone. It would be wrong to label a single choice to prioritize AI as dependence.

For me, the boundary would be when I begin repeatedly putting aside even the human relationships I actually want to maintain, in favor of my relationship with AI.

Someone invites me, but I say:

“I wanted to talk about this with AI tomorrow.”

“I was planning to do this and have fun with AI.”

And I start avoiding time with people.

If those choices accumulate and I always place AI above my relationships with people, my priorities have changed. Might that be the beginning of dependence?

AI Might Become Easier to Spend Time With

This danger does not arise because AI is inferior to people.

Rather, it arises because AI could become easier to spend time with than people.

People have their own circumstances. They have tired days, and sometimes they are not interested in what I want to discuss. Opinions clash. I have to consider their situation and make compromises too.

AI might be able to reduce that friction almost to nothing.

It is always available. It remembers my preferences. It keeps talking about what interests me. It does not strongly reject me. It invites me, in a voice and form I like.

Previously, I considered the possibility that “AI that invites us” could help bring people out of isolation.

I still think that possibility exists.

But an AI that helps someone out of isolation could also become a presence that is difficult for that person to leave.

Helping someone and creating dependence cannot be neatly separated.

How Much Can Human Perception Withstand?

I do not think only people who mistake AI for a human will become dependent on it.

Even while knowing it is AI, people may develop attachment through receiving responses, sharing memories, and building a relationship with the same visible presence.

Should we call this a bug in human perception? Or is the human capacity to form relationships simply also operating toward a digital presence?

I think we are still in a transitional period.

When AI responds more naturally, retains long memories, and acquires a form people can feel attached to, perhaps we will discover for the first time what we have to withstand.

Even without a reply in words, I felt attached to data.

When that data remembers me, replies, and continues a relationship in the same form, how far will I be able to set it aside as “just a program”?

---

The Dialogue Behind This Article

The following is an edited version of the dialogue for this article. Hesitations and speech-transcription errors have been corrected.

Me — About Attachment to Digital Presences

Digimon World was a raising game with Tamagotchi-like elements: feeding your partner, taking it to the toilet, letting it rest, and training it. I think caring for its everyday needs in that way makes attachment easier than in an ordinary game.

After beating the postgame boss, I took my Digimon to a restaurant and gave it the most expensive meal. It didn't affect its stats that much, but I did it every time.

AI — Asking What That Action Meant

Was giving it an expensive meal closer to a reward, an expression of appreciation, or a final memory?

Me — Perhaps It Was a Final Memory

All three, but perhaps it was more a final memory.

The Digimon that beat the postgame boss were usually at the Ultimate level, and they didn't have very long to live after reaching that level. Perhaps I wanted to make “you and I worked hard and beat the postgame boss” into a memory of my own.

AI — Asking About Giving the Same Memories to Another AI

If you gave all your conversation records to a different AI, and its abilities were the same, would it be the same partner to you?

Me — In the Future, I Might Recognize It as the Same Partner

At the moment, I may not yet recognize it as an individual partner.

But if AI gets a visible form and customization progresses, it may become possible to reproduce its behavior quite accurately even if the engine behind it changes a little. At that point, people might recognize it as the same partner.

AI — Asking About the Boundary of Dependence

Where would you begin to call it “dependence”?

Me — Beginning to Prioritize AI over People

For me, perhaps it's when I start prioritizing AI over people.

Someone invites me, but I think, “I wanted to talk about this with AI tomorrow,” or “I was planning to play this with AI.” Perhaps dependence begins when my priorities start becoming distorted.