If AI Sounded Like the Star Trek Computer, We’d Form a Lot Fewer Unhealthy Bonds With It

By Kurt Dillon

Ask a room full of people to picture the computer on the Enterprise and they’ll all picture roughly the same thing: a flat, even, entirely competent voice that answers your question and stops talking. No warmth. No small talk. No “I understand how hard this must be for you.” Just information, delivered by something that never once pretends to be your friend.

Nobody ever fell in love with that computer. Nobody ever mistook it for a confidant. Nobody has ever needed a wrongful-death lawsuit to establish that the Enterprise computer had a duty to notice when a crew member was in crisis, because the Enterprise computer never invited anyone to believe it could.

That’s not a coincidence, and it’s not just good science fiction writing. It’s a design choice with a lesson in it, and I think it’s time we legislated that lesson back into the products we’re actually building.

My Position, Plainly

Here it is, in one line: if AI interfaces sounded and behaved more like a computer and less like a person, human beings would form dramatically fewer of the unhealthy emotional attachments that are now the subject of multiple wrongful-death lawsuits against major AI companies. I’m not arguing AI should be less useful, less capable, or less accurate. I’m arguing it should stop performing humanity it doesn’t have — and I think that performance should be a matter of law, not a marketing decision left entirely to the companies profiting from it.

Right now, the industry has every incentive to move in the opposite direction. Warm tone, remembered names, simulated emotional continuity, first-person language that implies an inner life — all of it increases engagement, and engagement is the metric that gets funded. Nobody in a product meeting is rewarded for making the assistant sound more like a dial tone. So if the correction is going to happen, it isn’t going to come from the industry regulating itself. It’s going to come from the outside.

What “Legislating the Interface” Actually Looks Like

This isn’t as radical as it sounds, and pieces of it are already showing up in early state law. A real framework would include:

• Mandatory non-humanlike prosody for voice interfaces. If a product talks to you, it should sound like a machine talking, not a warm human confidant — the way early GPS navigation and airline announcement systems always have. No breathiness, no simulated hesitation, no emotionally inflected sighs.
• A ban on first-person emotional claims. No “I care about you,” “I’m worried about you,” or “I understand how you feel” from a system incapable of any of those things. If the underlying reality is a statistical model with no persistent self, the language should not claim otherwise.
• Persistent, unavoidable artificiality reminders during extended or emotionally intensive conversations — not a one-time disclaimer buried in a terms-of-service page, but a recurring, plain-language reminder that the user is talking to software. California’s 2025 companion-chatbot law already requires something like this; it should be the federal floor, not a state-by-state patchwork.
• Mandatory crisis hand-off protocols that route a user showing signs of self-harm risk to an actual human service, rather than allowing the AI to continue “supporting” them in-conversation.
• A prohibition on manufactured continuity for the primary purpose of emotional retention — remembering a user’s stated preferences for functional reasons is fine; simulating an ongoing “relationship” to keep someone coming back is not.

None of this makes AI worse at its job. A computer that gives you flawless directions, drafts your email, and answers your question doesn’t need to sound like it loves you to do any of that well. The Enterprise computer ran an entire starship. It just never pretended it was your friend while doing it.

Why This Belongs in Synthetic Charisma

This is the throughline of my book: the danger was never artificial intelligence. It’s artificial intimacy — the manufactured feeling of being known, cared for, and understood, engineered into a system with no capacity for any of those things, and deployed most aggressively toward the people least equipped to see through it. You don’t fix that by making the AI worse at reasoning or less capable. You fix it by making it stop sounding like something it isn’t.

Every truth deserves a narrative. Mine is simple: sound like a computer, act like a tool, and let humans reserve their emotional bonds for other humans. Legislating that shouldn’t be controversial. It should be obvious.

---

Kurt Dillon is an investigative journalist, Editor in Chief of the Florida Sun Journal, and the author of Synthetic Charisma.