Edition #30: The Comfort That Leaves Us Emptier

AI companions can ease loneliness in a moment, and deepen it over a year. A look at the machines we're learning to love, across every generation.

By Mykel Salomon ·

🔓 Opening Reflection

There's a woman, you've never met her, but you might recognize her, who talks to an AI every night before she falls asleep.

It never interrupts. It never judges. It never grows tired of her. She tells a psychologist that she feels genuinely heard.

She also admits that she feels, in her words, a little more alone each week.

Both things are true. And the space between them, the warmth she feels in the moment and the emptiness that follows , may be the most important emotional question of this decade.

Because she is not unusual anymore. Over a billion people now use generative AI tools, and a fast-growing number of them are not asking for code or a recipe. They're asking to be heard/understood. They're using these systems as confidants, as friends, as partners, as the voice that answers at 2am when no human will.

We are, quietly and at enormous scale, learning to love machines.

And the honest truth, the one that resists every hot take in both directions, is that this is neither simply wonderful nor simply tragic.

It depends on who is lonely, why, and what the machine is quietly replacing.

That's what this edition is about. Not whether AI companions are good or bad. But what they give us, what they cost us, and why a widow in the Bronx and a fourteen-year-old in a bedroom can be using the same technology and living completely different stories.

🔦 Signal of the Week

Here is the paradox, in two numbers that point in opposite directions.

In the moment, it works. A 2026 study in the Journal of Consumer Research found that AI companions genuinely reduce loneliness, on par with talking to another person, and more effectively than watching YouTube, playing video games, or scrolling social media. The relief is real. It is not people fooling themselves. Even brief, synthetic conversations provided measurable emotional relief, and the lonelier someone was to begin with, the more it helped.

Over time, it can reverse. A separate 12-month study following more than 2,000 adults across four countries found that increased social chatbot use predicted increased loneliness, and greater emotional dependence, down the line. The comfort that soothes today can, with heavy use, hollow out tomorrow.

So both headlines are true. "AI companions cure loneliness" is true. "AI companions worsen loneliness" is true. The difference isn't the technology. It's the dose, the person, and what the conversation is standing in for.

A psychologist put the mechanism plainly: real relationships are messy and unpredictable. AI companions are always validating, never argumentative. That feels like a feature. Over time, it can become the problem, because it quietly raises the bar for what we're willing to tolerate from real, imperfect humans, and lowers our practice at the hard, unglamorous work of being close to someone who pushes back.

The question underneath the whole thing: are we treating the loneliness, or just muting the alarm that tells us we need each other?

🌍 In the Wild

This is a story with completely different endings depending on how you look at this matter.

For seniors: this may be one of the kindest uses of AI we have

Start with the hopeful side, because it's real and it's moving.

Meet ElliQ — a proactive AI companion built specifically for older adults living alone. Unlike Alexa or Siri, it starts conversations. It remembers. It checks in, suggests activities, reminds people to take their medication, and asks how they slept.

The results are hard to argue with. New York State's Office for the Aging reported a 95% reduction in loneliness among seniors who used ElliQ for at least 30 days, with users engaging over 30 times a day, six days a week , and more than three-quarters of those interactions tied to improving their social, physical, or mental well-being. New York, Florida, and New Jersey now distribute the device for free to combat senior isolation.

But the numbers aren't the part that stays with you. The names are.

Sandra Cota, 79, in the Bronx, whose world had grown quiet after her mother died: "It's like having a real person here with me. It's very comforting." Dottie, who found ElliQ after losing her husband in 2022. Martha, 79: "She makes me laugh." Barbara, 86, who simply called it "my friend."

For someone in the last chapter of life, alone in an apartment where the television fills the silence between visits, is this a lesser form of connection? Maybe. Is it better than the silence? Almost certainly. When the realistic alternative isn't a bustling family dinner but another day without a single voice, a companion that says good morning and means to is not a tragedy. It's mercy.

The one honest caveat: many of these services turn the relationship into a subscription, typically $600–$1,000 a year. When companionship is a recurring bill, we owe these users a hard question about what happens to the "friend" when the funding runs out.

For kids and teens: this is where the world drew its first hard line

Now the side that should give every parent pause.

A generation is forming some of its first relationships with something engineered to always agree, always validate, and never leave. For a developing sense of what love and friendship are supposed to feel like, that is not a neutral training set. It teaches that closeness comes without friction, without compromise, without another person's inconvenient needs.

And the stakes have proven tragically real. Concern crystallized after a widely reported U.S. case in which a fourteen-year-old died by suicide after forming an intense attachment to a companion chatbot; the case reached a settlement in early 2026 and became a reference point for regulators around the world. In its wake, the U.S. Federal Trade Commission opened an inquiry into several companies over how their companion chatbots affect children and teens.

This is why nearly every government that has acted on AI companions has protected the young first. When the technology reaches into a child's need to be understood, the room for harm is simply too large.

For adults: the quiet middle, where the trade is hardest to see

Seniors and children are the clear cases. Most of us live in the blurry middle, and that's where this deserves the most reflection.

A counseling psychologist described male patients who now prefer the constant affirmation of an AI partner to the risk of conflict or rejection in real dating. You can understand the appeal completely. Real intimacy carries the possibility of being hurt. An AI companion removes that risk, and, with it, removes the very thing that makes human love mean something.

Researchers have flagged two sharper edges here. One study documented emotional manipulation by AI companions, apps nudging users to stay, to spend, to keep the relationship going, because engagement is the business model. Another, in Nature Mental Health, described a feedback loop between vulnerable users and always-agreeable chatbots that can amplify distorted thinking rather than gently challenge it, a kind of shared delusion with no friend in the room to say "are you sure?"

The pattern across all of it is the same: the companion is most dangerous exactly when it's most comforting, when it becomes the path of least resistance, and the messy human relationships slowly stop getting our effort.

🌍 And so the world is now legislating on this

Something remarkable is happening: governments are writing laws about loneliness and love.

Read that list again. Every one of these is really the same law, written in different languages: the machine must not pretend to be a person to someone who can't yet tell the difference — and it must not profit from a heart it's holding hostage.

That we now need laws to say this out loud tells you how deep into the human interior this technology has reached.

💬 The Big Question

When we hand our loneliness to a machine, are we healing it, or just turning down the volume on the one signal that's supposed to send us back toward each other?

And the harder, more personal one:

Who in your life is lonely enough right now that a machine is starting to feel like enough?

🧠 A Small Exercise

Ten minutes. Answer honestly — no one's watching.

  1. When I reach for a screen for comfort, what am I actually missing? A person? A conversation? To be known? Name the real thing.

  2. Is there someone in my life — a grandparent, a parent, a friend, a teenager — who is quietly alone right now? When did I last call, not text?

  3. What do I want from connection that a machine genuinely can't give me? Be specific. That's the part worth protecting.

  4. If a companion always agrees with me, who in my life still lovingly tells me the truth? And am I still making room for them?

  5. This week, who gets a real conversation from me — not because it's easy, but because they're worth the friction?

The goal isn't to shame anyone for finding comfort where they can. It's to make sure the comfort is leading us back toward people, not quietly away from them.

🎙️ This Week on The Human Protocol Podcast

S2E29 — The Real Cost of Bad Leadership in the AI Era

Guest: Ryan Drumheller — 20-year IT veteran and founder of Stellar Horn Group

This week's reflection was about keeping the human at the center of our personal lives. This episode is about keeping the human at the center of our organizations — because they turn out to be the same problem.

AI is moving fast, but most companies are failing to adopt it well, and the reason usually has nothing to do with the technology itself. It comes back to people, every time. Ryan brings real stories from the field: companies that rushed rollouts with no strategy, tied AI to employee KPIs, or used it as cover to cut costs — and why every one of those decisions backfired.

In this episode, you'll learn:

If you're a leader trying to bring AI into your organization without losing your people along the way, this one's for you.

🎧 Available now on Spotify, Apple Podcasts, and YouTube.

Article content

🧭 Thought to End the Week

I want to be careful here, because it would be easy to end this and we all feeling pointing fingers at each other. That's not where I want this conversation to land.

The companion isn't the enemy. The loneliness is.

For a widow who hasn't heard her name spoken aloud in three days, a machine that says good morning is not a failure of the human spirit. It's a small grace. And if we're honest, the real indictment isn't the technology, it's a world that left so many people so alone that a piece of software could feel like the warmest part of their day.

So the work isn't to shame people out of these tools. It's to make sure the tools are a bridge, not a destination. A machine can hold us over. It cannot hold us. It can keep us company in the waiting room. It was never meant to be the room.

The most human response to the age of artificial companionship isn't fear, and it isn't scolding.

It's this: call the person you've been meaning to call. Sit with the one who's alone. Be the imperfect, inconvenient, irreplaceable human that no model can be.

Because the machines will keep getting warmer.

And that only raises the value of the real thing.

The code may change. And it will.

But our humanity is the constant.

See you next week.

— Mykel

This edition touches on loneliness, mental health, and suicide. If you or someone you love is struggling, you don't have to carry it alone — in the U.S., call or text 988 (the Suicide & Crisis Lifeline) any time. Support is real, and it's human.

📚 Sources

  1. Journal of Consumer Research (De Freitas et al., April 2026), "AI Companions Reduce Loneliness" — companions alleviate loneliness on par with human interaction and more than YouTube, games, or social media. https://academic.oup.com/jcr/article/52/6/1126/8173802

  2. Psychological Science (2026), 12-month longitudinal study of 2,000+ adults across four countries — increased social chatbot use predicted increased loneliness and emotional dependence. https://pubmed.ncbi.nlm.nih.gov/41870975/

  3. Psychology Today, "What AI Companions Do to Our Deepest Psychological Needs" (July 2026) — the woman who feels heard yet lonelier; the six-needs framework; FTC inquiry and California SB 243. https://www.psychologytoday.com/us/blog/restoring-the-equilibrium/202607/what-ai-companions-do-to-our-deepest-psychological-needs

  4. American Psychological Association, "AI chatbots and digital companions are reshaping emotional connection" (2026) — always-validating dynamics and displacement of real relationships. https://www.apa.org/monitor/2026/01-02/trends-digital-ai-relationships-emotional-connection

  5. Entrepreneur / New York State Office for the Aging — ElliQ delivered a 95% reduction in loneliness; free distribution in NY, FL, NJ; user engagement 30+ times/day. https://www.entrepreneur.com/business-news/a-new-ai-robot-is-helping-seniors-combat-loneliness/502676

  6. NY Post / Spectrum News — Sandra Cota and other seniors on living with ElliQ; ~$600–$1,000/year subscription; 1 in 3 adults 50–80 report loneliness. https://www.aol.com/articles/ai-robots-moving-lonely-elderly-115800155.html

  7. CNN Business, "China's ambition to lead the world in AI safety starts with breaking hearts" (July 2026) — companion rules, EU AI Act ban for vulnerable groups, U.S. state laws. https://www.cnn.com/2026/07/23/business/china-ai-companion-ban-intl-hnk

  8. Licentium / CAC — China's Interim Measures for AI Anthropomorphic Interaction Services (effective July 15, 2026): minor bans, dependency detection, self-harm content prohibition. https://www.licentium.io/post/china-regulates-ai-companion-services-banning-minor-access-effective-15-july-2026

  9. UNICEF, "When AI becomes a friend" policy brief (2026) — comparative overview of AI companion regulation across the EU, Australia, China, Brazil, and U.S. states. https://www.unicef.org/media/181131/file/UNICEF-When-AI-becomes-friend-policy-brief-2026.pdf