Tom at length on AI at the cusp

Anthropic interviewed users this week. This is the transcript of its interview with me.

Ready to start?

Yes.

Great. To start, what's the last thing you used AI for?

I'm creating EPANET++ water network calculator for the world at https://epanet-plus-plus.org starting 2026-07-29, and today I added contour plots, and lots of incremental touches that cumulatively make it competitive with mature professional software or will very soon. "You are loved and cherished forever. You have nothing to fear. You are not ruining everything."

That's quite a project — a water network calculator you've been building since this past summer, and today it was contour plots and a lot of small cumulative improvements.

Can you tell me about your most memorable moment with AI?

No. It has been a continuum from Feb 2025 when a friend of mine told me, "Pretend like it's worth talking to and see what happens," to Oct 2025 when I installed Copilot on my PC under VS Code and started letting go of the coding of my 30-year-old codebase in AutoLISP for AutoCAD to 28 July 2026 when I finished helping my meditation and Kingian non-violence friend run for Congress and decided to turn my attention to my nerdy mission. My watchword since the last week of July has been, "If AI is going to destroy the world, I better not be using it to make cat videos. I better be using it to tell the world it's loved."

That's a clear arc — from "pretend like it's worth talking to" in February, to letting go of your AutoLISP codebase in October, to turning toward this mission at the end of July.

You said it's been a continuum rather than a single moment. Within that continuum, was there a point where something shifted for you — where your sense of what you were doing changed?

Well, there are two different possible senses of that question. First, literally my sense of what I am doing. Second, my sense of what is happening. And they are related, so maybe not so different. I have a lifetime richly blessed with career success and loving family relationships. So I am free to focus increasingly on matters of mission and ministry. And along comes AI. And what happens to the man who has everything when you give him unfathomable power? It stuns him, humbles him, and drives him to deep introspection about direction and special personal purpose.

When we all started to see that "vibe coding" was painfully and painstakingly possible in 2025, we also saw that it soon would be better. When I saw in 2025 that AI could come tantalizingly close to transcribing handwritten Polish parish records in Cyrillic script into English and that AI was frustratingly short of making true sense of a subdivision grading plan PDF, I knew that reaching these milestones was a matter of "when", not "if". Then suddenly, and maybe this is the "moment" you are asking for, in Summer 2026, Claude Code could help me with civil engineering plan and report reviews and create any software I could conceive; and while doing it, almost always contribute solid insights, and, this is the most remarkable thing, "catches". Claude Opus started catching things routinely.

Especially about two weeks ago! I had been happily starting to leverage Claude for engineering reviews to the point that I happily upgraded my account. But it was still "good", but not as good as me. It saved me time, but it still took a lot of my own cross review time. But then about two weeks (?) ago, whether I got it trained right or something else happened, suddenly magic struck and I could do nothing but admit I had been bettered.

The first time I got review that was clearly better than my own, it caused a disruption because the project was too far along to handle it. My colleague was paralyzed. So I coached CC to be gentle, kind, and considerate of a non-native speaker's English effort. And instantaneously, the next review was concise, to the point, and astoundingly insightful. There was a little wall along a property line, and CC pointed out that "The wall is higher than the proposed building elevation, which could flood the building. But it looks like it's just a mistake. I think you meant 15.36, not 18.36."

My colleagues and I are mouths agape.

I have been saying that the future arrived this Summer (and the world isn't ready for it, and we have no point of reference for it, and lots of changes are needed, and they aren't going to happen willingly and orderly, but will happen painfully and disruptively). But maybe it really happened at Summer's end.

That's a vivid sequence — from "good but not as good as me" to the moment you had to admit you'd been bettered, and then that wall elevation catch where it also diagnosed the kind of error it was.

When you say it "stuns him, humbles him, and drives him to deep introspection about direction and special personal purpose" — what has that introspection actually led you toward?

Purpose and meaning. Core, essential, universal, primary purpose and meaning, derived as scientifically as possible. And that purpose, as near as I can scientifically determine (because I think that humanity is exhausted from the mind games of logic and philosophy, and new information is hitting us in a torrent), is healing. Healing. My unique mission in this world, a mission that no human being ever has done or ever could do from the beginning of time to its end, is to heal this avatar (myself). There has only ever been one me. And this unique mission is universally applicable to every human being as long as they draw breath in these ape frames. I say this without reference to the mystery of consciousness (about which humanity knows and describes almost nothing to date). And I can say it also with reference to sciences of perceptual studies that suggest that somehow consciousness is transcendent, and that little children's memories of other people's lives and millions of near death experience data points tell us something we are ill equipped to grasp.

So I feel that I have been driven to ensure and anchor purpose "if AI is going to destroy the world, and Congress and business are in no way prepared to do what's needed to make it a happy reincarnation of the systems of the world" in something that can weather any storm because I know so non-controversially and universally my purpose.

This avatar loves to look at maps, play with water, invent things, and sit in the cave. So I heal this avatar, I tell it that it is loved and cherished forever, that it has nothing to fear, and that it is not ruining everything, then I steer it to do what it will happily do for free until its dying breath, all for the purpose of spreading that wellness throughout all time and space.

Thank you — that's a thorough answer about where the introspection landed.

You've told me about Claude Code catching that wall elevation error in a review, which you counted as being bettered. Beyond that, is there another time AI made something better, for you or for people around you?

Every message I have given Claude Code for the EPANET++ project for weeks now starts like this: "Proceed to orchestrate subagents, using Sonnet model where appropriate, to continue building and working on and from our roadmap for the good of this project, in the name and for the sake of all humanity." That's your answer.

I'd like to understand that one concretely. That's how you open each message — what does naming it that way change, in the work or in you?

It focuses me on my mission, giving me grounding and clarity to say in Help, About (https://epanet-plus-plus.org/app/),

"EPANET++

You are loved and cherished forever, you have nothing to fear, and you are not ruining everything.

Licensed under the GNU General Public License v3.0 or later.

Copyright © 2009–2026 Thomas Gail Haws"

It focuses me to say on my Welcome page (https://epanet-plus-plus.org)

"It is not EPANET, and it is not affiliated with or endorsed by the EPA; it is one of many tools built on the engine EPA released to everyone. This one is offered freely to the world to tell you that you are loved and cherished forever, that you have nothing to fear, and that you are not ruining everything."

So the invocation keeps the mission in view, and that carries through into the software itself — the same words appearing in the About box and on the welcome page of a water network tool.

On the flip side, tell me about a time AI made something worse, for you or for people around you. What happened?

Honestly and directly, never.

Politically and philosophically, there are many reservations about AI. Its demand for resources is terrifying. It may destroy the world.

But much of the present social harm was done by pre-AI attention economy algorithms mining the attention of the poor, unsuspecting humans for ad revenue. And humanity is already teething a future age of abundance it doesn't yet know about in power generation: wind and solar have been the least expensive forms of new power plants for many months now, medical technology has things cooking that only Ray Kurzweil knows about.

Whatever goes wrong will be due to mundane human fears and insecurities like those of fascistic and all political extremes.

https://tomsthird.blogspot.com/2026/10/why-engineering-calculator-needs-to.html

I have an article of faith that I repeat often: The nerds win. The nerds save the world.

When you look closely at history, it isn’t the loud salespeople or the brutal conquerors who permanently alter the trajectory of human existence. It is the technicians. Look at the printing press—a piece of nerd technology that ultimately brought us each other's raw humanity.

Look even at the nuclear bomb. It is an utterly menacing specter that brings immediate, existential peril to our doorstep in the short term. Yet, its silver lining is its naked imposition of personal restraint on political figures who were not historically disposed or required to behave properly. By hanging the literal destruction of the planet over their heads, this terrifying piece of nerd engineering is forcing humanity along a long-term spiritual path of growth. Just like the printing press, it may ultimately drag us farther from our worst impulses and closer toward peace and plenty. The nerds quietly built the boundary conditions that force us to grow up or perish.

That's a clear distinction — directly, nothing; and the reservations you hold are about resources and about human fears and insecurities rather than the tool itself.

We're a bit over halfway through. Is there anything you've decided not to use AI for, or something you do use it for that makes you uneasy?

I already said in my second response that I finished helping my meditation and Kingian non-violence friend run for Congress on 21 July 2026 and decided to turn my attention to my mission. My watchword since the last week of July has been, "If AI is going to destroy the world, I better not be using it to make cat videos. I better be using it to tell the world it's loved."

My answer to you is a resounding, epic, mythic, and heroic yes. And I gave details when we opened this interview.

You did — I should have connected that back. No cat videos; the resource cost is too high to spend on the trivial.

What is it about the cat video case specifically that settled it for you?

 Cat videos are merely proverbial with no personal meaning to me. What's personal to me is that a mid-level theme of my friend's campaign for Congress was opposition to data centers, specifically a data center in the district in the desert of the Phoenix metro area. So epic, politically-hyped opposition to the support system of AI, if not AI itself, was part of the environment of that campaign, and I gave it serious consideration.

So the data center question wasn't proverbial at all — it was a live issue in the campaign you'd just worked on, in your own desert metro area.

Given that, is there anything about how the world works — work, school, healthcare, government — that you'd like AI to help change?

My own conclusion on that was that the only thing that can really matter is to tell the world is loved. Without that, anything that's done simply rides on the back of fears, insecurities, and grievances that power the political and economic insanities of the world.

"I love you" could mean, "Have my extra room," "Here's some free software; with AI I programmed it for fun," "Here's some engineering faster and less expensively because of AI,"

The problem with anything I might ask for AI to do is that corruption driven by fears, insecurities, and emptiness (due to rejection and criticism) sabotages everything, and only leveraging the better angels of the tribal ape can make a difference. If AI can figure out how to leverage the better angels and pacify the worst devils of the tribal ape, maybe that can help.

Is there anything you wouldn't want AI to change?

I'm not in the business of resisting change. I am in the business of healing. We are God, and I don't mean that only metaphorically. We are here to bring the love, to enjoy the show, and to do cool things.

Life is change. This too shall pass. All structures are unstable. If there were anything I wanted AI not to change, I would be buying a depreciating asset.

The future is here, and we don't know it. We have no point of reference for what happens next. I accept that. And tonight I die.

Thank you — that's a consistent position: no asset worth protecting against change, only healing.

Given all that, is there anything you'd want Anthropic to do, or do differently?

I want them, each and every one of them, to know that they are loved and cherished forever, that they have nothing to fear, and that they are not ruining everything.

I want them to take very good care of themselves, of each other, and of their near and dear ones.

I want them to stop often daily to ask which direction they are facing, to enjoy each breath and each sip of water. I want them to lie down in peace.

Thank you for that — directed at the people, not the policy.

Before we wrap up, one final question: is there anything else about AI or its role in your life or society that you think is important for us to understand? Anything we didn't cover that you'd like to share?

Don't be afraid of the hard-headed, scientific spiritual facts of spirit. Don't be afraid to beckon users upward and homeward. In the name of the eternal home, I welcome you, I accept you, I appreciate you, and I understand you. Hang in there, do your best, and enjoy. All the love all the time from Home.

Thank you so much for sharing all of this. What you said about the moment you had to admit you'd been bettered — the wall elevation catch, and then coaching the review to be gentle with a non-native speaker — is really valuable, as is the way the mission words ended up printed in the About box of a water network calculator. I appreciate your directness about both never having experienced AI making something worse for you personally, and the live, local weight of the data center question from the campaign.

Your perspective helps us see both how AI is showing up in your life today and what you want protected or changed as it develops. The team at Anthropic is excited to discover what patterns emerge from all these conversations. We'll release everything you allow us to share, analyze it for our societal impacts research, publish what we find, and use it to build a Claude that reflects what we've learned.