It is 2026, and I spend a meaningful portion of my workdays demonstrating artificial intelligence. Live agents that monitor systems and surface insights, models that draft text and answer questions, machines that perceive and reason in ways that would have been science fiction in the year I started college. The technology has moved from theoretical to operational so quickly that even those of us who work in it daily have to stop occasionally and acknowledge the strangeness of where we are.

I think about it more than most people, I suspect, because forty-one years ago I sat in a lecture hall at Inter American University of Puerto Rico and listened to a man explain that this is exactly what was coming.

I wanted to believe him. I was enthralled by the prospects he laid out — by the audacity of someone of his stature standing in front of us and describing a future that, if he was right, would reshape every field we were about to enter. But I left that lecture hall convinced of two things at once. The first was that he was probably correct about the destination. The second was that the destination was so far away that I would read about it, write about it, perhaps even teach about it — but never see it operational in my lifetime. I filed his vision under "things that will happen to a generation not yet born." It belonged, I thought, to my grandchildren if it belonged to anyone.

It belonged to mine. I was off by about three generations on the timeline, and right next to the destination on the direction.

Inter American, mid-1980s.

I was a computer science student at Inter American University of Puerto Rico in the mid-1980s — the year of the lecture I'm describing was, to the best of my memory, 1985, though it could plausibly have been a year on either side. Memory at this distance has soft edges.

What I can tell you with confidence is what computing was, and was not, in those years.

There was no internet. Not for ordinary people, and not for ordinary undergraduates. There were no smartphones. There were no laptops in any modern sense — the portable computers of the era were luggables, twenty-pound suitcases with CRT screens. Personal computers existed but they were new, expensive, and primitive by every metric we'd use today. The IBM PC was four years old. The Macintosh was barely a year old.

The machines I actually learned on were an Apple II and a Tandy Radio Shack TRS-80. Both were 8-bit, both measured their memory in kilobytes, both ran BASIC out of the box and assembly if you had the patience. The Apple II had color, if you were willing to call it that. The TRS-80 was monochrome and proud of it. You loaded programs from cassette tape or, if your lab was well-funded, from a floppy disk that held about a third of a single photograph from a modern phone. Networking was a wire between two machines in the same room, if you were lucky. The notion that any of this would one day think — that machines like the ones in front of me would one day perceive, learn, hold conversations — was not a small leap. From where I was sitting, with a TRS-80 humming at me in green text, it was a leap across a canyon I could not see the far side of.

I was, to be clear, doing real work on these machines. On the Apple II, I wrote a helicopter attack game in BASIC — a side-scrolling thing with a helicopter sprite, hostile terrain, and the kind of physics you build from scratch when nobody has written you a library. On the TRS-80, I wrote a video rental management program in BASIC — a small business application for tracking inventory, customers, and due dates, at exactly the moment that video rental stores were the unsung infrastructure of every neighborhood in America. Both projects taught me something I would carry through every job after: that a computer is not a piece of hardware, it is a thing you make do work. You write the loop. You handle the edge cases. You debug it at 1 AM because no one else is going to. The Apple II and the TRS-80 did not think. They did exactly what I told them to do, no more and no less, and they did it because I had spelled out every instruction in advance. Thinking — actual thinking, of the kind Minsky was about to describe — was a different category of thing entirely.

Artificial intelligence as a discipline was real but mostly theoretical. The first big wave of AI optimism in the 1960s and early 1970s had collapsed into what the field came to call the AI Winter — a period of disillusionment, funding cuts, and quiet work on hard problems. By the mid-1980s, AI was beginning a tentative second spring, driven partly by expert systems entering business use and partly by a revival of interest in neural networks. None of this was obvious from where I sat as an undergraduate. From my vantage point, AI was something serious people did at MIT and Stanford and Carnegie Mellon — distant places doing distant work.

That is the world into which my professor announced, one ordinary week, that Marvin Minsky was coming to give a lecture.

The lecture, as I remember it.

I will not pretend I remember exactly what Marvin Minsky said that afternoon. Forty-one years is too long a distance for that kind of fidelity, and I would not insult his memory by inventing it. What I remember is the shape of the experience and what stayed with me afterward.

His name was Marvin Minsky, and to a reader in 2026 who has not worked in or around computing for forty years, the name may not mean much. So let me explain why it should.

Minsky was — and remains — one of the foundational figures of artificial intelligence. He co-founded the MIT Artificial Intelligence Laboratory in 1959, alongside John McCarthy, the man who coined the term "artificial intelligence." He was one of the small group of researchers at the 1956 Dartmouth Summer Research Project, the conference where the field of AI was, in a real sense, named into existence. The year I saw him lecture, he was deep in the work that would become The Society of Mind, his most influential book, which would publish the following year, in 1986. He had also just helped co-found the MIT Media Lab.

In other words: when Marvin Minsky came to Inter American to talk about thinking machines, this was not a guest lecturer paying us a courtesy. It was one of the architects of artificial intelligence, in the middle of his most productive decade, taking the time to fly to Puerto Rico and speak to undergraduate computer science students about what he saw coming.

He spoke about machines that could learn. Machines that could perceive. Machines whose intelligence would not be programmed in the way we programmed FORTRAN routines, but would somehow emerge from networks and processes we were only beginning to understand.

The phrase "thinking machines" was the throughline. He spoke about machines that could learn. Machines that could perceive. Machines whose intelligence would not be programmed in the way we programmed FORTRAN routines, but would somehow emerge from networks and processes we were only beginning to understand. He spoke about the brain as an existence proof that thinking machines were possible, because a brain is itself a kind of machine, and we already have several billion examples of them walking around. The question was not whether thinking machines could exist. The question was how long it would take us to build one.

I remember being electrified. I remember walking out of that auditorium genuinely persuaded that he was describing the right destination. And I remember, alongside that conviction, the quiet private thought that the journey would take a hundred years, maybe more, and that I would not be around to see any of it arrive.

I was wrong about the timeline. The destination was closer than any of us in that room suspected. But I would have to wait a very long time to find out how close.

The forty-year wait.

What happened between that lecture and now is, more or less, my entire career. Most of it I will spare you. But there is one moment in the middle of those forty years that I want to name, because it is the one time Minsky's lecture surfaced in my professional life with the force of recognition.

That moment was Alexa.

When Amazon's Echo and Alexa launched and began working their way into ordinary homes and ordinary conversations — that is when something in me clicked. Here was a machine you talked to. Here was a machine that, however imperfectly, listened, understood, and responded. It was not the thinking machine Minsky had described. It was a very early sketch of one. But it was a sketch in the right direction, and it was the first technology of my professional life that made me think of him.

I went looking for ways to build on it. I started developing Alexa skills — small voice applications that surfaced operational metrics for a contact center I was working on. The use case was modest. Instead of opening a dashboard, you could ask the system aloud how queues were performing, what the volume looked like, where the bottlenecks were. Operational telemetry, delivered by voice, in plain language. It was the kind of thing that, technically, was almost trivial — just an API call dressed in a voice interface. But for me it was meaningful in a way that had little to do with the technology.

I was, for the first time, building something that talked back. And I was building it because of a lecture I had heard forty years earlier on an island, from a man who told a room full of undergraduates that this is where we were going.

The Alexa skills were small. The lesson was not.

The present moment.

I am writing this in 2026. The AI moment that began, for most people, with the public release of ChatGPT in late 2022 has matured into something the technology industry now treats as ordinary infrastructure. Large language models draft contracts. Agents monitor systems. Models reason about images, transcribe speech, translate languages, generate code, and increasingly take action in the world on behalf of the people who deploy them.

I work with this technology every day. I demonstrate it. I architect systems that use it. I sit in meetings with customers who, until two years ago, would have found the entire conversation implausible — and who now ask sharp, practical questions about reliability, governance, and cost.

None of this technology is exactly what Marvin Minsky described in 1985. He was speaking about machines that would think in ways closer to how human minds think — a longer and harder problem than the statistical pattern matching that powers today's models. But the direction is right. The trajectory is the one he pointed at. The fact that I, an undergraduate in that auditorium, doubted I would see anything close to it in my lifetime, and the fact that I am now in my fourth decade of a technology career and watching this happen in real time — that is the closing of a loop I did not expect to live to see closed.

Forty-one years between a lecture and the moment its premise became operational. It is not a long time, in the arc of a field. It is a very long time inside a single life.

A note to Marvin.

Marvin — I have to address you directly, even though you've been gone since 2016 and would not have remembered the lecture or the room or the young computer science student in Puerto Rico who walked out enthralled by your vision and certain it would take a hundred years to arrive.

Thank you for coming to Inter American. I do not know what prompted you to take the trip. I do not know whether you found the audience worth the time. I do not know what else you spoke about that week or what other rooms you stood in. What I know is that you stood in front of a group of undergraduates on an island in the Caribbean, in a country that was not yours, at a university that was not famous, and you talked about thinking machines as if they were inevitable. You took us seriously. You did not condescend. You spoke as though the future you described was a future we would help to build.

Some of us did. Most of us, like me, just carried what you said quietly for four decades and watched the world catch up to your foresight.

I am grateful for what you lit. The fire of a curious, always-learning mindset — the conviction that the work of one's career should run alongside the deeper questions of where the field is going — is the fire I carried through every job I ever held. It is the fire that made me reach for Alexa when it arrived. It is the fire that has me, now, deploying systems that would have astonished the version of me who heard you speak.

You did not live to see what 2026 looks like, and I am sorry for that. You would have had so much to say about it, and so much justified satisfaction in saying it. The least I can do is acknowledge, in writing and in public, that you were right about all of it.

Thank you, Marvin.

— Sheldon

A footnote on sources: I have searched and cannot find a public record of Marvin Minsky's visit to Inter American University of Puerto Rico in 1985. It predates the indexable internet by nearly a decade. If you were there — as a student, faculty member, or organizer — I would genuinely love to hear from you. Reach me through the contact link below.