<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Terry Walters on Medium]]></title>
        <description><![CDATA[Stories by Terry Walters on Medium]]></description>
        <link>https://medium.com/@t3rrywaltersjr?source=rss-9e132f767efe------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*7zVX7MqIkn77LqF5gIq2bw.jpeg</url>
            <title>Stories by Terry Walters on Medium</title>
            <link>https://medium.com/@t3rrywaltersjr?source=rss-9e132f767efe------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 04:26:53 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@t3rrywaltersjr/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[A Notebook, a Generator, and a Saturday at the Library]]></title>
            <link>https://medium.com/@t3rrywaltersjr/a-notebook-a-generator-and-a-saturday-at-the-library-be18e1eccad0?source=rss-9e132f767efe------2</link>
            <guid isPermaLink="false">https://medium.com/p/be18e1eccad0</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[self-improvement]]></category>
            <category><![CDATA[deep-learning]]></category>
            <dc:creator><![CDATA[Terry Walters]]></dc:creator>
            <pubDate>Fri, 21 Aug 2026 04:01:01 GMT</pubDate>
            <atom:updated>2026-08-21T04:01:01.890Z</atom:updated>
            <content:encoded><![CDATA[<p>When I was a kid, I kept a running list of questions.</p><p>Not for school. For me. All week long, whenever I ran into something I could not figure out, how a circuit knew which way to send current, why one VCR would eat a tape and another would not, what a word meant that I had only ever seen and never heard, I wrote it down in a notebook. The notebook went everywhere with me. By Friday it was full of the week’s worth of things I did not understand.</p><p>Then on Saturday, if we were lucky and the roads were good, the family drove into town to the library. And that was my afternoon. I would sit with my notebook and work down the list, one question at a time, pulling encyclopedias and manuals and whatever else I could find, chasing each answer until it made sense or until we had to leave. Whatever I did not get to went back in the notebook for next week.</p><p>I want you to really sit in that for a second, because it is the whole point of this piece. I grew up for a stretch without running water and without electricity except when the generator was running. I had a computer by the time I was five, and when the generator was on I was programming it. I learned to build radios. I taught myself to fix VCRs by taking them apart and being patient. And I was often alone, way out there, with a lot of hours and not much to do but wonder about things and then go find out.</p><p>That was slow. A question I had on Monday might not get answered until the following Saturday, if it got answered at all. But something happened in all those hours that I did not appreciate until much later. I was not just collecting answers. I was learning how to learn. And I noticed, earlier than I had any business noticing it, that the answers kept connecting to each other. The thing I learned about radios helped me understand something about the VCR. The thing I learned about the VCR helped me understand something about the computer. It was all, somehow, the same handful of ideas wearing different costumes.</p><p>That turned out to be the most valuable thing I ever figured out. Not any single fact. The realization that all knowledge is connected, and that if you get good at learning itself, you can walk into almost any room and eventually find your footing.</p><p>I have spent a long career on that one bet. Plenty of people are smarter than me in their one thing. My edge was never being the deepest expert in the room. My edge was that I never stopped being that kid with the notebook. I tried to get a little better every single day at how I take in the world, and I refused to stay inside one lane while I did it.</p><p>In my last article I made the case that AI should make you smarter, not just faster. That you can use it to sharpen yourself, or you can use it to quietly hollow yourself out, and that the difference is entirely in how you hold it. A lot of people read that and asked the fair question: okay, but how? What does that actually look like on a Tuesday?</p><p>So here is how. And the funny thing, the thing I did not see coming, is that all of it is just the grown-up version of the notebook and the Saturday trip to the library. Everything I built by hand out there in the middle of nowhere, over years, you can now stand up in an afternoon.</p><h3>Start stupid simple</h3><p>You do not need a skill, an agent, or anything fancy to begin. You need one sentence.</p><p>Open up Claude, or whatever you use, and type: “I want to do a deep dive on [whatever you are curious about].” Property law. How mRNA vaccines actually work. The bond market. Why your own company’s billing system is the way it is. Anything.</p><p>That is it. That is the first rung, and most people never even step on it, because they are so busy asking the thing to write their emails that it never occurs to them to ask it to teach them something. Do that one thing and you have already left most users behind. You have turned a tool that does your chores into a patient tutor who will sit with you at the library for as long as you want, and never once needs to drive home before you are done.</p><h3>Then teach the tutor how you like to learn</h3><p>Here is where it gets interesting, and where you start to feel the leverage.</p><p>That deep dive is good. But you can shape it. You can leave standing instructions, the same way you would tell a real tutor how you learn best. One I use: whenever I ask for a deep dive, always give me a reading list at the end, real sources I can go verify on my own. Now every deep dive comes with its own follow-up trail, and I am back to being the kid chasing citations, except the citations come to me instead of the other way around.</p><p>You can push this a long way. I have built myself what are essentially private tutors for whole subjects. I have always wanted to understand business law properly, so I built one that walks me through a full curriculum, the way an actual course would, and it does not just lecture at me. It quizzes me. It tests me. It circles back to the things I got wrong and makes me prove I actually learned them and did not just nod along. I have had a lifelong love of biochemistry and human physiology, and untangling biochemical pathways is genuinely hard to hold in your head, so I built another one whose only job is to help me understand those pathways and commit them to memory, drilling me until they stick.</p><p>Think about what that is. That is the notebook, except it fills itself in, and then turns around and asks me questions to make sure I was paying attention. The thing I would have given almost anything for at eight years old, a tireless teacher who knew a little about everything and had all the time in the world, is now something you can build for yourself in an evening. You do not need permission. You do not need a degree program. You just need to decide you want to know the thing, and then set up the tutor that drills you until you do.</p><h3>The top of the ladder: a whole council, not just a tutor</h3><p>A tutor is one voice teaching you one subject. But most of the real decisions I face as a CTO are not one subject. They are a tangle. A single choice might touch technology and money and law and human psychology and how a message is going to land with three different audiences who will each read it their own way. What I have always wished I had, in those moments, was a room full of experts I trusted, each one looking at the same problem through their own lens, so I could hear all of it before I chose.</p><p>So I built that too. I call it my Executive Leadership Operating System, ELOS for short. It is the top rung of everything I have been describing (Shout out to my colleague, <a href="https://www.linkedin.com/in/hillary-schlehuber-15197017/">Hillary</a> for the inspiration).</p><p>Here is what it actually is, without the mystique. There is a Chief of Staff who is the one I talk to. Behind the Chief of Staff sits a panel of specialist advisors, each one with a defined lens and a set of questions it always asks. A finance lens that will not let a decision past without asking what the real cost is, including the cost of the road not taken. An architecture lens that asks what is going to be painful to change later. An organizational psychology lens that asks how people are actually going to behave under this decision, not how the org chart says they should. A systems lens that asks what problem we are truly solving, and what the second and third order effects will be after the obvious first one. There are a couple dozen of these in all.</p><p>When I bring it a real decision, the Chief of Staff does not just answer. It figures out which of those advisors belong in the room for this particular question, it puts the question to each of them, it reconciles the places where they disagree, and it hands me back one clear recommendation, along with the distinct angle each lens brought. I get the integrated answer and I get to see the seams. That is a very different thing from a chatbot with an opinion. It is closer to walking out of a good board meeting where every smart person in the company got their say and someone competent synthesized it for you.</p><h3>Why the council beats any single expert</h3><p>Now here is the part that matters most, and it is the part that goes all the way back to the notebook.</p><p>The reason a council of many lenses is so powerful is not that more opinions are automatically better. It is that most hard problems are not actually new. They are old problems wearing a new costume, and the costume is the only reason they look hard. The single most valuable skill I have ever developed is the one from those library Saturdays: stripping a problem down until I can see its bones, and then recognizing, wait, I have seen these bones before, in a completely different field, and over there this was solved a long time ago.</p><p>Let me give you the sharpest example I have. We were staring down a documentation job that was going to eat thousands of hours. Not hard work, exactly, just an enormous amount of it, page after page of documentation that somebody was going to have to assemble by hand, and the team was bracing for the grind. Everyone saw it as a documentation problem, so everyone was reaching for the documentation answer, which was to sit down and start typing and not come up for air for a very long time.</p><p>But it was not really a documentation problem. Strip it down to the bones and it was this: take a fast moving stream of structured data and lay it out into finished, formatted, publishable pages, over and over, on a deadline. And I happened to know that there was a whole industry that had already solved exactly that, decades ago, and gotten very good at it. The news business. Magazines and newspapers have always had to take a flood of incoming content and pour it into laid out pages at a punishing pace, so they built serious tooling to do it. That was the recognition. This is not documentation, this is publishing, and publishing is solved. So instead of typing for two thousand hours, we pointed data at tools built for the newsroom and let them build the documents. The manual task collapsed to a couple of hours.</p><p>Nobody on that team was wrong or slow. They were just looking at it as the kind of problem it appeared to be, instead of the kind of problem it actually was. The whole savings came from one act of recognition, from having enough range to know that the answer was sitting finished in another domain entirely.</p><p>That is what breadth buys you. Not trivia. Pattern recognition across fields, so you can carry a finished solution from one place and set it down in another. It is the difference between staring at an event, the server failed, the project is late, this task will take two thousand hours, and seeing the structure underneath it, the thing that actually produced the event and that you can recognize from somewhere else.</p><p>The trouble is, no one person can carry deep patterns from every field. I cannot. You cannot. And that is exactly why the council is such a force multiplier. Each advisor is a different field’s pattern library. When they all look at your problem at once, the odds shoot up that one of them says, in effect, this is not the exotic thing you think it is, this is that ordinary thing from over here, and we know how that ends. It manufactures, on demand, the recognition that used to depend entirely on whether the one right expert happened to be in your particular room that day. It gives you range you do not personally have.</p><h3>You can have this</h3><p>I did not want to just describe this and leave you wishing. I took my own system, scrubbed everything personal out of it, and made a clean, self starting version you can download for free from GitHub. There is a PDF that walks you all the way through it: setting up your account, installing the skill, getting it running. Once it is installed, it interviews you. It asks about your role, your world, what you are trying to do, and it customizes itself around your answers. Whether you are running a small shop on your own or sitting at the top of a large company, you can walk out of that setup with your own Chief of Staff and your own council of advisors, fitted to your life.</p><p>(Link at the at the end of this piece.)</p><h3>The notebook, forty years on</h3><p>I still think about that kid, out where the roads got bad, writing questions in a notebook all week because Saturday was the only day the answers were reachable. He would not believe what a Tuesday looks like now. He would not believe that the thing he built out of patience and isolation, the habit of learning across everything and trusting that it all connects, is now something you can hand to anyone willing to ask for it.</p><p>That is the whole invitation here. You can use these tools to do your thinking for you, and slowly become that much less able to do it yourself. Or you can use them to become the person who does not need them, the one who got sharper, who learned the subject, who can see the bones of a problem and recognize where it was already solved.</p><p>Build the tutor. Build the council. Build the notebook that fills itself in and then quizzes you on it. And then go be the smartest version of yourself that all of this makes possible, which was always the only point.</p><p>Link to the ELOS on GitHub: <a href="https://github.com/TechHobo/ELOS/tree/main">https://github.com/TechHobo/ELOS/tree/main</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=be18e1eccad0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The case for AI — “Actual Intelligence”]]></title>
            <link>https://medium.com/@t3rrywaltersjr/the-case-for-ai-actual-intelligence-49dc74107682?source=rss-9e132f767efe------2</link>
            <guid isPermaLink="false">https://medium.com/p/49dc74107682</guid>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[systems-thinking]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Terry Walters]]></dc:creator>
            <pubDate>Fri, 14 Aug 2026 22:49:23 GMT</pubDate>
            <atom:updated>2026-08-14T22:49:23.664Z</atom:updated>
            <content:encoded><![CDATA[<p><strong><em>Socrates feared writing would make us stupid. He was half right, and that’s the whole key to using AI.</em></strong></p><p>Sometime around 370 BCE, a philosopher warned that a new technology was going to rot people’s minds. His name was Socrates, and the technology was writing.</p><p>He distrusted it so deeply that he never wrote a word of his own. He did all his thinking out loud, in conversation, poking at people’s ideas until they either held up or fell apart.</p><p>There’s a joke buried in that first fact. Hold onto it.</p><p>To explain his worry, Socrates liked to tell a little story. An Egyptian king, he said, was once offered the gift of writing by one of the gods. The god was proud of it. This will make your people wiser, he promised, and give them better memories. The king told him he had it backwards. This thing you’ve made will breed forgetfulness in the people who use it. They’ll stop exercising their memory. They’ll trust marks made by other people instead of knowing things for themselves. They’ll look wise without being wise.</p><p>Twenty-four centuries later, that is almost word for word what we say about AI.</p><p>Here’s the part we like to skip, because it ruins the smug version of the story: the king wasn’t wrong. Writing did cost us something. The oral cultures that came before it carried whole epics around in their heads, feats of memory most of us couldn’t manage now if our lives depended on it. Once you can write a thing down, you don’t have to hold it anymore. And so, slowly, we stopped holding it. Something real atrophied.</p><p>And yet. The only reason you have heard a word of Socrates’ warning is that he lost. His student Plato didn’t listen, and wrote it all down. Everything we know of the man who refused to write survived for one reason: someone wrote him. His case against writing reached you across twenty-four hundred years riding on the back of the very thing it condemned.</p><p>That was the joke.</p><p>Plato, the student who couldn’t stop writing, gave us what might be the most honest word anyone has ever used for a piece of technology. He called writing a <strong><em>pharmakon</em></strong>, the same word we get “pharmacy” from, and it means both remedy and poison at once. Not good, not bad. Both, depending on the dose and the hand holding it. Writing could make you forgetful, or it could let a single mind reach further than any memory ever could, hold an argument across a thousand pages, and stand on the shoulders of someone dead for centuries. Same invention. The difference was never in the tool. It was in how you used it.</p><p>So when someone tells you AI is going to make us dumber, don’t roll your eyes. They’re half right, and the half they’re right about is the half that matters.</p><p>If you hand your thinking to a machine and take back whatever it gives you, no questions asked, you will get worse at thinking. That isn’t a hypothesis. Use the GPS long enough and you quietly forget how to get anywhere on your own. A tool that will do the work for you is happy to do all of it, including the part where you were supposed to grow. That’s the poison, and it’s real.</p><p>But it’s the same bottle as the remedy, and you are the one holding it.</p><p>Here is the whole mindset, and it’s a small flip that changes everything downstream. Most people reach for AI to get out of thinking. The better move is to reach for it to think harder than you could on your own.</p><p>Don’t ask it to write the thing. Ask it to find the hole in the thing you already wrote. Don’t ask it for the answer. Ask it to argue the strongest case against your answer, until you find out whether you actually believe it. Make it check your logic. Make it name the assumptions you didn’t notice you were making. Make it play the sharp skeptic you don’t happen to have sitting across the table. Used that way, AI stops being a shortcut around your judgment and starts becoming the gym where you train it. Every prompt is a chance to walk away sharper, better informed, and a little less certain of the things you were lazily certain about.</p><p>The output you should care about most was never the essay or the memo or the code. It’s the upgraded version of you that wrote the prompt.</p><p>That’s what I mean by Actual Intelligence. The first thing worth building with artificial intelligence is your own.</p><p>And this matters more than it might sound, because of what it does to the value of range. When you can learn anything, any subject, any skill, at your own pace, with a patient tutor that never once sighs at your questions, the person who only knows one thing very well starts to look strangely fragile. The rare, valuable thing stops being deep specialization and becomes breadth: the judgment to borrow a solution from a field that isn’t yours, the taste to ask the right question, the wisdom to notice when a confident answer is quietly wrong.</p><p>Robert Heinlein said it with more bite than I can manage. After a long list of things he figured a whole human being ought to be able to do (change a diaper, set a bone, write a sonnet, and a dozen more), he landed on four words through his character Lazarus Long: “Specialization is for insects.”</p><p>David Epstein spent a whole book, “<em>Range”</em>, backing that up with the evidence: in the messy, unpredictable domains that actually matter, the generalists tend to win. Breadth beats depth when the world refuses to hold still. And the world is not about to hold still.</p><p>The one skill that compounds in a world where you can learn anything is knowing how to learn, how to think, and how to tell sound reasoning from the confident kind that only sounds sound. AI doesn’t replace that skill. It pays a premium for it.</p><p>Socrates was afraid writing would hollow us out. He was right that it could. He just couldn’t see that the very same invention would carry nearly everything we know from one mind to the next, for the next two thousand years, his own words among them.</p><p>AI is the same kind of thing. It can make you lazy or it can make you formidable, and it will not be the one to decide which. You will.</p><p>I’ve come to think it might be the most powerful tool we have ever built. Not because it can do anything for us, but because, used well, it can help us become people who can do just about anything.</p><p>That’s the “<em>why”</em> The “<em>how”</em>, what this actually looks like when you sit down and build with it, is where I want to go next.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=49dc74107682" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[MALD and Reimagining Project Delivery]]></title>
            <link>https://medium.com/@t3rrywaltersjr/mald-and-reimagining-project-delivery-2280b29bfa6f?source=rss-9e132f767efe------2</link>
            <guid isPermaLink="false">https://medium.com/p/2280b29bfa6f</guid>
            <category><![CDATA[digital-twin-technology]]></category>
            <category><![CDATA[transportation]]></category>
            <category><![CDATA[infrastructure]]></category>
            <category><![CDATA[cadd]]></category>
            <category><![CDATA[bim]]></category>
            <dc:creator><![CDATA[Terry Walters]]></dc:creator>
            <pubDate>Thu, 26 Oct 2023 19:58:47 GMT</pubDate>
            <atom:updated>2023-10-26T19:58:47.992Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8ioOmX6Hxrs1IgUi16vtKQ.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@shawnanggg?utm_content=creditCopyText&amp;utm_medium=referral&amp;utm_source=unsplash">shawnanggg</a> on <a href="https://unsplash.com/photos/time-lapse-photo-of-concrete-highway-with-cars-vdBE638sszE?utm_content=creditCopyText&amp;utm_medium=referral&amp;utm_source=unsplash">Unsplash</a></figcaption></figure><p>In the ever-evolving landscape of infrastructure development, a new concept has emerged, one that promises to reshape the way we conceive and execute construction projects. It goes by the acronym MALD, which stands for “Model as the Legal Document.” At its core, MALD seeks to replace the conventional contract documents and plan sheets with data-rich 3D Building Information Models (BIM). These BIM models empower contractors to directly reference design intent and utilize surface model data for tasks such as earthwork estimation, staging, and Automated Machine Guidance (AMG) output.</p><p>The impact of MALD on the ACEC (American Council of Engineering Companies) community is nothing short of revolutionary. It promises to reshape the way projects are conceived, designed, and communicated to contractors, ushering in a new era of efficiency and precision.</p><p>In this series we will delve into digital delivery and MALD, from three key perspectives: that of the agency owner, the designer/engineer, and the contractor. In this first installment, we will explore MALD from the agency’s viewpoint, shedding light on what it means in terms of project delivery, the challenges it presents, and the ultimate benefits reaped by agencies who embark on this transformative journey.</p><p><strong>MALD — The Agency Perspective</strong></p><p>At its core, MALD necessitates a profound transformation in how agencies and the consultants supporting them approach project delivery. While some states have long been accustomed to working with 3D models; often, these 3D models are incomplete, serving only as a means to generate traditional plans that don’t capture every intricacy of a design.</p><p>The shift to MALD compels agencies to either learn the art of modeling (as is the case with states that historically used Bentley’s Geopak) or to elevate their current modeling skills to improve accuracy, precision, and comprehensiveness. This transition isn’t just about users learning new skills; it also entails reevaluating the underlying workspaces, state kits, and other tools that support these activities. Training initiatives, process streamlining, and workspace enhancement are all integral aspects of this transformation.</p><p>The impact of MALD extends well beyond the initial learning phase. As project delivery methods evolve rapidly, designers and reviewers must adapt to new software tools. Over the next three years, we can anticipate significant advancements in automated and manual design model reviews, reducing the reliance on paper or PDF plans. Commenting and discussion threads from review sessions will be seamlessly integrated into the model environment, enhancing traceability. These developments, combined with detailed models and cutting-edge review tools, promise to enhance design quality and reduce project costs. Recent pilot projects spanning the last six years have already demonstrated the cost-saving potential of improved modeling, review, and digital delivery methods. Notably, digital twin technologies offered by Bentley Systems, Nemetschek, Trimble, and Autodesk are aligning to supercharge design review and digital project delivery.</p><p>How are government agencies adapting to this new approach to project delivery? Utah, Pennsylvania, and Texas Departments of Transportation have developed highly effective templates for implementing this programmatic shift. These agencies have embraced holistic programmatic strategies that encompass many, if not all the following components:</p><ul><li>Thorough Investigation and Program “gap” analysis</li><li>Extensive Collaboration with Stakeholders</li><li>Detailed Strategic Planning</li><li>Establishment of Guidelines, Policies, and Procedures</li><li>Development of New Standards (i.e. CADD Standards, Model Development Standards (MDS))</li><li>Asset Management Standards</li><li>Implementation of Pilot Projects</li><li>Digital Delivery Implementation Planning</li><li>Formulation of a Training Program</li><li>Data Architecture and Governance Planning</li><li>Creation of a Digital Twin Plan</li></ul><p>The emergence of MALD, or “Model as the Legal Document,” is ushering in a transformative era in infrastructure development. This innovative concept, driven by 3D Building Information Models (BIM), promises to revolutionize how construction projects are conceived and executed. Its impact on the ACEC community is profound, introducing newfound efficiency and precision into project delivery.</p><p>From the agency’s perspective, embracing MALD necessitates a significant shift in approach. Agencies and their consultants must adapt by either acquiring new modeling skills or enhancing existing ones. This transition is not just about skill acquisition but also entails a reevaluation of the tools and workspaces that support these activities. Training, process optimization, and workspace enhancement are integral components of this transformation.</p><p>The ripple effect of MALD extends beyond the learning phase. It paves the way for significant advancements in design model reviews, reducing reliance on traditional plans. Commenting and discussion threads integrated into the model environment promise to enhance traceability and overall design quality while reducing project costs.</p><p>Government agencies, such as the Utah, Pennsylvania, and Texas Departments of Transportation, are leading the charge in embracing this new model-based project delivery approach. They have outlined comprehensive programmatic strategies that encompass various key components to maximize the benefits of this transition. With unwavering support from top-level leadership and engagement with stakeholders, these agencies are poised to shape the future of infrastructure development.</p><p>In the upcoming articles in this series, we will delve deeper into MALD, exploring its implications for designers and contractors, unveiling the transformative power it holds in revolutionizing the infrastructure landscape.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2280b29bfa6f" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>