Now I Get It, with Dr. Andy

I've been thinking a lot lately about AI, a subject I explored back in grad school and postdoc. In this episode, I walk through a hackathon project to democratize AI by distributing neural network training across home computers. I explain why it failed: first the obvious reason, a bad actor can silently poison a distributed model, then a subtler one rooted in the physics of computation. I break down how neural networks work, why a single nonlinearity lets them approximate almost anything, and why these models need the low latency bandwidth only a data center can provide.From there I connect to something bigger: why AI is destined to concentrate power into a few organizations, and why that oligopoly could tip into an outright monopoly. I use the mathematics of pandemic spread to show how a tiny edge in AI self improvement compounds into a runaway advantage, and I compare it to UK land ownership to explain what it means to be a rent seeker over an entire economy. I close with why every business is now a software business, why AI generated code brings new unintended behavior, and why we need to stay engaged rather than assume it's someone else's problem.In this episode you will learn:(00:00) The hackathon idea that started this, distributing AI training across home computers so everyone could have a stake, and why it failed, starting with undetectable poisoning by bad actors.(03:44) How neural networks work, from linear functions in algebra to the single nonlinearity that lets them approximate virtually any function.(06:36) The matrix computations and graphics chips these models need, and why low latency communication means the work must happen inside one data center, not distributed machines.(10:00) Why AI is destined to be an oligopoly controlled by a handful of corporations and nation states, and how AI improving AI can turn a breakout model into a runaway one.(12:57) The mathematics of pandemic spread, the replication number separating exponential growth from decline, and how a small edge in AI self improvement compounds into an overwhelming advantage.(15:56) The idea of a rent seeker, illustrated through UK land ownership dating back nearly a thousand years, and how a runaway AI could end up in a similar position over the global economy.(18:53) Dan Kennedy's insight that whoever can spend the most to acquire a customer wins, and why that means every business is now a software business built on AI.(21:32) The unintended behaviors in AI generated software, and why engaging with these risks now matters more than treating them as a distant problem.Let’s connect!linktr.ee/drprandyRead, with computer assistance, the best literature the world has ever producedandrew-winkler-reading-room.overskill.appLove Quotient: Stop Dying of Thirst in an Ocean of Love Kickstarterhttps://www.kickstarter.com/projects/amwphd/love-quotient

I'm taking you through my own origin story this episode, from getting swept into a Johns Hopkins study on intellectual talent as a sixth grader to skipping straight into college at fifteen. I share how that unusual path, being identified as gifted, accelerated through school, and landing in a college honors program that clashed with my own academic ambitions, shaped the way I think about learning today. Along the way I talk about my late AP English teacher Adrian Morris, and how I completely missed the point of literature even while being taught by the best.That early misstep, wanting to read everything in its original language rather than in translation, ended up defining my life's work. I explain why I built reader.winklers.us, my free computer assisted reading tool with over a hundred classic texts, and why I deliberately left out syntax support so readers still get the meditative wrestling that makes literature transformative. I also dig into the psychology behind why this matters: how literature builds theory of mind, how naming our emotions helps us process implicit memories, and why untranslatable words like Schadenfreude and saudade reveal blind spots in our own emotional vocabulary.In this episode you will learn:(00:00) How a Johns Hopkins study on intellectual talent led me from taking the SAT in sixth grade to skipping into college at fifteen(02:43) Why my college honors program wanted humanities depth I didn't have, and how I vastly overestimated my ability to learn languages fast enough to read literature in the original(04:42) What inspired me to build reader.winklers.us, a free tool that handles the tedious dictionary work while deliberately leaving syntax to the reader for a deeper, more meditative experience(07:09) How literature builds theory of mind and why that matters for long term relationships, a concept I explore further in my book Love Quotient: Stop Dying of Thirst in an Ocean of Love(08:00) What "name it to tame it" means, and how explicit versus implicit memory processing shapes our emotional coping(09:35) Why words like Schadenfreude, Waldeinsamkeit, and saudade reveal the blind spots baked into any single language or cultureLet’s connect!linktr.ee/drprandyRead, with computer assistance, the best literature the world has ever producedandrew-winkler-reading-room.overskill.appLove Quotient: Stop Dying of Thirst in an Ocean of Love Kickstarterhttps://www.kickstarter.com/projects/amwphd/love-quotient

I open this episode with a story from my childhood: a dam that broke above my uncle's farm, and two California kids who survived the flash flood only because they knew how to surf a wave instead of fighting it. That's exactly how I think about AI right now. It's a massive wave bearing down on all of us, and whether it becomes a wonderful tool or something that crushes us depends entirely on whether we learn to ride it. I walk through what happened when I went from writing off AI-generated code as "laughable" last fall to building real, deployable tools with it today, including the Andrew Winkler Reading Room, a free app that lets you read over a hundred of the world's greatest books in their original languages, from Homer's Greek to modern Japanese.In the second half, I get honest about what it's actually like working with AI day to day. I talk about building "Gentle Reader," a feature that lets language learners dive into real literature with only 70 to 75% vocabulary instead of waiting until they hit 95%, and the surprising detective work it took to get ancient and modern Japanese texts working properly. I also share where AI still struggles: its short-term memory lapses, its tendency to patch symptoms instead of solving root problems, and what my ongoing project to restore Don Quixote from a centuries-old scan taught me about "wisdom of crowds" and a 250-year-old quote from Sir Joshua Reynolds that turned out to apply just as much to machines as it does to us.In this episode you will learn:(00:00) Why I compare AI to a wave that can either be surfed or can crush you, using a story about a dam breaking above my uncle's farm(03:14) How the Andrew Winkler Reading Room lets you read the Odyssey and other classics in their original Greek, with every word's form, meaning, and pronunciation just a click away(06:06) What "Gentle Reader" is and why lowering the vocabulary threshold from 95% to 70-75% makes it possible to start reading real literature much sooner(09:04) Why I chose books like Wizard of Oz, Aesop's Fables in Chinese, and Nicholas Nickleby in Italian to fill gaps where no natural "great work" existed at the right reading level(12:01) What made Japanese the hardest language to build, including the challenge of AI handling ancient versus modern Japanese grammar and meaning shifts(14:54) Why AI models have what I call a "short-term memory slash auditory processing disorder," and how the "Attention Is All You Need" paper changed the game but didn't fully solve it(17:48) What my Don Quixote restoration project, removing decades of ink bleed and yellowing, then building a font from "wisdom of crowds," taught me about AI's tendency to avoid the real labor of thinkingLet’s connect!linktr.ee/drprandyRead, with computer assistance, the best literature the world has ever producedandrew-winkler-reading-room.overskill.appLove Quotient: Stop Dying of Thirst in an Ocean of Love Kickstarterhttps://www.kickstarter.com/projects/amwphd/love-quotient

I'm digging into something that's genuinely changed how I spend my free time: using AI to build the tool I've wished existed for decades, one that helps me read the world's great literature in its original language. I tell the story of a chance dinner with a Hertz Fellow who turned out to be a major figure at Anthropic, and how that connection eventually led me back to a project I'd tried and abandoned before, a computer-assisted reader for Greek, Latin, Hebrew, and Aramaic texts. Six months ago, the AI models I tried couldn't get past writing a single working line of code. This time was different, and I walk through exactly what changed, what these models are actually good at, and where they still fall apart.I also get into the weeds a bit, sharing my process for getting better results out of these tools, why my public writing on Quora turns out to matter, and a wild multi-round debugging session with Gemini that kept insisting it had "finally" found all the bugs. I close with a personal aside into a physics insight I've been chasing since the 1970s, and how an AI helped me push it further, even if I lost most of the conversation to a very avoidable software glitch. Along the way I make my case for why AI right now is best thought of as a tireless, brilliant, but occasionally hallucinating army of interns, not a replacement for human judgment.In this episode you will learn:(00:00) How a chance dinner with a Hertz Fellow connected to Anthropic changed my skepticism about the current state of AI(03:37) Why I've always wished a computer could do the tedious lexical lookup work of reading Greek, Latin, Hebrew, and Aramaic texts(06:35) What large language models actually are under the hood, and why they can "spout like an idiot" and "pontificate like a sage" with equal ease(08:57) The multi-round debugging saga where Gemini kept insisting it had found the last bug, and the last bug, and the last bug(11:48) What gets lost in translation, using the hidden joke behind Don Quixote's horse Rocinante as an example(14:43) Inside the Andrew Winkler Reading Room, from the Septuagint to Shakespeare, and why computers and humans make a great reading team(17:10) My feature requests for Google and Apple after losing valuable AI conversations to a silent clipboard failure(19:36) Why I think of AI right now as an army of knowledgeable but wisdom-less interns, and what that means for companies rushing to replace workers with it(22:01) The physics insight I've been sitting on since the 1970s, and how AI helped me extend it into the standard model(24:28) What was lost when my breakthrough AI conversation on unit-free physics vanished, and why I'm sharing the story anywayLet’s connect!linktr.ee/drprandyRead, with computer assistance, the best literature the world has ever producedandrew-winkler-reading-room.overskill.appLove Quotient: Stop Dying of Thirst in an Ocean of Love Kickstarterhttps://www.kickstarter.com/projects/amwphd/love-quotient

Today I'm talking about something that's been on my mind as I've been writing Love Quotient: the two feedback loops that quietly corrode almost every relationship we have. I explain how our dominant cognitive function becomes our personal "channel" for expressing and receiving love, and why a mismatch with the people we care about can spiral into the exact same withdrawal on both sides, until the relationship falls apart.From there I dive into Pride and Prejudice, using what I know about personality and temperament to offer a fresh, more sympathetic read on Mr. Bennett and the entailment that hangs over the Bennet family. I also share a personal story about my border collies that mirrors his situation surprisingly well. Then I take a hard turn into artificial intelligence, walking through how I used "vibe coding" to build a free tool that helps you read the world's great literature, in the original Greek, Latin, Hebrew, and Aramaic, with computer-assisted definitions built right in.In this episode you will learn:(00:00) Why our relationships are being steadily corroded by two feedback loops we rarely notice(02:27) How our dominant cognitive function becomes our personal channel for expressing and receiving love(05:15) What entailment meant for the Bennet family and why it set the entire plot of Pride and Prejudice in motion(08:22) Why Mrs. Bennett was so openly focused on marrying off her daughters to rich men(11:57) How "goodness of fit" and temperament theory give us a more sympathetic read on Mr. Bennett's parenting(15:32) What my border collies taught me about Mr. Bennett's distance from his younger daughters(18:45) Why artificial intelligence is simultaneously overhyped and underhyped right now(18:45) How human and machine intelligence complement each other when it comes to parsing language(22:01) How I used "vibe coding" to build a reading tool from scratch in a single day(24:45) What's inside the Andrew Winkler Reading Room, from the Septuagint to Shakespeare to UlyssesLet’s connect!linktr.ee/drprandyRead, with computer assistance, the best literature the world has ever producedandrew-winkler-reading-room.overskill.appLove Quotient: Stop Dying of Thirst in an Ocean of Love Kickstarterhttps://www.kickstarter.com/projects/amwphd/love-quotient

Today I'm tackling something a little different: taxes. After spending years writing my book Love Quotient and other projects in physics and math, I funded that work through earnings in the tech sector, specifically long-term capital gains. That led me into a maze of federal versus California tax treatment, amended returns, and a frustrating "screw you letter" from the IRS demanding paperwork they could have processed themselves if they'd just shared their own software with taxpayers. I talk about why that doesn't happen, who benefits from keeping it that way, and the surprising silver lining: California's tax agency was actually reasonable to deal with, which genuinely surprised me.Then I shift gears completely into something I've been working on for the past year: a new way of understanding scalar and vector quantities in physics. I unpack why "scalars" like mass, temperature, and length aren't really scalars at all, but one-dimensional vectors tied to units, and how units themselves function as bases for these vector spaces. From there, I show how ordinary vector quantities like force and velocity are tensor products of geometric vectors with these one-dimensional magnitude spaces, and how this framework recovers the classic magnitude-and-direction picture of physics in a cleaner, more unified way.In this episode you will learn:(00:00) Why I ended up filing amended tax returns after years of writing my book and research projects(03:43) Why government tax agencies don't give taxpayers access to their own processing software(04:18) How political incentives keep tax prep companies like H&R Block and Liberty Tax in business(06:00) Why "scalar" quantities like mass and temperature aren't actually scalars(09:27) How units function as bases for one-dimensional vector spaces(10:00) Why changing units works exactly like a change of basis for vectors(10:47) How to recover both magnitude and direction from this unified frameworkLet’s connect!linktr.ee/drprandy

In today's episode, I'm sharing the core insights from my new book, Love Quotient: Stop Dying of Thirst in an Ocean of Love — and my goal is to give you everything you need to transform your most important relationships.Welcome to Now I Get It with Dr. Andy. I'm talking about one of the most painful and preventable dynamics in relationships: the slow drift that happens when two people are pouring love into a connection, but neither one can feel it. The culprit isn't a lack of caring — it's a mismatch in cognitive functions. I walk through the four core ways we process the world — sensing, thinking, feeling, and intuition — and why we each only mature some of these functions while others stay dormant. That gap is precisely where love gets lost.Tune in as I explore how to identify your loved one's dominant cognitive type through something as simple as their gestures or walk, and how to bridge the gap between the love you're giving and the love they actually feel. Whether you're navigating a marriage, a friendship, or a professional dynamic, this episode gives you a practical framework for turning a drifting relationship into one that deepens every day.In this episode, you will learn:(00:27) We only fully mature half of our cognitive functions in a lifetime(05:09) Most relationships break down because love is sent in a form the other can't perceive(07:45) Each cognitive function has its own unique language of love (08:30) Four hand gestures reveal a person's dominant cognitive type(13:00) The way someone walks maps to their relationship dynamic (19:54) Combining gesture and walk pinpoints exactly what your loved one needs (22:10) Attuning to your loved one triggers them to give you the love you need in returnLet’s connect!linktr.ee/drprandy

What if artificial intelligence doesn't replace human intelligence — it amplifies it? And what if the quality of what you bring to AI is exactly what determines what you get back?Welcome to Now I Get It with Dr. Andy. I'm Andrew Winkler, and in this episode I'm taking a deep dive into one of the most consequential technologies of our time: large language models. I break down how these systems are built on surprisingly elegant mathematics, why language itself has a hidden statistical structure that makes AI possible, and what it really means for how we interact with these powerful tools.Tune in as I explore the neural network foundations that underpin modern AI, unpack the "garbage in, garbage out" principle in its most precise form, and reveal why the most important thing you can bring to an AI conversation is your own intelligence and curiosity.In this episode, you will learn:(00:27) Neural networks are built on elegant mathematics(01:15) One nonlinearity unlocks AI's power to model anything(02:47) Models extract signal, not just memorize data(04:30) Language has a hidden statistical structure AI can learn(08:30) AI defaults to average intelligence without strong context(09:03) Smarter input produces smarter AI output(09:45) AI amplifies human intelligence — it doesn't replace itLet’s connect!linktr.ee/drprandy

Welcome back to Now I Get It with Dr. Andy. I'm sitting down once again with Angela McEwen, our resident expert in early childhood development, to dig into one of the most overlooked secrets of raising healthy, capable kids — proprioception. Angela has spent decades in childcare, including helping coordinate San Francisco's childcare response during the pandemic, and what she's discovered about the role of natural play environments is something every parent and educator needs to hear.In this conversation, Angela shares the remarkable results from a nature-based outdoor redesign pilot program at her San Francisco preschool — and what happened blew even her away. For the first time in her career, children developed proprioceptive skills entirely on their own, without any formal instruction, simply by playing with logs, mud, and the natural world around them. We also get into why full-body sensory experiences — from jumping into pools to rolling in the dirt — matter more than flashcards or structured fine motor activities, and how giving kids a little controlled risk teaches them to trust themselves for a lifetime.In this episode, you will learn:(01:01) How animal movement games build proprioception in young children(02:54) Why rushing kids into formal schooling before age seven backfires(05:30) The nature-based playground redesign — and its surprising results(09:49) The Olympian study: why trampolines and full-body impact matter(14:57) Why peeling bark and picking berries beat fine motor activities(19:59) Why a little controlled stress builds lifelong resilienceLet’s connect!linktr.ee/drprandy