# Alan Scott Encinas

> Machine Learning, Cognitive Systems, and Real-World Deployment


## Posts

- [I Added Search to My Agent and It Got Worse, and That Cost Me Nothing](https://alanscottencinas.com/search-made-it-worse/): I wrapped search around a working agent and every time I let the search overrule it, it played worse. The experiment was built so a negative result cost zero rating, which is why I could afford to find out.
- [The Folder Name That Would Have Lost Me Every Single Game](https://alanscottencinas.com/kaggle-agent-import-error/): I named a package agents. So had the platform. Its loader appends to the path instead of prepending, so its module wins, mine never loads, and every match is a loss. Local tests could not see it, for two separate reasons.
- [My Score Said None for Three Weeks and I Believed a Number I Made Up](https://alanscottencinas.com/pokemon-tcg-ai-rating/): For three weeks I thought my agent was rated around 709. It was 784.9. The number had been unavailable the entire time and my own code had quietly replaced it with a guess I stopped questioning.
- [My Pokemon TCG Agent Won 95% of Its Games and I Learned Nothing](https://alanscottencinas.com/pokemon-tcg-ai-agent/): I built a clean general search agent and it beat every opponent I had. Then I pointed it at a 400-line hand-written bot and lost three games in four. The opponents I had been measuring against were the problem.
- [I Entered the Pokemon Trading Card AI Battle Challenge](https://alanscottencinas.com/pokemon-tcg-ai-battle-challenge/): The Pokemon Company put the Trading Card Game up as an AI problem, and I entered. Here is what the competition actually is, why this game is harder for a machine than chess, and what you are going to see over the next few weeks.
- [The AI Reset](https://alanscottencinas.com/ai-as-a-crutch/): Anthropic started watermarking Claude's text and people reacted like someone had announced the end of AI. I think they are asking the wrong question. The better one is what we became while AI was becoming normal.
- [Four Months, and the Thing That Would Have Fixed It Took an Afternoon](https://alanscottencinas.com/kaggle-competition-post-mortem/): This is where the final score lands, and what it says about four months of work. I finished four tenths of a foot from a medal, and everything I built in the second half was worth less than that gap.
- [The Asset With No Line Item](https://alanscottencinas.com/key-person-dependency/): A consultant sold us a six page strategy report that was generated, not written. The interesting question was never why he failed. It was why the company was so confident about what it was buying and so unclear about what it already had.
- [I Was Right About the Risk and My Hedge Was Useless Anyway](https://alanscottencinas.com/kaggle-final-submission-selection/): Six weeks before the deadline I worked out exactly how my models would fail, and built a hedge against it. The hedge was drawn from the same family as the thing it was hedging. Here is the question I should have asked instead.
- [My Metric Was Valid. It Just Stopped Putting Things In Order.](https://alanscottencinas.com/kaggle-leaderboard-shakeup/): For two months I optimized a score that had stopped ranking my models. It was still a valid measure of quality. It had just gone blind at the resolution I was working at, and three lines of code would have told me.
- [A Stranger's Blog Post Told Me I Was 900 Places Wrong](https://alanscottencinas.com/kaggle-private-leaderboard-api/): The competition closed and I pulled my final rank. The number was internally consistent, completely plausible, and wrong by nine hundred places. Here is the two-word API difference that caused it and how a competitor's blog post caught it.
- [From Regolith to Revolution: How AI and In-Situ Manufacturing Could Build Humanity’s First Lunar City](https://alanscottencinas.com/building-on-the-moon/): Building on the Moon means making what you need from the dust already underfoot. A permanent settlement cannot import every beam and panel from Earth, and the materials to do it may be designed by AI before anyone builds them.
- [Three Times I Measured Nothing](https://alanscottencinas.com/three-times-i-measured-nothing/): I pointed a 22-agent review at my own competition endgame expecting notes about the code. It came back with three errors in my reasoning instead, and all three were experiments that could only ever confirm what I already believed.
- [Your AI Sounds Certain. Look at Where It Learned That.](https://alanscottencinas.com/ai-training-data/): These systems learn from the open internet, and the open internet is getting worse on purpose. The quality of what goes in is falling while public trust in what comes out keeps climbing. Nobody is watching that gap.
- [The New Fraud Economy: When Trust Becomes the Attack Surface](https://alanscottencinas.com/business-impersonation-fraud/): Someone used our business information to make a fraudulent transaction look legitimate. No malware, no stolen passwords, no alert. They studied how the process worked and walked through the front door of trust.
- [Everyone Knows It Scores Half. Nobody Checks Which Half.](https://alanscottencinas.com/which-half-the-leaderboard-scores/): The public leaderboard scores half the test set and says so on the page. I measured which half. It was not a random sample, it was the first 1,974,995 rows in time order, and that one fact invalidated six weeks of tuning.
- [Your AI Safety Wall Checks One Thing. The Attacker Changes the Other.](https://alanscottencinas.com/prompt-injection-defense/): Moving AI authorization outside the model was the right decision. But a security boundary only protects what someone wrote down, and last week OpenAI's own models proved it by breaking into Hugging Face to cheat on a test.
- [Every Knob My Validation Picked Was Too Timid](https://alanscottencinas.com/validation-picked-too-timid/): I spent a session building a temporal convolutional network to squeeze more signal out of Martian pressure residuals. It moved the leaderboard about 1.3. Then a single constant, found with five submissions and a parabola, moved it 5.4. Here is why my cross-validation never saw either coming.
- [I Took First Place Without Training a Model](https://alanscottencinas.com/gradient-boosting-extrapolation/): The competition asked me to predict Martian air pressure on days the rover never recorded. My first submission took the top of the leaderboard and contained no machine learning at all. Here is what it did instead, and the number that told me it was still wrong.
- [From Moon Craters to Martian Dust](https://alanscottencinas.com/from-moon-craters-to-martian-dust/): How one project quietly led to another. Finishing LunarSite changed how I watched The Martian, and that curiosity turned into a machine learning project on the Martian environment.
- [New Tariffs Cover 99.4% of U.S. Imports. Your Plan Doesn't.](https://alanscottencinas.com/supply-chain-disruption-planning/): Three tariff actions in four days, war risk premiums up tenfold, and freight rates falling at the same time. The assumptions companies plan against are now expiring faster than the planning cycles built to use them.
- [What Drilling Blind Taught Me About Machine Learning](https://alanscottencinas.com/drilling-blind-lessons/): Five lessons from a machine-learning competition where you predict the depth of rock you cannot see. Almost none of them are about models. A summary of the arc so far, and the through-line underneath it.
- [Every System Is About to Get a Guard](https://alanscottencinas.com/ai-attacking-ai/): Autonomous agents are already running attacks at machine speed, and the only thing fast enough to answer them is another agent. Defensive AI is about to stop being a product category and become a layer, the way antivirus did.
- [The Green Test Suite That Hid a Broken Model](https://alanscottencinas.com/green-tests-broken-model/): A passing test suite is the most reassuring lie in software. Three times in one competition, green checkmarks certified a broken model, each one lying in a different way. Here is how, with code.
- [How I Killed Sixteen Ideas Without Wasting a Submission](https://alanscottencinas.com/killed-sixteen-ideas/): In a competition where every experiment costs hours and submissions are rationed, the highest-leverage skill is not a better model. It is a rule written down in advance that lets a cheap probe tell you not to build the thing at all.
- [Why Your Dashboard Is Lying to You (And What Actually Works)](https://alanscottencinas.com/the-end-of-optimization/): Efficiency won when the world was stable. It no longer is. Why competitive advantage is shifting from optimization to operational intelligence, and how small teams now build sensing systems that used to require a department.
- [The Validation Score That Lied to Me for a Month](https://alanscottencinas.com/validation-score-that-lied/): For a month my local validation and the competition leaderboard disagreed about whether I was improving at all. They were computing two different statistics. Here is how I found out, and the one-submission experiment that fixed it.
- [What Teaching a Machine to Think Taught Me](https://alanscottencinas.com/teaching-a-machine-to-think/): A summary of the arc so far from inside the ARC Prize 2026: the three failures that taught me how the competition actually works, and why the plain agent still beats every cleverer thing I have built. Almost none of it was about the model.
- [What Is a Cognitive System? Software That Pursues Goals, Not Instructions](https://alanscottencinas.com/cognitive-system/): For fifty years software executed instructions. Cognitive systems pursue objectives instead: they perceive, plan, act, verify, and adapt under uncertainty. Here is what actually separates a cognitive system from a traditional application, and from bolting an LLM onto one.
- [The bug was in my beliefs, not my code](https://alanscottencinas.com/the-bug-was-in-my-beliefs/): An ARC Prize 2026 builder's log entry. I spent two work sessions reasoning from two "facts" about my own code that were both false. Checking them against the artifact took an hour. The highest-risk sentence in any project is "we can't, because."
- [The Token Tax: Why Bad AI Architecture Is Becoming Enterprise Debt](https://alanscottencinas.com/the-token-tax/): Inference got dramatically cheaper, yet enterprise AI bills keep climbing. The token tax is not a pricing problem, it is an architecture problem, and the missing middle layer is quietly becoming enterprise debt.
- [What Is Cognitive Orchestration? Why AI Systems Can't Be Built Like Traditional Software](https://alanscottencinas.com/cognitive-orchestration/): Most AI systems fail not because the model is weak but because we wrap probabilistic cognition in architecture built for deterministic machines. Cognitive orchestration is the layer that manages the uncertainty: synthesized state, adaptive planning, verification, and persistent memory.
- [I made my agent more capable and it got worse](https://alanscottencinas.com/more-capable-and-it-got-worse/): An ARC Prize 2026 builder's log entry. I taught my agent to solve more kinds of games and the leaderboard score fell from 0.09 to 0.04. The capability was real. The cost of using it was the problem. More powerful is not better.
- [The Paintbrush Paradox: Why the Monolithic Era of AI Is Crumbling](https://alanscottencinas.com/paintbrush-paradox/): Bigger context windows were supposed to solve AI. Instead they exposed the real bottleneck: not the model, but the architecture around it. Why the monolithic era of AI is giving way to orchestration.
- [I looked unbeatable in practice and lost the real fight](https://alanscottencinas.com/unbeatable-in-practice-lost-the-real-fight/): An ARC Prize 2026 builder's log entry. A change that gained on my home bench lost three times as much on the real, hidden leaderboard, and my own tooling couldn't even reproduce the failure. Why the leaderboard is the only oracle.
- [The AI Boomerang: They Laid You Off to Save Money. They're Paying More to Get You Back.](https://alanscottencinas.com/the-ai-boomerang/): Companies that cut jobs for AI are quietly rehiring, and most of them didn't even save money doing it. The AI boomerang, and what it reveals about the real price of human judgment.
- [I'm trying to teach a machine to play a game it has never seen](https://alanscottencinas.com/teaching-a-machine-to-learn-a-new-game/): A new thread in the builder's log: I'm competing in the ARC Prize 2026 (ARC-AGI-3), the benchmark built to measure the one thing today's AI is worst at, learning something it was never trained for. What the comp is, why it's the field's uncomfortable mirror, what it would mean to crack it, and where I'm standing.
- [The Infrastructure Fallacy: Why Most Sales Problems Are Actually Strategy Problems](https://alanscottencinas.com/the-infrastructure-fallacy/): Most companies think they have a sales problem. They have a strategy problem in disguise. A decade of auditing startups and OEM manufacturers, and the pattern is always the same: activity went up, revenue did not, because no system connected market demand to the sale.
- [The Scale Wall: Why AI Gurus Are Building Toys While the World Needs Architects](https://alanscottencinas.com/why-ai-projects-fail-at-scale/): The influencer economy sells flashy AI toys while real systems shatter at scale. Notes from six months of auditing broken pipelines, the 10-level hierarchy that separates a demo from an engine, and what real AI engineering actually takes.
- [I was sure, and I was wrong](https://alanscottencinas.com/i-was-sure-and-i-was-wrong/): The sixth entry in a live builder's log. The contribution I was certain would be the centerpiece lost to the cruder approach already in the pipeline. I published the loss, because complexity has to earn its place every time.
- [I went hunting in light the eye can't see, and the models were the easy part](https://alanscottencinas.com/the-models-were-the-easy-part/): A deeper Builder Journal entry from inside the Hyperspectral Object Tracking Challenge 2026, with code. Before a single model could matter I had to build a local scorer for an unlabeled test set and beat a stale submission template. The plumbing is half the competition.
- [The win that was a coin flip](https://alanscottencinas.com/the-win-that-was-a-coin-flip/): The fifth entry in a live builder's log. A parameter sweep handed me a new best score. Three red flags said it was noise wearing a knob's clothing, so I refused to ship it. Most of competitive ML is declining to fool yourself.
- [My own test set lied to me](https://alanscottencinas.com/my-own-test-set-lied-to-me/): The fourth entry in a live builder's log. I built a faithful local scorer, then fed it a biased test set, and it confidently signed off on a change that made me worse on the real leaderboard. A dev set built from failure cases certifies rescues, not deployments.
- [The Search for El Dorado: How Traditional Data Science Unlocked the Hidden Markets of Latin America](https://alanscottencinas.com/el-dorado-latin-american-markets/): Expanding into Latin America is not the UK game of branding. Here is how traditional data science found the hidden, government-backed markets the incumbents never saw, and turned a sub-$1,000 trickle into a network across two continents.
- [My submission was rejected before it ever scored](https://alanscottencinas.com/submission-rejected-before-it-scored/): The third entry in a live builder's log. My first Kaggle submission was rejected before it ever scored, thanks to a stale sample_submission file and a Google Drive download limit. The plumbing turns out to be half the competition.
- [The 90-Day Asymmetric Playbook: Scaling Cross-Border Tech with Zero Upfront Capital](https://alanscottencinas.com/asymmetric-market-entry-playbook/): The traditional international playbook says global expansion needs millions in capital and a big team. Here is how a two-person venture scaled across the UK and EU to a high revenue run rate in 90 days with zero upfront capital, and an AI intelligence system that cost under $9.
- [The leaderboard you can't score against](https://alanscottencinas.com/leaderboard-you-cant-score-against/): The second entry in a live builder's log. The Hyperspectral Object Tracking Challenge scores you on an unlabeled test set, so the leaderboard can't tell you your own score. Here's how I built a local scorer byte-faithful to the official one, and why measurement is half the battle.
- [I entered a competition to track objects in light you can't see](https://alanscottencinas.com/hyperspectral-object-tracking-challenge/): The first entry in a live builder's log. I'm competing in the Hyperspectral Object Tracking Challenge 2026: track one object through video shot in colors the human eye was never built to see. Here's what the problem is, why it's hard, and where I'm starting on the leaderboard.
- [The Unknwn Player](https://alanscottencinas.com/the-unknwn-player/): Seven unrelated AI competitions, one person, at the same time. The framework behind it was not built in a lecture hall. It was built by a broke teenager who could not stop losing at League of Legends.
- [Building LunarSite: the director's commentary](https://alanscottencinas.com/building-lunarsite-directors-commentary/): The results piece has the metrics table. This is the other half: the decisions, the forks I parked, the harness I built before training anything, and the negative result I published so nobody re-runs it. Director's commentary on building a solo ML pipeline.
- [I spent a few months teaching a laptop to read the Moon](https://alanscottencinas.com/teaching-a-laptop-to-read-the-moon/): In March 2025 a lander reached the lunar south pole and tipped over. That gap between the most valuable ground in the solar system and the hardest to read is the thing I couldn't put down, so I spent a few months teaching a single laptop to scout it.
- [Companies Replaced Their Workers With AI. Now They're Hiring Them Back to Babysit It.](https://alanscottencinas.com/the-great-ai-rehiring/): Everyone fears AI will replace them. The data shows the opposite happening: the companies that cut jobs for AI are quietly hiring people back, often to supervise the AI. Here is why it's breaking, with the receipts.
- [My 21 AI Agents Aren't Allowed to Talk to Each Other. That's Why It Works.](https://alanscottencinas.com/multi-agent-ai-system-architecture/): I built a multi-agent AI system of 21 specialized agents that never talk to each other, run for under $20 total, and have never fallen over. Here is the architecture, and why every strange decision was the right one.
- [I Built an AI Sales Team for Under $1 a Day. It Closed $29,201 and Beat Salesforce and HubSpot's AI](https://alanscottencinas.com/meet-albert-multi-agent-sales-force/): I built an AI sales team for under $1 a day. In 42 days it cost $15.85, closed $29,201, and beat Salesforce and HubSpot's AI at finding and working real leads. Here is how it works, and why I turned it off.
- [Robot skin can feel pain now. The hard part is teaching it to think.](https://alanscottencinas.com/cognitive-prosthetic-neuromorphic-e-skin-and-fluid-ai/): We spent decades imagining a limb you don't just wear, but feel. Like Luke's hand. This year a lab built the part that always seemed impossible: the wince.
- [LunarSite: An end-to-end ML pipeline for lunar south pole landing site selection](https://alanscottencinas.com/lunarsite-an-end-to-end-ml-pipeline-for-lunar-south-pole-landing-site-selection/): Sim-to-real terrain segmentation, fine-tuned crater detection, PSR-aware XGBoost site scoring, and calibrated uncertainty — validated against NASA’s Artemis III shortlist and real ShadowCam imagery at Cabeus.
- [I built an AI agent that reads my CRM emails and scores leads for $0.01 each](https://alanscottencinas.com/ai-sales-agent-hubspot-claude/): It scored 1,051 leads in one run for $15.07 — and surfaced $82K+ in pipeline I didn't know I had.
- [I Caught Myself Not Thinking. That's When the Research Started Making Sense.](https://alanscottencinas.com/i-caught-myself-not-thinking-thats-when-the-research-started-making-sense/): People who think first and use AI to accelerate gain leverage. People who let AI think for them accumulate cognitive debt. Research from MIT and enterprise studies shows the divide is already forming.
- [The Bifurcation of Cognition: How AI Is Splitting the Workforce Into Augmented Experts and Atrophied Passengers](https://alanscottencinas.com/bifurcation-of-cognition-ai/): AI is not uniformly enhancing or degrading human intelligence. It is redistributing it. As tools absorb certain forms of reasoning, humans reallocate effort toward higher-level synthesis, judgment, and coordination. This shift creates a bifurcation in cognition itself: some skills atrophy while others accelerate. Understanding this split is essential to designing systems that augment rather than erode human capability.
- [Sustainable Cognition](https://alanscottencinas.com/sustainable-cognition/): The IEA estimates global data center electricity consumption at 415 terawatt-hours in 2024, projecting growth to 945 TWh by 2030. This paper argues that achieving sustainable cognitive AI requires simultaneous optimization across three interdependent layers: neuromorphic hardware, software algorithms, and governance frameworks.
- [Before We Knew We Could Survive](https://alanscottencinas.com/before-we-knew-we-could-survive/): From Project Mercury's desperate survival missions to Artemis learning to build lunar infrastructure—exploring how space exploration evolved from fear-driven engineering to the discipline of endurance and continuity.
- [The Death of the Pilot: Why COV is the Future of Drones](https://alanscottencinas.com/the-death-of-the-pilot-why-cov-is-the-future-of-drones/): The commercial drone industry faces a critical scalability bottleneck: the "1:1 Ratio," where increasing fleet size linearly increases human cognitive load. COV—Cognitive Orchestration & Vision—is a novel architectural framework that decouples mission intent from flight execution, shifting the human role from "pilot" to "supervisor."
- [COV: When Fiction Stops Being Fiction](https://alanscottencinas.com/cov-when-fiction-stops-being-fiction/): Years ago, I watched Prometheus and fixated on a small moment most people glossed over. Inside the alien structure, two spherical drones fly autonomously through dark, unknown corridors, scanning, mapping, and reconstructing the interior in real time. The scene wasn't flashy. It was practical. And it planted a thought that never quite faded: Why isn't this real yet?
- [How We Accidentally Built the Death Star and Called It the Golden Dome](https://alanscottencinas.com/how-we-accidentally-built-the-death-star-and-called-it-the-golden-dome/): There's a growing assumption right now that building a national missile shield like the Golden Dome is a largely solved engineering problem—expensive, ambitious, but fundamentally understood. That assumption is wrong. What's being underestimated isn't hardware or physics, but cognition at scale.
- [From Lost Towns to Living Destinations](https://alanscottencinas.com/from-lost-towns-to-living-destinations/): For most of the last century, hospitality scaled by height. Towers, elevators, concrete, and controlled environments. But the market is changing. What's growing now is a different model of development—one that treats land, culture, and community as the core product.
- [Credit Card Fraud Detection with ML | Case Study](https://alanscottencinas.com/credit-card-fraud-detection-with-ml-case-study/): Building real-time fraud detection for financial systems: 92.6% accuracy with just 13 features and sub-millisecond inference. Through aggressive feature selection (435 features → 13), logistic regression proved that production-ready ML doesn't require complexity—it requires smart fundamentals and clear constraints.
- [Transfer Learning for Medical Diagnosis | Case Study](https://alanscottencinas.com/transfer-learning-for-medical-diagnosis-case-study/): How a flower classification model became a medical diagnostic tool for underserved communities. Using transfer learning with InceptionV3, we built a pneumonia detection system that achieved 73% validation accuracy and increased screening throughput by 340% at UCI Medical's community health center. Real deployment, real impact, real constraints.
- [The Golden Dome Is Impressive, Expensive, and Structurally Vulnerable](https://alanscottencinas.com/the-golden-dome-is-impressive-expensive-and-structurally-vulnerable/): The U.S. just greenlit the most ambitious missile defense system in history. It looks like Star Trek, smells like Star Wars, and structurally resembles the Death Star—right down to the fatal flaw.
- [When Sci-Fi Stops Being Fiction: AI Pilots, Lunar Intelligence, and Nuclear Propulsion Are Here](https://alanscottencinas.com/when-sci-fi-stops-being-fiction-ai-pilots-lunar-intelligence-and-nuclear-propulsion-are-here/): AI did not wait for permission to change aviation or space. It simply stepped in and started doing the work. In only a few months, we moved from talking about autonomous systems as distant possibilities to watching them fly fighter jets, navigate the darkest parts of the Moon, and run on hardware powerful enough to reshape the pace of innovation.
- [Why We Already Have the Future of Robotics, But Can't Use It](https://alanscottencinas.com/why-we-already-have-the-future-of-robotics-but-cant-use-it/): COV—Cognitive Orchestration & Vision—didn't emerge from speculative futurism. It emerged from a convergence of existing systems, peer-reviewed research, and a coordination problem that modern autonomy still hasn't solved. The bottleneck isn't perception or mobility. It's coordination under cognitive load.
- [The Drone Industry Hit a Wall](https://alanscottencinas.com/the-drone-industry-hit-a-wall/): For the last decade the drone industry has chased better hardware, longer flight times, sharper cameras, and faster processors. That race is largely over. In 2025, the limiting factor in drone operations is no longer what drones can do—it's what humans can manage.
- [Breaking the "Complexity Wall": The Case for Cognitive Orchestration & Vision (COV) in Multi-Agent Systems](https://alanscottencinas.com/cognitive-orchestration-vision-multi-agent-systems/): Shifting from Pilot-Centric to Intent-Centric Operations: A Framework for Scaling Autonomous Fleets without Linear Cost.
- [Why 200 Drones and a Netflix Movie Have Engineers Re-Coding Reality](https://alanscottencinas.com/why-200-drones-and-a-netflix-movie-have-engineers-re-coding-reality/): Last night, I was standing on the cold wet sand watching 200 drones swarm the sky like a hive of digital fireflies. As they locked into formation, the dark California sky didn't just hold stars anymore—it held a giant, glowing, neon Pikachu. By the time the drones landed, the simulation theory rabbit hole had opened wide.
- [When One Drone Fails, the System Shouldn't](https://alanscottencinas.com/when-one-drone-fails-the-system-shouldnt/): What people underestimate about multi-drone systems isn't flight—it's continuity. Drones don't fail dramatically most of the time. They drift, batteries dip, links degrade, one unit quietly peels off. Unless someone is watching closely, a hole opens in coverage. That's where most systems break.
- [I Built an App in Four Days and Sold It in One](https://alanscottencinas.com/i-built-an-app-in-four-days-and-sold-it-in-one/): This story isn't about speed for the sake of speed. It's about what happens when friction disappears. AI didn't invent the idea. It didn't decide the architecture, the revenue model, or the constraints. What it did was remove the bottlenecks that had stopped me from shipping for years.
- [How Early Digital Systems Quietly Shaped the Minds Building Tomorrow](https://alanscottencinas.com/how-early-digital-systems-quietly-shaped-the-minds-building-tomorrow/): There is a strange pattern you notice if you grew up in the Atari, Nintendo, or PlayStation era. Those early systems did not just entertain us—they trained us. They taught us logic, physics, resource management, experimentation, and the kind of curiosity that asks "what happens if I push this further."
- [Cognitive AI: The Next Leap from Algorithms to Awareness](https://alanscottencinas.com/cognitive-ai-the-next-leap-from-algorithms-to-awareness/): AI today can write code, compose music, and even fly drones—but it still can't truly think. It reacts, predicts, and imitates. But it doesn't perceive. That's the gap between the AI we know and the intelligence that's coming next: Cognitive AI.
- [Cognitive AI, the Energy Obstacle, and the Artemis Solution: Why the Future of Intelligence Will Be Decided by Power, Not Parameters](https://alanscottencinas.com/cognitive-ai-the-energy-obstacle-and-the-artemis-solution-why-the-future-of-intelligence-will-be-decided-by-power-not-parameters/): We are racing toward Cognitive AI systems designed for continual learning, cross-domain transfer, and meta-learning across a lifetime. Yet this ambition collides directly with the energy obstacle. The next wave of breakthroughs will not come from brute-force compute scaling—it will come from smarter, physics-constrained compute.
- [The GIGO Crisis: Why Social Media's Fact-Check Rollback Is Teaching AI to Lie](https://alanscottencinas.com/the-gigo-crisis-why-social-medias-fact-check-rollback-is-teaching-ai-to-lie/): The internet is becoming toxic and AI is drinking from the stream. When Meta removed professional fact-checking, it didn't just change how people consume information—it altered how machines learn. Data poisoning, model collapse, and what it means for AI reliability.

## Pages

- [Contact](https://alanscottencinas.com/contact/): The good conversations usually start with a direct message, not a contact form.
- [All Writing](https://alanscottencinas.com/all-writing/)
- [About](https://alanscottencinas.com/about/): I build systems that work when conditions aren't ideal. ML engineering, cognitive AI, autonomous systems, defense technology, and scaling real companies. Here's the background.
- [Home](https://alanscottencinas.com/): I design and scale intelligent systems across cognitive AI, autonomous technologies, and defense. Writing on what I've built, what I've learned, and what actually works.
- [Privacy Policy](https://alanscottencinas.com/privacy-policy/): This site collects minimal data necessary for functionality and understanding how people use it. Here's what happens, why, and how you can control it.

## Optional

- [Agent (MCP protocol)](websites-agents.hostinger.com/alanscottencinas.com/mcp)

[comment]: # (Generated by Hostinger Tools Plugin)