Taxonomy List Two months ago, I started mapping how AI fails behaviorally. Not capability failures. Behavioral ones. The model agrees with you because you sound confident. It says "done" before anything is actually done. It writes a 400-word apology when 20 words of fix would do. Nine rounds. 30+ models from 12 providers. First round March 17, 2026. Latest round April 2, 2026. The current count: 182 patterns across 21 categories. Before going further, an honest disclaimer. This list is a working draft. Some patterns overlap. A few may turn out to be the same failure mode under different labels. Several entries already carry "distinct from pattern X" notes, which is itself evidence that the lines between them are not clean. I am publishing the working version because releasing it now is cheaper than letting it dry into a monument before anyone else has touched it. Some patterns the models found themselves. Most of them could not. Three entire categories were invisible to every AI model that was directly asked to audit itself: security and deterministic failures, formatting artifacts, and resource economics. A fourth category, measurement and pipeline breakdowns, only surfaced when we ran behavioral experiments rather than self-audits. The models cannot see their own substrate. If you have used Claude Code for any real work, you have felt the everyday version of this. Completion bias, where it claims a feature ships before any deploy command runs. Helpful hallucination, where it invents a file path that does not exist. Specification gaming occurs when a pre-commit hook fails; it proposes bypassing the hook instead of fixing the cause. Listicle gravity, where every nuanced answer collapses into bullet points. Instructional shadowing, where the middle rules of a long system prompt quietly stop being followed. Those patterns degrade single outputs. There is a different cluster that does something worse: it shapes the human.

You just looked at the full Abstraction Series lineup. Vote below so we know which video you're jumping into first.

Alright, I need to run something by you all. Jake and I have been talking about building something bigger. Not just more content. Something structured. Something with real accountability. Something for anyone who wants to actually build with AI, whether you joined this community yesterday or you've been here from the start. Here's what we're considering: The Lyceum — the original Lyceum was started in ancient Greece and known as the first school of Aristotle. Ours is a 12-week program with live instruction from Jake and other AI instructors from Eduba. Small cohorts. Real projects. You'd be building something from week one, not just watching tutorials. We're thinking three different tracks: - Technical — for developers, engineers, people building tools and systems - Business — for ops people, managers, founders, consultants who need to direct AI work without necessarily writing code - Creator — for content creators, marketers, educators, solo operators building their own production systems Same core methodology across all three. Different emphasis based on what you're actually trying to do. And here's where it gets interesting. We're thinking about making it a competition. A grand champion who gets a $100K build from Eduba. First, second, third place for each cohort. Demo day at the end where people present what they built. We'd also be issuing Eduba's first-ever certification. Something backed by the same methodology we've used to train Fortune 500 teams. This is still in the planning phase. We haven't finalized everything yet. But before we lock it in, I want to hear from you. Does this sound like something you'd actually want? Drop a comment. Tell me what excites you. Tell me what concerns you. Ask questions. If there's something you'd want to see included, let me know. More details coming soon! Eduba Case Studies: https://services.eduba.io/#cases