TT5: Good Taste, Questionable Ideas
Jan 15, 2026
January 15, 2026
Happy New Year! I hope 2026 is off to a great start for you.
Over the holidays, I found myself at an arcade with my kids. While they ran off to the racing games, I gravitated toward the classic. Standing there pumping quarters into a vintage machine from 1981, a ridiculous thought popped into my head: What if I built something like this inside Revit? Normally, I'd laugh that off. But this time, I didn't. More on that below.
That experience got me thinking about a theme that runs through this week's links: what stays human when AI can do so much? The answer, it seems, is judgment—knowing what's worth building, what problems matter, and what "good" actually looks like. AI can accelerate execution, but it can't tell you which ideas are worth pursuing. That's still on us.
With that in mind, here are the links that caught my attention this week.
#1: I Built a Video Game Inside Revit
Yes, I really did this (I never said my judgment was perfect). BIM Defender is a Space Invaders-style game where you blast rogue users, in-place families, and corrupt files descending on your model. Could I have built this without AI assistance? Sure. Would I have? Absolutely not—40+ hours for a joke wasn't worth it. But two hours? That changes the equation entirely. The real unlock isn't that AI writes code for you; it's that ideas you'd never justify building suddenly become possible.
#2: Architecture After Certainty
Archinect asked design leaders to share their predictions for 2026, and the responses are sobering. Economic uncertainty, the scramble to define AI's role, and fundamental questions about what architects should even be doing right now. OMA's Reinier de Graaf warns the profession is at a breaking point. MAD's Yosuke Hayano argues architects need to become "optimistic future-gazers." Whether you find this energizing or exhausting probably depends on the day.
#3: EULAs - 28 Days Later
Remember that AEC Magazine cover story on software license agreements quietly claiming rights to your data? The conversation hasn't stopped. Martyn Day reports that large firms are now pushing back on EULA terms at the boardroom level, and a cross-industry working group is drafting a "Tech Stack Bill of Rights." If you're not paying attention to what you're agreeing to when you click "Accept," this is your wake-up call.
#4: The One AI Skill That Truly Matters
Luke Johnson kicks off a new series with some pointed observations: prompt engineering is becoming obsolete as fast as you can learn it, and most AI "wrapper" products are living on borrowed time. What actually matters? The intersection of your unique skills, your industry experience, and how you leverage these tools to get real work done. It's a good reminder that the future belongs to domain experts who build, not people who memorize magic words.
#5: How to Learn Taste
Scott Young explores why "taste"—the ability to discern good ideas from bad ones—might be the skill that matters most in an AI-saturated world. AI can find solutions, but it struggles to identify which problems are worth solving. Interestingly, taste appears to be transmitted through close contact with people who have it, not through books or tutorials. A compelling case for why mentorship and enculturation might matter more than ever.
That's all from me. Here's to a new year of good ideas worth building!
Michael
Responses