Radiologists are not only fine but doing more volume than ever haha, timelines are the hardest part of AI predictions
@whats_ai
-
Clear Content Beats Hype in AI Communication
By
–
For anyone creating content in AI: We’re in a space where hype spreads fast.
Which makes this even more important. Clear > flashy.
Useful > viral. -
Optimizing for Respect Over Retention in Tech Design
By
–
One way I’m starting to think about this: Instead of optimizing for retention, optimize for respect.
Respect for people’s time, attention, and expectations. It’s a higher bar. -
The Hidden Cost of Wasting 125,000 Hours
By
–
I just wasted 125,000 hours of people’s time.
At least… that’s what I thought at first. A while ago, I gave a 1-hour workshop to ~100 people.
And I remember thinking: “If this is bad… that’s 100 human hours gone.” That idea stuck with me. So today, out of curiosity, I -
Hinton’s Best Interviews When Thinking Deeply
By
–
Hinton's best interviews are the ones where he's thinking about something else. Looking forward to this.
-
Economists Debate Task-Level versus Aggregate AI Impact Outcomes
By
–
Even among those economists, the task-level camp tends to predict messier outcomes than the aggregate-effect one.
-
Designers Still Critical for AI Output Validation
By
–
Knowing when the output is wrong is where designers still win for a good while I think!
-
Does 4.7 Model Tighten Hedging Language or Tool Use?
By
–
Did the 4.7 prompt tighten around hedging, or is it mostly tool-use phrasing?
-
Single-shot vs branching prompts: structural variants exploration
By
–
Is the prompt single-shot, or does it branch on variant dimensions (length, tone, thumbnail)?
-
Agent Context Loss in Design Tools: Structured Data vs Screenshots
By
–
external design tools lose agent context on every trip. Are your agents reading the design artifacts back as structured data, or as rendered screenshots?