Concept
A model has no reliable sense of its own uncertainty.
It cannot tell you what it does not know, because it does not represent knowing.
Concept
What AI Genuinely Cannot Do Yet
Learn the real current limits of AI systems and why they persist, so you can plan and promise accurately. Free, one sitting.
Start Quick LessonFree to explore · ~3 minutes
Why this matters
Overpromising on AI is how teams lose credibility and ship features that quietly fail. Knowing which limits are structural — reliable arithmetic without tools, genuine long-horizon planning, knowing what it does not know — lets you design around them instead of being surprised in production.
What you'll cover
Drawing an Honest Line
- Hype in both directions
- Structural versus temporary limits
- Why honesty pays
- Reasoning from the mechanism
- A map of the boundary
Knowing What It Does Not Know
- No calibrated uncertainty
- Confidence scores mislead
- Self-verification fails
- What this forces on you
- Designing for it
Precision and Computation
- Arithmetic without tools
- Character-level operations
- Exact recall of specifics
- Deterministic repeatability
- Tools close most of these gaps
Long-Horizon Work
- Planning over long horizons
- Holding a goal across many steps
- Learning from its own mistakes
- Genuine novelty
- Accountability and judgement
Designing Around Limits
- Mitigations that work
- Residual risk
- Setting expectations
- Choosing not to use AI
- Staying current on the boundary
What you'll understand
- Name the limits that are structural rather than temporary
- Explain why each one follows from the mechanism
- Set expectations that survive contact with production
- Design around a limit instead of denying it
A look inside
Three moments from this Quick Lesson
The real thing — not a mockup of it.
How it fits
Limit → Workaround → Residual risk
Most limits can be mitigated with tools or humans. None of them simply disappear.
Apply
Which is a structural limit?
Writing fluent prose
Knowing reliably when it is wrong
Summarising a document
How it works
01
Understand the concept
A plain-language walkthrough of the idea itself, no prior context assumed.
02
See it in practice
A simple diagram or example showing how it actually fits together.
03
Apply what you learned
One quick check that you can recognise it, not just recall it.
Useful for
Product managersFoundersExecutivesDevelopers
Ready to understand What AI Genuinely Cannot Do Yet?
Want to go deeper? Explore AI Product Manager →
Frequently asked
Will these limits go away with the next model?+
Some will narrow. The ones rooted in how prediction works — calibrated self-knowledge above all — have proved stubborn across many model generations.
What will I learn?+
Which AI limits are structural, why they persist, and how to design products that account for them.
Is this Quick Lesson free?+
Yes. Quick Lessons are free, short, and do not require a paid plan — sign in only to save your progress.