Lesson 7 of 1596
Narrow, General, AGI, ASI: What We Mean and Why It Matters
The terminology ladder of AI capability is loaded. Clarify your definitions and you clarify your whole view of the field.
Creators · AI Foundations · ~24 min read
Why Definitions Are Political
If a CEO says their product is AGI, they are simultaneously making a scientific claim, a marketing claim, and often a trigger for contractual clauses with investors. Definitions are not neutral. Pin yours down before you use the words.
The classical ladder
Compare the options
| Level | Working definition |
|---|---|
| Narrow AI (ANI) | Performs a specific task, sometimes superhumanly |
| General AI (AGI) | Matches human-level performance across most cognitive tasks |
| Artificial Superintelligence (ASI) | Vastly exceeds human performance across all domains |
Google DeepMind's Levels of AGI
A 2023 paper by Morris et al. at DeepMind proposed a two-axis framework: performance (from emerging to superhuman) and generality (narrow vs general). This escapes the binary trap and lets you ask, how general, how competent, and at what percentile of humans.
- Level 0: No AI
- Level 1: Emerging (equal to or better than unskilled human)
- Level 2: Competent (50th percentile of skilled adults)
- Level 3: Expert (90th percentile)
- Level 4: Virtuoso (99th percentile)
- Level 5: Superhuman (outperforms 100 percent)
Why the AGI term is contested
- 1No agreed benchmark means anyone can claim it
- 2The label can trigger contracts and IPO milestones
- 3Different stakeholders prefer different goalposts
- 4Anthropomorphizing leads to bad policy decisions
- 5Emergent capabilities blur thresholds further
Alternative framings worth knowing
Compare the options
| Framing | Core concept |
|---|---|
| Transformative AI | AI that triggers changes of industrial revolution scale |
| Human-level AI | Performs any cognitive task humans can do |
| Automated AI researcher | Can itself do the work of an AI researcher (recursive) |
| AGI as economic test | Can do 50 percent of remote work as cheaply as a human |
Implications for your thinking
- Whenever you see AGI, ask whose definition
- Benchmarks should be paired with generality claims
- Short timelines and long timelines both have credible defenders
- Plan products for a range of outcomes, not a point estimate
“AGI is the horizon — it keeps receding as you approach it.”
Key terms in this lesson
The big idea: intelligence is multidimensional, and the AGI label collapses too much nuance. Use multi-axis frameworks and be explicit about what you are measuring.
End-of-lesson quiz
Check what stuck
8 questions · Score saves to your progress.
Tutor
Curious about “Narrow, General, AGI, ASI: What We Mean and Why It Matters”?
Ask anything about this lesson. I’ll answer using just what you’re reading — short, friendly, grounded.
Progress saved locally in this browser. Sign in to sync across devices.
Related lessons
Keep going
Creators · 45 min
What Is Intelligence, Really? A Working Framework
Before we can judge whether an AI is intelligent, we need a framework for what intelligence even means. Draw on Chollet, Dennett, and modern evals.
Creators · 55 min
The Three Ingredients: Data, Compute, Algorithms (Capstone)
Every AI breakthrough of the past decade rests on three interacting ingredients. Synthesize everything you have learned into one working model.
Creators · 30 min
Searle's Chinese Room: Understanding Without Meaning?
A 1980 thought experiment asked whether symbol manipulation alone could ever amount to real understanding.