Synthetic Intelligence
The central analogy
Reid Hoffman argues that for the first time in history, synthetic intelligence — not just knowledge — is becoming as flexibly deployable as synthetic energy has been since the rise of steam power. “Intelligence itself is now a tool — a scalable, highly configurable, self-compounding engine for progress.”
The parallel is structural, not metaphorical:
| Dimension | Synthetic energy (steam, 1700s+) | Synthetic intelligence (AI, 2020s+) |
|---|---|---|
| Pre-existing constraint | Energy was scarce — everything required human/animal labor | Intelligence was scarce — everything required human cognition |
| What it augmented | Physical capability — tilling, digging, manufacturing | Cognitive capability — analysis, creativity, decision-making |
| Scale mechanism | Mechanization → factory system → mass production | AI models → agent systems → cognitive automation |
| Democratization | From elite workshops to mass manufacturing | From specialized labs to ChatGPT for everyone |
| Agency expansion | Humans could do physically superhuman things (travel 60mph, lift tons) | Humans can do cognitively superhuman things (diagnose across specialties, analyze millions of documents) |
Superagency
When a critical mass of individuals are personally empowered by AI, the effects compound through society — a state Hoffman calls superagency. The downstream effects reach even non-users:
- Your doctor diagnoses with AI precision
- Your mechanic identifies obscure problems instantly
- ATMs, parking meters, and vending machines become multilingual and adaptive
Superagency is the societal-level expression of synthetic intelligence, just as industrialization was the societal-level expression of synthetic energy.
The humanizing force
Hoffman rejects the framing of technology and humanism as oppositional forces. He argues they are integrative: “Every new technology we’ve invented — from language, to books, to the mobile phone — has defined, redefined, deepened, and expanded what it means to be human.”
The Industrial Revolution, despite its “violently dehumanizing” early period (factory conditions, child labor, urban anomie), ultimately proved to be a radically humanizing force: fairer laws, economic mobility, social welfare, individual rights. Steam-powered mechanization “expanded the possibilities of what it meant to be human, in ways that were exponential in scope.”
The same trajectory applies to AI. The early disruption period is real, but the long arc bends toward expanded human agency — if synthetic intelligence is broadly distributed rather than concentrated.
The two levers of human agency
Hoffman frames human agency as driven by two complementary forces:
- Intelligence gives the capacity to weigh options, envision scenarios, and plan
- Energy enables action on those plans
Every major technology in history augmented one or both. AI is unique in directly augmenting intelligence itself — the lever that was historically bottlenecked by human cognitive limits. This makes AI’s potential impact comparable to steam power, which unlocked the energy lever at scale for the first time.
Connection to existing frameworks
The synthetic intelligence framing enriches several existing frameworks:
-
Enabling vs Replacing Technologies: Synthetic intelligence is enabling by default — it augments human cognitive capability. But the direction depends on deployment choices (Acemoglu), not inherent properties. AI deployed as pure automation is replacing; AI deployed as capability amplification is enabling.
-
Technological Revolution Cycles: Perez’s framework predicts that the current AI frenzy phase requires a turning point (institutional reform) before the synergy/golden age arrives. Hoffman’s framing adds: the turning point for synthetic intelligence requires broad distribution, just as synthetic energy’s turning point required the factory system to spread beyond Britain.
-
The Expanding Work Frontier: If intelligence becomes cheap and abundant (like energy after steam), the frontier of what organizations can do expands in proportion. The 95% of agent tokens spent on never-before-done work is the expansion that abundant synthetic intelligence enables.
The four schools of AI
Hoffman categorizes perspectives on AI’s trajectory:
| School | Stance | Approach |
|---|---|---|
| Doomers | Existential risk from superintelligence | Pause or ban development |
| Gloomers | Near-term harms (jobs, disinfo, bias) | Prohibitive regulation |
| Zoomers | Benefits far exceed costs | Full autonomy for developers |
| Bloomers | Optimistic + participatory | Iterative deployment, broad access, mass engagement |
Hoffman places himself firmly as a Bloomer: “the freedom to innovate and the obligation to regulate are both important.” The Bloomer stance aligns with iterative deployment — incremental releases that give society time to adapt while maintaining innovation momentum.
Related
- Enabling vs Replacing Technologies — Whether synthetic intelligence enables or replaces depends on deployment
- Technological Revolution Cycles — AI as the fifth technological revolution
- The Expanding Work Frontier — Abundant intelligence expands what organizations can do
- The Two Machine Ages — Brynjolfsson/McAfee’s framework for the same transition
- The Productivity Paradox — Organizational restructuring lag applies to synthetic intelligence too