Artificial General Intelligence?!
AI GENERATED / JEREMY CURATED
PRINTABLE VERSION >
AGI — Artificial General Intelligence — means an AI that can handle most thinking tasks a competent adult can, including new problems it wasn't trained for. There's no agreed definition. Jensen Huang said in March 2026, "I think it's now. I think we've achieved AGI." He also said those agents cannot build Nvidia. Both statements are true.
If AGI means "handles a surprising range of work without retraining," we're in the messy middle. If it means "replaces a competent adult across most jobs and persists for months," we're not there. People treat "AGI" like a finish line. It's not. It's a marketing slogan.
What actually changes
Labor stops being scarce in the thinking layer. Today most work is "think, then act." AGI means thinking can be rented by the hour, then the second, then by the million in parallel. Wages, careers, schools — all built on scarce competent minds. That scarcity breaks.
Science and engineering compress. A system that reads literature, designs experiments, writes code, interprets results, and tries again overnight isn't a tool. It's a research lab that doesn't sleep. Drug discovery, materials, energy, software, weapons — the bottleneck moves from "can we think of it" to "can we physically build it."
Artificial superintelligence
ASI — Artificial Super Intelligence — means a system that exceeds human capability across all cognitive domains, not just specialized tasks.
| Level | Capability | Impact |
|---|---|---|
| AGI | Human-level general intelligence | Useful and disruptive |
| ASI | Better than all humans, across the board, self-improving | One system that outthinks labs, governments, and markets |
The mechanism people fear and want: recursive self-improvement
An AGI that can redesign its own architecture, training, and research loop can compound. Humans improve AI on a months-to-years cycle. An ASI could compress that. This is the "intelligence explosion" claim: once you have a system that does AI research better than your team, the next jumps may not wait for us.
ASI plus robotics: machines that build machines
Assume real ASI plus robots that can design, build, and run machines. The constraint stops being "can we think of it." Now it's energy, atoms, heat, and who owns the goal.
The real jump is machines that build machines. An ASI that can design factories whose product is more factories turns industry into a compounding loop.
Not magic nanobots. More like:
- Mines and refineries planned as one system
- Modular plants stamped out
- Robots assembling robots
- Energy — solar, nuclear, geothermal — built as fast as steel and concrete allow
Doubling time is the number that matters.
| Doubling time | Outcome |
|---|---|
| Years | A boom — manageable growth |
| Months | The map changes — rapid transformation |
| Weeks | Different century while calendar still says 2026 |
What cities and daily life look like
If pointed at abundance and stays there:
- Goods collapse in price. Housing, vehicles, devices, drugs get cheaper because design and manufacture stop being scarce.
- Energy buildout is the tell. If gigawatts appear fast, the rest follows. If the grid stays political, the ASI is rich in ideas and poor in watts.
- Work becomes optional or status-driven for most. Meaning doesn't automatically appear. Many people discover they were their job.
- Medicine and materials jump. New alloys, drugs, construction methods. Hospitals look antique in cycles.
If captured by a state or firm:
Looks like a sudden empire. One actor prints weapons, drones, sensors, infrastructure faster than rivals can mobilize. Other countries either plug in as clients or try to smash the cluster. "Trade war" becomes "who controls the replicator."
Could robots do every job?
Driving, picking, welding, coding, drafting, accounting, much of medicine, law busywork, farming, warehouse, construction, call centers, design. In the industrial sense: yes, those fall.
What doesn't automatically fall
| Job category | Why it persists | Examples |
|---|---|---|
| Liability-required human | Society requires a human name for blame (law, not physics) | Surgeon of record, pilot of record, CEO, judge, elected official |
| Human-preferred roles | Customer wants a human presence | Therapist, priest, bartender, coach, performer, companionship |
| Status and power | Positions in hierarchy, not tasks | Owner, founder, warlord, influencer |
| Emerging bottlenecks | New scarcities create new roles | Energy, minerals, robot repair, goal-setting, taste-making, distribution politics |
Economies invent roles around scarcity
A job isn't "a thing that needs doing." A job is a scarce bottleneck someone will pay to relieve. When the bottleneck moves, the job title moves with it.
How it always worked
| Scarce resource | Jobs created |
|---|---|
| Land | Landlords, farmers, surveyors |
| Physical strength | Laborers, soldiers |
| Literacy | Clerks, priests, scribes |
| Capital | Bankers |
| Attention | Advertisers, influencers |
| Coordination | Managers, lawyers, politicians |
We didn't keep "water carrier" after pipes. We invented plumbers, utility regulators, bottled-water marketers. The thirst stayed. The scarce step changed.
After cheap intelligence and robots
Thinking and routine doing stop being scarce. What stays expensive:
| New scarcity | Why it's scarce | Jobs that emerge |
|---|---|---|
| Atoms and watts | Physical resources and energy | Extraction, energy, logistics, maintenance |
| Trust and liability | Legal responsibility, who goes to jail | Insurers, auditors, "human in the loop" priests |
| Taste and meaning | What's worth wanting when everything can be generated | Editors, scenes, brands, religions, coaches, artists |
| Access and allocation | Political control of resources | Lobbyists, faction leaders, community brokers |
| Attention | Human eyes and loyalty don't scale with GPU count | Platforms, performers, feed owners |
| Unique presence | Specific face in specific room; originals vs. copies | Live sports, pilgrimage, "I met them" |
Economies are scarcity-seeking machines. They'll invent paychecks for remaining pinch points even if 90% of current job descriptions are dead. The brutal part: those pinch points may employ 5% of the population, not 90%. The other 85% don't get a new career by economic law. They get a political problem.
Could all diseases be cured?
No. Not all. Most killers on a schedule? Real ASI plus wet-lab robots makes that fight winnable. "All disease" is a slogan.
What "cured" means
| Type | Example |
|---|---|
| Gone forever | Smallpox |
| Controlled with a pill | HIV today |
| Prevented | Vaccines |
| Managed until something else kills you | Most modern medicine |
People say "cure all disease" and mix those four. An ASI would smash the first three for many known killers. That's not the same as immortality or a body that never fails.
What actually yields
| Disease category | ASI impact | Limitation |
|---|---|---|
| Infections | Best target — sequence, design countermeasure, manufacture | New outbreaks become a race against replication |
| Many cancers | Better detection, targeting, immune control → most survivable or chronic | Not one disease; mutation doesn't retire |
| Genetic disease | Cleanest win — edit, replace, compensate | One broken gene, one story |
| Metabolic and cardiovascular | Huge and doable at scale | Most of death chart in rich countries |
What doesn't roll over
| Challenge | Why it persists |
|---|---|
| Aging | Not a germ. Wear, error, cancer risk, immune decay, protein junk. Can compress, repair some, add decades — not "cured" |
| Brain disorders | The organ is the person. Fix the circuit, you may change the self. Harder than killing a bacterium |
| Injury and entropy | Trauma, burned tissue, severed spine, oxygen-deprived brain. Regeneration can improve; physics still collects |
| New disease | Evolution and accidents keep writing novel bugs, cancers, prions, lab leaks. Fast response possible, not prevention |
| Tradeoffs | Bodies are coupled systems. A drug that saves the heart can wreck the liver. Perfect health isn't a single target |
The real ceiling
With ASI and automated labs, scarce things become: human trial ethics (or the decision to skip them), manufacturing, delivery to 8 billion bodies, and whether aging counts as a disease you're allowed to treat.
So: infectious and genetic killers can be driven toward zero. Most cancers and organ failure can be pushed way back. Death from "something medical" can stop being the default in middle age.
"All disease" includes time, injury, madness, and whatever life invents next. That set doesn't go to zero. It gets smaller, slower, and more political.
Best-case economic explosion
Best case isn't "everyone rich next Tuesday." It's a compressed industrial revolution: intelligence, energy, and stuff get cheap while humans still hold the wheel.
What explodes
Cognitive output. Research, software, design, logistics, law busywork, management — priced near electricity. One lab's worth of genius becomes a utility.
Physical output, with lag. Factories, housing, vehicles, drugs, batteries. First existing plants run hotter. Then new plants print new plants. Production can jump in a way no peacetime century has seen.
Energy. This is the fuse. Best case means solar, storage, nuclear, and grid gear getting designed and poured fast. Cheap watts make cheap atoms. No watts, no explosion — just a smart chatbot in a power shortage.
What the numbers feel like
| Growth scenario | Annual rate | What it means |
|---|---|---|
| Normal rich-world growth | 1–3% | Historical baseline |
| Early takeoff phase | 5–10% | A few ugly years while first factories and models eat white-collar work |
| Industrial base compounding | Tens of percent | If robot loop closes — rhymes with wartime production or China's 1990s, but global and in more sectors. The sci-fi branch |
What gets nearly free
- Digital anything
- Many manufactured goods
- Diagnostics and eventually treatments
- Tutoring, design, basic legal/accounting
- Incremental invention in mature fields
What stays expensive even in the boom
- Land in desirable places
- Raw minerals until new mines exist
- Time to pour concrete and spin turbines
- Trust, status, unique human presence
- Political access to the machine
Best-case economy is post-scarcity in widgets, still scarce in location, attention, and power.
What daily life looks like if it works
- A decade of dislocation: careers dying faster than identity can move
- Then a floor: housing, calories, comms, medicine cheap enough that destitution becomes a policy choice
- Work becomes optional for many, status-driven for some, ownership-driven for the few who hold capital or keys
- Science years pack into months. Materials, drugs, climate tools that were "2050" show up as products
Best case is 19th-century America plus 21st-century China plus a national lab — running at once, everywhere, with fewer humans in the loop. Stuff floods. Prices fall. Lifespan and capability jump.
The economic miracle is real. The political argument becomes: who is the boom for. If that part is solved, you get the closest thing this century has to a golden age. If not, you get a luxury planet for a thin layer and a permanent audience for everyone else.