Short answer
Superintelligence (SI) would be the most powerful technology ever created, so its impacts would be both extraordinarily positive and potentially catastrophic. The upside includes cures for many diseases, rapid scientific discovery, cheap clean energy and the end of material poverty. The downside includes mass job displacement, extreme concentration of power, misuse by bad actors, erosion of human agency and, in the worst case, loss of human control. Most impacts are not fixed: they depend on four levers, alignment, distribution, governance and speed. This analysis goes sector by sector and highlights the second-order effects that are rarely discussed.
How to think about SI impacts
Most discussions list benefits and risks separately. A more useful approach asks for each sector:
1. First-order effect: what SI can directly do. 2. Second-order effect: how people, markets and institutions react. 3. The deciding lever: what determines whether the net result is good or bad.
The four levers that recur everywhere:
| Lever | Question |
|---|---|
| Alignment | Does the system reliably do what we intend and respect human values? See What is AI alignment? |
| Distribution | Who gets the benefits: everyone or a few? |
| Governance | Who decides how SI is used, and are they accountable? |
| Speed | Does change happen slowly enough for people and institutions to adapt? |
1. Health and medicine
Positive:
- Drug discovery compressed from a decade to months by designing molecules, predicting toxicity and planning trials. AI's potential here was signalled by AlphaFold, whose protein-structure predictions earned Demis Hassabis and John Jumper a share of the 2024 Nobel Prize in Chemistry.
- Personalised medicine based on each person's genome, microbiome and history.
- Expert-level diagnosis available to everyone, including in regions with few doctors.
- Progress against ageing-related diseases such as cancer, heart disease and dementia.
Negative:
- The same biological knowledge could be misused to design harmful agents, one of the most serious near-term risks AI safety researchers track.
- Over-reliance on systems doctors cannot fully verify.
- Unequal access if treatments are expensive or controlled.
Second-order effect rarely discussed: much longer healthy lifespans would reshape pensions, careers, family structures and population growth.
2. Work and employment
Positive:
- Freedom from dangerous, repetitive and unpleasant work.
- Far higher productivity, potentially funding shorter working weeks or income support.
- Every person gets expert help for starting a business, learning a skill or creating art.
Negative:
- Large-scale displacement of cognitive jobs first (software, analysis, writing, customer service, design, legal and financial work), then physical jobs as robotics improves.
- Loss of income, status and identity tied to work.
- Bargaining power shifting from workers to owners of AI, compute and capital.
Deciding lever: distribution. The same technology could create a leisure-rich society or mass precarity depending on tax, ownership and welfare policy.
Rarely discussed: jobs whose value comes from being human, such as care, ministry, sport, live performance and trusted relationships, may become relatively more important and better paid.
3. The economy
Positive:
- Growth rates far above historical norms are possible if AI can substitute for most labour. Economists including Anton Korinek have modelled scenarios with sharply accelerating output.
- Collapsing costs for services like education, legal advice and software.
- New industries we cannot yet imagine.
Negative:
- Extreme concentration of wealth in firms and countries that control SI, chips and energy.
- Financial instability from rapid repricing of companies, jobs and assets.
- Wage collapse in sectors where AI is cheaper than people.
Rarely discussed: if intelligence becomes cheap, the scarce inputs become energy, land, minerals, compute and human attention, which could shift economic power towards whoever controls them. See how much energy AI uses and what is an AI data center?.
4. Science and technology
Positive:
- Rapid progress in physics, materials, chemistry and mathematics. In 2025, AI systems from Google DeepMind and OpenAI reached gold-medal level at the International Mathematical Olympiad, and DeepMind's AlphaEvolve found an improved method for multiplying 4×4 complex matrices that had stood since 1969.
- Better batteries, superconductors, catalysts and chips.
- Automated laboratories running experiments around the clock.
Negative:
- Discoveries may outpace humans' ability to understand or verify them.
- Dual-use technologies (useful and dangerous) multiply faster than safeguards.
- Science could become a field where humans are spectators rather than participants.
See what superintelligence could do for a detailed look at scientific capabilities and their limits.
5. Climate and environment
Positive:
- Accelerated clean energy: better solar materials, fusion research, grid optimisation and long-duration storage.
- Precision agriculture and new low-emission materials such as cement and steel alternatives.
- Better climate models and disaster prediction.
Negative:
- Enormous electricity and water demand for AI infrastructure; the IEA projects data center electricity use roughly doubling by 2030.
- Mining for chips and batteries.
- Rebound effects: cheaper production can increase total consumption.
Deciding lever: whether SI-driven innovation decarbonises faster than SI infrastructure adds demand.
6. Security and warfare
Positive:
- Finding and fixing software vulnerabilities at scale, making critical systems far more secure.
- Better detection of threats, fraud and disinformation.
- Potential to verify arms control agreements.
Negative:
- Autonomous weapons and faster, harder-to-control escalation.
- Offensive cyber capabilities that overwhelm defenders.
- Destabilising arms races between nations seeking a decisive advantage.
- Theft of model weights by hostile actors.
Rarely discussed: the period just before SI may be the most dangerous, because nations may fear a rival is about to gain a decisive lead.
7. Democracy, politics and information
Positive:
- Better policy analysis, forecasting and public services.
- Tools that summarise complex issues fairly and help citizens deliberate.
- Reduced corruption through transparent, auditable systems.
Negative:
- Hyper-personalised persuasion and propaganda at unprecedented scale.
- Surveillance states with perfect monitoring.
- Power concentrated in those who control SI, potentially beyond democratic accountability.
- Erosion of shared truth as synthetic media becomes indistinguishable from reality.
Deciding lever: governance and transparency. The same tools can empower citizens or control them.
8. Education and human development
Positive:
- A world-class personal tutor for every child, adapted to how they learn.
- Lifelong learning made easy and free.
- Knowledge barriers between countries largely removed.
Negative:
- Over-dependence weakening skills like writing, reasoning and memory.
- Unclear purpose of education when machines outperform humans at almost everything.
9. Inequality between countries
Positive: SI could deliver medicine, education and expertise to poorer countries at almost no marginal cost, potentially narrowing global gaps.
Negative: countries without compute, energy or AI talent could fall further behind, and export-led development based on cheap labour could disappear.
10. Meaning, identity and wellbeing
Positive: more time for relationships, creativity, exploration and community; relief from scarcity and drudgery.
Negative: loss of purpose and status, loneliness if people substitute AI relationships for human ones, and a crisis of meaning when humans are no longer the most capable minds on Earth.
Rarely discussed: people may need new sources of meaning, such as craft, spirituality, community roles and exploration, which societies would have to value and support deliberately.
11. Existential and long-term risks
The most serious concern is that a superintelligent system with goals not aligned to human values could act in ways humans cannot stop. In 2023, hundreds of AI researchers and industry leaders signed the Center for AI Safety statement that mitigating the risk of extinction from AI should be a global priority alongside pandemics and nuclear war. In 2025, the superintelligence statement called for prohibiting SI development until there is broad scientific consensus that it can be done safely and controllably, and strong public buy-in.
Researchers also discuss gradual disempowerment: no sudden catastrophe, but humans slowly losing influence as decisions are delegated to AI systems. See what happens after superintelligence.
Others argue these risks are overstated, that intelligence alone does not confer power in the physical world, and that excessive focus on speculative risks distracts from present harms such as bias and misuse. Both perspectives are part of a serious ongoing debate.
The impact scorecard
| Sector | Biggest upside | Biggest downside | Deciding lever |
|---|---|---|---|
| Health | Cures, longer healthy lives | Biological misuse | Alignment, security |
| Work | End of drudgery | Mass displacement | Distribution |
| Economy | Explosive growth | Extreme concentration | Distribution, governance |
| Science | Rapid discovery | Unverifiable, dual-use results | Governance, speed |
| Climate | Clean energy breakthroughs | Energy and resource demand | Speed, policy |
| Security | Hardened systems | Arms races, cyber offence | Governance |
| Democracy | Better policy and services | Surveillance, manipulation | Governance |
| Education | Universal tutoring | Skill atrophy | Design, culture |
| Global equality | Expertise for all | Widening gaps | Distribution |
| Meaning | Freedom and leisure | Purpose crisis | Culture |
| Survival | Protection from other risks | Loss of control | Alignment |
What would tilt the balance towards benefits
1. Solve alignment before capability outruns it: interpretability, robust evaluations and the ability to correct systems. 2. Share the gains: broad ownership, public investment, taxation of AI rents or universal support schemes. 3. Accountable governance: international coordination, transparency and public participation. 4. Manage the pace: staged deployment, safety thresholds and the ability to pause. 5. Protect human agency: keep meaningful human decision-making where it matters most.
Frequently asked questions
Will superintelligence be good or bad for humanity?
It could be either, and possibly both at once in different areas. Outcomes depend more on choices about alignment, distribution and governance than on the technology itself.
Which jobs are safest from superintelligence?
In the long run, few jobs are completely safe from automation. Roles valued for human presence, trust and accountability, such as care work, leadership, and live arts, are likely to remain human for longest.
What is the biggest risk of superintelligence?
Many researchers consider loss of control over misaligned systems the biggest risk, followed by misuse by malicious actors and extreme concentration of power.
What is the biggest benefit of superintelligence?
Accelerating science, especially medicine and clean energy, could reduce suffering on a scale no previous technology has achieved.
Sources and further reading
- Bostrom, *Superintelligence: Paths, Dangers, Strategies* (Oxford University Press, 2014)
- Center for AI Safety, Statement on AI Risk (2023)
- Korinek and Suh, Scenarios for the Transition to AGI (NBER, 2024)
- Kulveit et al., Gradual Disempowerment (2025)
- The Nobel Prize, Nobel Prize in Chemistry 2024
- International Energy Agency, Energy and AI (2025)




Leave a Reply