What We Can't Predict
This chapter names the specific uncertainties that make predictions about the next forty years unwise: AI, climate, work, relationships, longevity.
The Honesty Discipline
It's worth being specific about what we don't know. Generic "the future is uncertain" is too vague to act on. Specific "we don't know what the job market will look like in 2050" is both true and actionable.
Below, five specific areas of genuine uncertainty. For each: what's changing, what honest forecasters say, what it means for parenting.
AI and Knowledge Work
What's happening: large language models and related AI systems are changing what knowledge workers produce, how fast, and at what cost. Already, drafting, summarising, translating, coding, and many kinds of analysis are partly automatable. Further capabilities are coming, fast.
What we don't know:
- How much of current knowledge work will be automated
- Which new kinds of work will emerge and how quickly
- What the resulting economy looks like
- Whether the transition is gradual or disruptive
- How much judgment and taste remain valuable
- What happens to professions whose value was "access to information"
Serious people disagree on all of these. Some think this is as big as the internet; some think it's bigger; some think the hype overstates the impact. The range of credible forecasts is wide.
What it means for parenting
Betting on any specific career prediction is rash. "My child will be a radiologist" or "a lawyer" or "a software engineer" assumes those jobs exist and pay well in twenty-five years. Maybe. Maybe not.
Betting on generalised capability makes more sense. A person who can learn new tools, think clearly, work with other people, and exercise judgment will find their footing in most plausible versions of the AI future.
Climate
What's happening: the global climate is warming; this changes where humans can comfortably live, grow food, avoid disasters. The baseline science is settled; the specific trajectory depends on policy, technology, and natural systems.
What we don't know:
- How much warming we'll see by 2050 or 2100
- Which regions become inhospitable, which become more viable
- How migration, politics, and economics respond
- Whether technological responses (geoengineering, fusion, direct air capture) will be significant
- How bad the cascading effects (food, water, disease) are
A few things are fairly likely: climate impacts escalate; displacement and disruption rise; stable places become more valuable. Most specifics remain uncertain.
What it means for parenting
Geographic certainty is weaker than it was. A family assuming their city, state, or country will remain as it is is making a bet with real uncertainty. This doesn't mean panic-preparing; it means raising someone adaptable enough to relocate, rebuild, or change plans.
For children, climate also shapes the psychological load. Many young people live with genuine climate anxiety. Taking that seriously is part of parenting.
Work and Careers
What's happening: the shape of work is changing. Remote work expanded enormously post-2020 and stayed partially. The gig and contract economy is larger. Career changes per lifetime are rising. Specific job titles are emerging and disappearing fast.
What we don't know:
- How much "stable career" remains as a concept
- What credentialing will look like when AI can generate the work that credentials used to signal
- How long careers last, given longer lifespans
- What work-life patterns replace the 9-to-5 era
- Whether "jobs" as we know them remain the main way adults earn a living
Serious observers expect work to keep transforming. The specifics are genuinely uncertain. Predictions that "in 2050 the big job category will be X" have been made and broken for fifty years.
What it means for parenting
Credentialism for a specific career is high-risk. Generic capability plus real interests plus the ability to change course produces a better adult than optimisation for a specific professional identity.
Also: income and stability may be harder to predict than they were. Teaching children to manage money, live below their means, and build some slack is more important than optimising for the maximum income in a specific field.
Relationships
What's happening: household structures are changing. Marriage rates are falling in many countries. Average age of first marriage is rising. Fertility is below replacement in most developed countries. Loneliness is rising. Friendship quality is measurably declining in some surveys.
What we don't know:
- Whether these trends continue or reverse
- What family structures replace the 20th-century nuclear-family default
- How AI relationships (companions, chatbots, more) change human connection
- How social media's effects on relationship formation play out over decades
- Whether loneliness becomes a chronic social condition or gets actively addressed
What it means for parenting
The skills of forming and maintaining real human relationships matter more, not less, as the default context for them weakens. A child who can make and keep friends, resolve conflict, stay connected through inevitable distance, and build a life with people in it is equipped for most versions of what's coming.
The opposite (optimising for individual achievement while neglecting relational skills) is a specific and common mistake of high-achieving households. Chapter 06 goes deeper.
Longevity
What's happening: life expectancy has increased across generations. A child born in 2020 has a reasonable shot at 90+ in many developed countries. Some researchers expect the next fifty years to bring further gains from medical and biotechnical advances; some don't.
What we don't know:
- How long lives get
- Whether those years are healthy or diminished
- What "old age" looks like when 70 is the new 50
- How work, relationships, and meaning shift across a longer lifespan
- Whether longer lives exacerbate inequality or compress it
What it means for parenting
You might be raising a 100-year-old. A person whose "adulthood" is 80 years, not 50. That's a different kind of life. Optimising for the first 25 of those 80 years at the expense of the others is a bad trade.
It also means: the stakes of adaptability rise. A skill that lasts 20 years is useful; a skill that lasts 70 is rare. Most specific skills don't. Meta-capabilities (how to learn, how to adapt) do.
The Compounding Uncertainty
Each area is uncertain on its own. They interact.
- Climate-driven displacement affects work, relationships, and meaning
- AI changes careers, which changes economic structures, which changes families
- Longer lives change career arcs, which change relationship patterns
- Each interacts with the others in ways neither has modelled
The result is not chaos. Humans adapt to changing conditions; they always have. But specific forecasts compound their errors. "AI + climate + demographic shifts + work changes over forty years" is not a scenario anyone forecasts well.
What This Doesn't Mean
Honesty about uncertainty is not the same as nihilism. It doesn't mean:
- Nothing matters
- All preparation is wasted
- You should freak out
- You should give up on specific hopes for your child
It means:
- The specific predictions of credentialism, career planning, and "optimise for this skill" are less trustworthy than they look
- The generic capacities (chapter 03's invariants) are more trustworthy
- Holding your plans loosely is appropriate; it is not the same as not planning
- Your child may surprise you by flourishing in a context you didn't imagine, or by struggling with a future you did
The Positive Side
A reasonable parent could read this chapter and despair. Don't.
Humans have been navigating unknowable futures forever. Our grandparents raised us for futures they couldn't predict; they mostly did fine. The bar is not "predict the future and prepare your child for it exactly". The bar is "produce a person who can meet the future that arrives". That is achievable.
Also: uncertainty cuts both ways. The future might be worse than today in ways we haven't imagined. It might be much better, in ways we also haven't imagined. A child raised with the invariants can do well in either.
Common Pitfalls
"So I shouldn't plan at all." Plan. Just hold the plan loosely. Prepare for the future you expect, with some buffer for the one that arrives
"The future is too scary to think about." Some children will think about it more than you do. They need you to have engaged with it enough to be a guide, not a deflector
"Focus on today; tomorrow will take care of itself." Partially wise. Also partly a dodge. Today-only thinking produces short-term-optimised kids without long-term orientation
"I'll do what my parents did." Your parents operated in different conditions. Some of what they did transfers; some assumes stabilities that no longer hold. Inherit selectively
"We're all doomed." We're not all doomed. Humans are adaptable. Kids raised well handle a remarkable range of futures. Panic is not a parenting strategy
Next Steps
Continue to 03-what-transfers.md for what holds up regardless of which specific future arrives.