uniteTOlove.ca
This is the futures library: a plain-language, honestly-sourced map of where credible people think things are heading, across eight domains. Every number carries a date and a named source — and where a figure could not be independently re-verified, we say so where it appears. Where sources disagree, the disagreement is reported as the finding — never smoothed into one house prediction.

The State of the Future

— a conversation starter

PUBLISHED · 2026-08-05 Domain: Futures Library — Synthesis · Source file: this library's internal records · Synthesis compiled 2026-07-23 · domain figures re-verified through 2026-08-05 (see each domain page for its own date)

What this library is

This is not a prediction. Nobody here claims to know what Toronto, Canada, or the world will look like in ten or thirty years. What this library offers instead is something more useful and more honest: a map of what credible, named researchers, institutions, and forecasters actually say about eight big questions right now — artificial intelligence, climate, democracy, work, energy, food, population, and Toronto’s own long-range planning — including exactly how much those credible sources disagree with each other, and why.

The point of building it this way is not to hand readers a verdict. It’s to build a shared, checkable starting point so citizens can have a real conversation about what comes next — and, eventually, help design it — rather than simply consuming someone else’s prediction. Every figure on every page below carries the date it was measured or published and a link to where it came from — and the few figures we could not independently re-verify are flagged in place. Where two credible sources give different numbers for what sounds like the same question, both numbers are shown, side by side, because the disagreement itself is usually the most honest thing that can be said.

The honest patterns, across every domain

Reading all eight domain files together turns up a few patterns that don’t show up clearly in any single one of them on its own.

The demographic shift is real, and it keeps arriving earlier than expected. Each new United Nations population projection has moved the peak of world population earlier and lower than the one before it. Fertility decline has outpaced demographers’ own models almost everywhere it’s been measured, and the rich world’s labour problem increasingly looks like a shortage of workers, not a surplus. Canada is living a policy-sensitive version of this in real time: its population fell for three straight quarters through early 2026, driven by a federal decision to cut immigration targets, not by any change in demography itself.

The energy transition’s cost story is no longer speculative — what’s still contested is pace and politics. Solar costs have fallen roughly 90% over the past decade; battery storage capacity is roughly 11 times what it was in 2021; a quarter of new cars sold worldwide are now electric. What genuinely divides credible institutions today is how fast the rest of the transition happens, and who controls it — visible in the open disagreement between the International Energy Agency and OPEC on when, or whether, oil demand will peak.

Warming’s central estimates cluster together; the real fight is over the tails. Current government policies point toward roughly 2.6°C of warming by 2100 — far below worst-case scenarios sometimes still invoked in headlines, and far above the Paris Agreement’s goals. The genuine scientific tension is over how close the world is to specific tipping points, where a slower, government-negotiated consensus process and a faster-moving scientific synthesis currently see the risk differently — a tension this library names rather than resolves.

Artificial intelligence is the domain where honest uncertainty is widest. Ask the same group of AI researchers the same underlying question two different ways, and their answers can differ by nearly 70 years. Surveys of how many businesses actually use AI differ by four to five times depending on who’s asked and how. What is measured and real: a narrow, replicated hiring slowdown for the youngest workers in the most AI-exposed jobs. What is not yet measured: any economy-wide employment effect.

Tested social programs return mixed, honest verdicts — not clean wins for either side of any argument. Three completed basic-income pilots produced three different findings on employment, alongside more consistent gains in wellbeing. Ontario’s own pilot was cancelled before its official evaluation could even run, so no verdict exists there at all — only smaller, independent surveys after the fact.

Venture-funded promises about food technology have repeatedly met physics and economics. Plant-based meat sales peaked and are now declining. Cultivated meat won regulatory approval years ago but still isn’t meaningfully purchasable almost anywhere. Vertical farms that promised to replace open fields largely went bankrupt; the ones that survived did so by abandoning that original pitch.

The clearest through-line across all eight files is participation, not prediction. The world’s most durable democratic innovations — permanent citizens’ assemblies, participatory budgeting, sortition-based panels — are growing in number, even though a real, honest scholarly critique says their actual policy impact is “patchy at best.” Toronto itself has real, currently under-used capacity here: a Toronto-founded organization that pioneered Canada’s civic-lottery model, a sortition-based city panel that ran twice and was then put on hold, and a youth cabinet that’s operated continuously since 1998.

How to read a forecast

Before trusting any prediction about the future — on this site or anywhere else — three things are worth knowing.

Track records matter more than confidence. Research led by psychologist Philip Tetlock found that trained, scored, and aggregated forecasters — people whose past predictions are tracked and graded over time — measurably outperform both subject-matter experts and professional intelligence analysts. But that edge only shows up on near-term, sharply defined questions, roughly three months to a year out. Accuracy for any forecaster, however skilled, decays toward pure chance by about five years out.

Some things really can be projected further out — but not all things equally. Global-level population projections are the best-performing long-range forecasting genre that exists: historical United Nations projections have generally landed within 1–2% of what actually happened, decades later. Country-level and regional population projections are much less reliable — both Canada’s recent population contraction and South Korea’s fertility rebound caught official forecasters by surprise, within the same year this library was built.

A range with a date beats a single confident number every time. Any credible claim about where something is heading should carry a range, not a point estimate; a date, so a reader knows how current it is; and the source’s own incentive or track record, named plainly. Detailed scenario documents — like “AI 2027” or “Situational Awareness: The Decade Ahead,” both discussed on the AI Futures page — are genuinely useful for thinking clearly about the future, but their own authors call them scenarios, not predictions, and one of those two authors has a large, disclosed financial stake in the outcome the scenario argues for. That kind of disclosure should travel with the claim every time it’s repeated.

Explore the domains