Independent agreement first
A finding needs the same forecast from at least three people working separately. The strongest come first.
NDX Foresight
What the evidence says
about growing up with AI.
Foresight gathers what researchers, institutions, industry and policymakers say about AI, work and learning. It keeps each claim in the speaker’s own words with its source, shows where voices agree and where they part, and checks dated claims when their date arrives. It does not pretend to predict.
Live from the ledger
The ledger, as it stands today
Every count below comes straight from the Foresight ledger and moves as it is rebuilt. Each prediction is kept in its speaker’s own words, with its source.
Snapshot · 2 November 2026
People followed
125
researchers, institutions, industry leaders and policymakers, including sceptics
- 0139,036Documents gathered
Posts, talks, papers and reports
- 0235,703Read so far
Read and searched for predictions
- 0346,039Predictions recorded
Kept word for word, with the source
- 0413,443With a date
Only a dated claim can be checked
- 051,516Ready to score
Their date has come
- 060Scored
Checked against what happened. Scoring has just begun.
How it works
A ledger, not a forecast
Most writing about AI and education repeats a view. Foresight records who said what, when, and how sure they sounded, so a ministry can weigh it.
Their words, not ours
Every prediction is stored as said, with its source. Nothing is paraphrased into a claim.
Coming due
Dated predictions are tracked to their date, so ministries can see what is about to be tested.
Scored in the open
When a date passes, the prediction is checked against what happened, and a track record builds for each voice.
How it works
Gather · Keep their words · Review · Score when due
- 01
Gather
Posts, talks, papers and reports from a broad roster of thinkers, including sceptics and economists, not only builders.
- 02
Keep their words
Each claim is kept in the speaker’s exact words and checked against the source, or it is dropped.
- 03
Review
Claims are reviewed for whether they are real forecasts, and whether they could ever be checked.
- 04
Score when due
When a dated claim reaches its date, it is checked against what happened. A track record builds over time.
What the sources are saying
The themes in the briefing
From the current briefing, Growing up with AI. These are what the sources say, not what we predict.
Work
- Junior office work changes first
- Work shifts to directing and checking AI
- Some see no near-term mass job loss
- Agents spread through everyday work
Learning
- Tutoring rises while the weight of degrees falls
- Prompting techniques date quickly
- Research itself is increasingly automated
Capability and risk
- Models improve almost month by month
- AI moves into science and robots
- Tasks become far cheaper to run
- Misuse, misbehaviour and security come first
Why it matters
Today’s pupils will live the answers to these questions. Their curriculum is being written now.
Where they agree
Common ground across the sources
The order of change
Office and coding work change first; physical work changes later, through robots.
Education is central
Education is among the sectors the sources discuss most.
Skills that date
Narrow AI techniques taught today are expected to date fast. Durable skills matter more.
Where they part
The disagreements that matter for a curriculum
A ministry plans for years. These are the questions where serious voices give very different answers.
One view
Large change within a low number of years, mostly from people building AI.
Another view
Large change over one or two decades, mostly from economists and engineers.
One view
Robots reach physical work within a few years.
Another view
Physical work stays largely human for decades.
One view
Society adapts as it has before, with new kinds of work.
Another view
If paid work shrinks, it is not clear how incomes would be funded.
What it cannot tell you yet
We show the gaps too
Foresight says plainly where its evidence is thin. That is what makes it usable.
Most claims carry no date
Undated claims cannot be scored, so they inform but do not test.
Voices lean to North America
Few sources speak to any one country. Local evidence is the gap we most want to close, with ministries.
No track record yet
Scoring starts as dated claims fall due. Until then, no voice has earned more weight than another.
The training route
Few sources set out a route for school leavers that holds if office and physical work change together.
For ministries
Growing up with AI: a living briefing
A briefing for ministries of education on what the sources say about the decade today’s pupils will grow up into, rebuilt as the evidence grows. We walk your team through it and connect it to your curriculum.
- What the sources say, and where they disagree
- Dates the sources give, and how many give none
- What it means for skills, assessment and teacher training
- How it connects to your curriculum map and a reform pack
Next step
Brief your team on the evidence
We present the briefing to ministry teams and connect it to their curriculum and training plans.