Superintelligence in Robots · Part of The Humanoid Group

Reading AGI Forecasts

Headlines often quote a year for when AI will match people. Those years come from surveys, trend lines and forecasting panels, each with its own assumptions. This guide explains the main sources and how to read them without being misled.

By Arjun Rao · Updated October 2026

The largest survey of AI researchers

In October 2023, a team led by Katja Grace surveyed researchers who had published at leading AI conferences and journals, receiving 2,778 responses, a 15% response rate. Asked when unaided machines could accomplish every task better and more cheaply than human workers, their aggregate forecast gave a 50% chance by 2047. That was thirteen years earlier than the 2060 estimate from the team's 2022 survey, a large shift in a single year that shows how quickly expert expectations can move.

Why wording changes the answer

The same survey asked a closely related question in a different form: when all occupations could be fully automated. The aggregate forecast for that version gave a 50% chance by 2116, almost seventy years later than the answer about tasks. Respondents were randomly assigned different phrasings, which is how the team could see the gap. The lesson for readers is simple. A forecast year means little without the exact question behind it, so check what was asked before quoting any date.

Trend lines: compute and task length

Other forecasts start from measured trends. Epoch AI, a research institute, found that the computing power used to train notable AI models grew about 4.1 times a year up to 2024, while warning that the trend is sensitive to choices of method. The non-profit METR measured how long a software task, in human time, AI models can finish half the time: about 50 minutes for frontier models in early 2025, a span that had doubled roughly every seven months since 2019. Its authors flag clear limits on how far that generalises.

Robots move on a different clock

Most of these measures track software. Physical skills improve on a different and generally slower curve, for the reasons set out by Moravec's paradox, and a surge in coding ability says little about a robot's hands. Forecasts are useful for long-range thinking but poor for purchasing. For robot decisions, rely on dated, task-specific evidence such as trials on your own work, and revisit the market as that evidence changes rather than waiting for a forecast year.

Sources and further reading

2023 Expert Survey on Progress in AI — AI Impacts (research project).
The survey team's own page for the 2,778-researcher survey, with its dates, definitions and headline forecasts.
Training compute of frontier AI models grows by 4-5x per year — Epoch AI (research institute).
Measured growth in the computing power used to train leading AI models, with the caveats behind the trend.
Measuring AI Ability to Complete Long Software Tasks — METR, Model Evaluation & Threat Research (arXiv).
Tracks how long a task, in human time, AI can complete, and how quickly that span has been growing.

Common questions

When do experts think AGI will arrive?

Estimates vary widely. In a 2023 survey of 2,778 AI researchers, the aggregate forecast gave a 50% chance that machines could do every task better and more cheaply than people by 2047.

Should I plan robot purchases around AGI forecasts?

No. Forecasts describe long-run possibilities with wide uncertainty. Buy for what a robot can do on your tasks today, and review the market as evidence changes.

What Superintelligence Means — Narrow AI, AGI and superintelligence: what each term means, and what each would mean for robots.
Levels of AGI — How researchers grade AI by performance, breadth and autonomy, and where today's systems sit.
Why Bodies Are Hard — Why AI that aces exams still struggles with chores a toddler manages, and what that means for robots.
Levels of Robot Autonomy — From remote control to full autonomy: the scales researchers use to say how independent a robot is.
Aligning Robots — Why robots can do what they were told rather than what was meant, and how researchers close the gap.
Off-Switches and Oversight — Why a capable AI might resist being switched off, and how researchers and safety standards respond.
Frontier AI Safety — The if-then safety plans AI developers publish, and the government institutes that test advanced models.
Reading AGI Forecasts — What expert surveys and trend lines say about when AGI might arrive, and how to read them sensibly.
Asimov's Laws and Ethics — Why Asimov's Three Laws fail as engineering rules, and the principles and standards that replaced them.
AI, Robots and Jobs — What the best evidence says about robots, AI and work, from US factory towns to global exposure estimates.
Latest updates — Ambitious talk of superintelligence now sits alongside new safety tools, shifting rules and sober expert forecasts. These figures and dated updates show what changed, with a source for every claim.
About us — What we cover, how we source every figure, and who publishes this site.

Visit The Humanoid Group

People also search for AGI timeline, when will AGI arrive, AGI predictions, AI expert survey, human-level AI forecast and AI progress forecast.