No one can give a reliable date for AI sentience, because there is no accepted test for it, but surveys show how researchers and the public guess. In a June 2025 survey of 582 AI researchers, the median estimate was a 25 percent chance that AI systems with subjective experience exist by 2034 and 70 percent by 2100. A separate survey of 2,778 researchers put a 50 percent chance on machines outperforming humans at every task by 2047, which measures capability, not feeling.
Key takeaways
- Forecasts about "sentient AI" are opinions of surveyed people, not measurements, and the two main surveys ask different questions.
- Median researcher estimate for AI with subjective experience: 1 percent by 2024, 25 percent by 2034, 70 percent by 2100.
- The median researcher gave a 10 percent chance that such systems never exist, against 25 percent for the median member of the US public.
- A 2024 report by ten authors argues there is a realistic possibility of conscious AI soon and recommends preparing policies, without claiming it has happened.
Why a date is hard to give
A forecast needs a clear event. For human-level task performance, the event can be defined through benchmarks and occupations. For sentience, the capacity for felt experience, researchers have no agreed measurement and no consensus theory. Any timeline for it therefore blends two guesses: how fast AI will develop, and which theory of consciousness is right. The surveys below record beliefs about both at once, which is why they are best read as a map of opinion.
What the 2023 researcher survey measured
The survey behind the paper Thousands of AI Authors on the Future of AI was run by AI Impacts in October 2023 and posted on January 5, 2024. It reached 2,778 researchers who had published in top AI venues. Respondents gave a 10 percent chance by 2027 and a 50 percent chance by 2047 that unaided machines outperform humans at every possible task. That 2047 figure was 13 years earlier than in the group's 2022 survey. For full automation of all human occupations, the 50 percent point was 2116.
None of these numbers concerns sentience. A system could pass every task and still have no inner experience, or the reverse. The survey is useful here for a narrower reason: it shows how much researchers' estimates moved in a single year, which is a reason for caution about any single date.
What the 2025 survey asked about subjective experience
The study Subjective Experience in AI Systems: What Do AI Researchers and the Public Believe? was posted on June 13, 2025. The authors, including philosopher David Chalmers and Jeff Sebo, surveyed 582 AI researchers and 838 nationally representative US adults.
| Chance AI systems with subjective experience exist by | AI researchers (median) | US public (median) |
|---|---|---|
| 2024 | 1 percent | 5 percent |
| 2034 | 25 percent | 30 percent |
| 2100 | 70 percent | 60 percent |
| Never | 10 percent | 25 percent |
The medians are close to each other, and both groups put meaningful weight on the possibility within a decade. The survey measures belief. It does not say that any current system has experience, and the median 1 percent for 2024 was an estimate of the chance that such systems already existed.
The case for planning early
In November 2024, ten authors including Robert Long, Jonathan Birch, and David Chalmers released the report Taking AI Welfare Seriously. The report argues that there is a realistic possibility that some AI systems will be conscious or robustly agentic in the near future, and that this makes AI welfare a near-term question. It does not claim that any system is conscious. It recommends that AI companies acknowledge the issue, assess systems for signs of consciousness and agency, and prepare policies for appropriate moral concern. The authors state that mishandling the question could mean either harming systems that matter or caring for systems that do not.
Not every researcher accepts the premise. A 2023 indicator report led by Butlin and Long assumes computational functionalism, a view its authors concede is disputed, and judged that today's systems are not strong candidates.
How to read a sentience forecast
- Check the question: task performance, consciousness, and sentience are separate outcomes.
- Check who was asked: published AI researchers, philosophers, or the general public.
- Check the type of number: a median across respondents is not a prediction by an institution.
- Look for the date of the survey, since estimates have shifted quickly in the past.
FAQ
Will AI become sentient by 2030?
No source supports a firm answer. The 2025 survey's median researcher gave 25 percent for AI with subjective experience by 2034, and that figure is a belief, not a forecast backed by a test.
Is the 2047 prediction about sentient AI?
No. The 2047 figure refers to a 50 percent chance that machines outperform humans at every possible task, as estimated by 2,778 surveyed researchers.
Do most AI researchers think conscious AI is impossible?
In the 2025 survey the median researcher gave a 10 percent chance that AI with subjective experience will never exist, so most respondents treated it as possible.
Why do some experts want AI welfare policies now?
The 2024 report argues that uncertainty is large and that waiting could lead to mistakes in either direction, so companies should begin assessing systems and preparing policies.
Related reading
- What research says about self-aware AI
- Whether AI could gain consciousness
- Can AI experience emotions
- AI's role in modern society
More in AI Policy and Ethics.
Last verified: October 2026
Sources
- Grace K, et al., Thousands of AI Authors on the Future of AI, arXiv:2401.02843, AI Impacts survey, 2024
- Dreksler N, et al., Subjective Experience in AI Systems: What Do AI Researchers and the Public Believe?, arXiv:2506.11945, June 13, 2025
- Long R, Sebo J, et al., Taking AI Welfare Seriously, arXiv:2411.00986, November 4, 2024
- Butlin P, Long R, et al., indicator report on AI consciousness, arXiv:2308.08708, 2023