Domains · Article 9 of 60
Models and AGI: the capability timeline
The largest domain in the collection, and the one with the least agreement.
This is where people argue about when systems stop being tools. The claims here are unusually hard to grade because 'AGI' is not a measurable quantity and because the people making the claims are also the people who get to define the term.
Read the rows for the shape of the disagreement rather than for a consensus. Where a domain is this contested, the useful information is the spread.
| Claim | Horizon | Verdict |
|---|---|---|
| <div><em>“It’s soon. I’m updating nearer and nearer.”</em><br /><span class="tiny">Demis Hassabis — WSJ Magazine profile, 2026-10-21</span></div> | 2028 – 2029 | <span class="verdict" data-v="open">Open</span> |
| <div><em>“Breathless claims about AGI and new capabilities fall apart pretty quickly under scrutiny.”</em><br /><span class="tiny">Timnit Gebru — MIT Technology Review, with Emily M. Bender, 2026-09-22</span></div> | 2026 claim cycle | <span class="verdict" data-v="partial">Partly borne out</span> |
| <div><em>“Of course I want to do AGI. This is the only meaningful thing to do over the next ten years.”</em><br /><span class="tiny">Yang Zhilin (Zhilin Yang) — TIME100 AI profile, 2026-08-27</span></div> | By 2036 | <span class="verdict" data-v="open">Open</span> |
| <div><em>“In my mind, I thought it was coming in four years instead of 10 years.”</em><br /><span class="tiny">Masayoshi Son — Public remarks, reported June 2026, 2026-06-06</span></div> | By mid-2030 | <span class="verdict" data-v="open">Open</span> |
| <div><em>“It seems hard to rule out even shorter [timelines]. Is there 1% chance of this happening in six, nine months? Yeah, that seems pretty plausible.”</em><br /><span class="tiny">Beth Barnes — 80,000 Hours interview, 2026-06-05</span></div> | By 2028 | <span class="verdict" data-v="open">Open</span> |
| <div><em>“Right now it’s like at the child level, they can do like half an hour ahead, they can plan half an hour ahead. But if the curve continues, that means in about five years they are at…”</em><br /><span class="tiny">Yoshua Bengio — Interviews referencing METR time-horizon data, 2026-02-17</span></div> | Around 2031 | <span class="verdict" data-v="open">Open</span> |
| Person | Organisation | Group | Claims |
|---|---|---|---|
| <div class="person"><span class="face sigil" aria-hidden="true">DH</span><div><strong>Demis Hassabis</strong><br /><span class="tiny">Chair, Google DeepMind; Alphabet chief scientist · Google DeepMind</span></div></div> | Google DeepMind | Frontier labs | 3 |
| <div class="person"><span class="face sigil" aria-hidden="true">EM</span><div><strong>Elon Musk</strong><br /><span class="tiny">xAI founder; SpaceX CEO · xAI / Grok (within SpaceX)</span></div></div> | xAI / Grok (within SpaceX) | Frontier labs | 3 |
| <div class="person"><span class="face sigil" aria-hidden="true">BB</span><div><strong>Beth Barnes</strong><br /><span class="tiny">Founder and CEO · METR</span></div></div> | METR | Researchers & evaluation | 2 |
| <div class="person"><span class="face sigil" aria-hidden="true">IS</span><div><strong>Ilya Sutskever</strong><br /><span class="tiny">Cofounder and CEO · Safe Superintelligence (SSI)</span></div></div> | Safe Superintelligence (SSI) | Frontier labs | 2 |
| <div class="person"><span class="face sigil" aria-hidden="true">MS</span><div><strong>Masayoshi Son</strong><br /><span class="tiny">Chairman and CEO · SoftBank Group</span></div></div> | SoftBank Group | Compute & capital | 2 |
| <div class="person"><span class="face sigil" aria-hidden="true">MS</span><div><strong>Mustafa Suleyman</strong><br /><span class="tiny">CEO, Microsoft AI · Microsoft AI</span></div></div> | Microsoft AI | Frontier labs | 2 |
Figures computed from docs/predictions/ledger.json at build time. Every claim quoted on this page carries its own sources in the ledger.