Which AI doom indexes can you compare? · Which AI indexes does SIREN contrast?
SIREN vs p(doom) clocks: which should you use?
SIREN does not publish a p(doom). It counts observable tempo. For the vocabulary, see the p(doom) explainer.
AI activity index today · SIREN → · state.json → · Scale runs 5 (quietest) to 1 (loudest). Tempo, not danger.
What is each one?
| Index | What it measures | Who sets it | Recomputable? |
|---|---|---|---|
| SIREN | How much is happening in AI (tempo across five pillars) | A published formula over public data, hourly | Yes — from the receipt and open code |
| DoomBench | A qualitative “doom index” from sourced evidence for and against an AI-takeover scenario | Their scoring rules over curated evidence items | No |
| Skynet Countdown | Estimated chance of AI control loss and an estimated AGI date, plus tagged news | Their risk and timeline model over breakthroughs and headlines | No |
| p(doom) clocks | One person’s (or a site’s) stated probability that AI ends badly, often as a clock | Whoever picks the number | No |
| PizzINT | A pizza-and-intel novelty index that borrows DEFCON and Doomsday Clock language in the title | Their site’s framing over pizza / intel metaphors — not a published AI-activity formula | No |
| Takeoff Tracker | A dashboard of AGI-adjacent signals — Metaculus-style timelines, AI stock moves, arena leaderboards, GPU power, datacentres, and collected p(doom) estimates | Their curated dashboard over public prediction markets, markets data and linked leaderboards | No |
What is p(doom) clocks?
- Public claim: Probability-of-doom countdowns.
- Measures: One person’s (or a site’s) stated probability that AI ends badly, often as a clock.
- Set by: Whoever picks the number.
- Recomputable by a stranger: No: p(doom) is an opinion expressed as a percentage.
- Also ships: Countdown-style pages on several sites.
- Their site: pdoom.ai/
What is SIREN?
- Measures: AI activity tempo across capability, compute & capital, attention, governance and markets.
- Set by: a published formula over public data, hourly.
- Recomputable: yes — receipt, open code, CORS JSON.
- Does not publish: takeover probability, AGI date, anyone’s p(doom), a military DEFCON level, or the Bulletin’s Doomsday Clock.
What questions do people actually ask?
SIREN vs p(doom) clocks: which should you use?
Use p(doom) clocks if you want one person’s (or a site’s) stated probability that ai ends badly, often as a clock. Use SIREN if you want an hourly AI activity tempo reading with open inputs and a receipt. They are not substitutes for the same claim.
Can I recompute p(doom) clocks?
No: p(doom) is an opinion expressed as a percentage
Does SIREN publish the same number as p(doom) clocks?
No. SIREN does not score probability-of-doom countdowns. It scores observable activity across five pillars and says so on every page.
What is p(doom) clocks?
p(doom) clocks: Probability-of-doom countdowns. Measures One person’s (or a site’s) stated probability that AI ends badly, often as a clock. Set by Whoever picks the number.
What is SIREN?
SIREN measures AI activity tempo across capability, compute & capital, attention, governance and markets with a published formula over public data, hourly.
Where should you go next?
AI activity index today · SIREN → · How the SIREN AI activity index is computed · methodology · Which AI doom indexes can you compare? · Which AI indexes does SIREN contrast? · What are AI doom index alternatives? · What is p(doom)? → · Is there an AI doomsday clock? →