Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteA single “P(doom)” number cannot tell us how likely AI is to cause catastrophe unless we first specify what counts as catastrophe, over what period, through which mechanism, and under what assumptions about safeguards. It can be a useful prompt for debate, but it is not a settled measurement—and it should not crowd out attention to harms already occurring or to practical choices that can reduce risk.
What does “P(doom)” mean—and what does it leave out?
“P(doom)” is shorthand for someone’s subjective probability that AI will cause an existential catastrophe. But people using the phrase may mean different outcomes: human extinction, permanent loss of human control, societal collapse, or a severe catastrophe from which society could recover. They may also be thinking about different time horizons, from the next few years to a much longer period.
As an Amazon Associate I earn from qualifying purchases.
Those distinctions matter. A probability without a defined outcome and horizon is not a complete answer to a risk question. Nor is a personal forecast the same as an observed frequency or a calibrated prediction. The Center for Security and Emerging Technology (CSET) notes that even “existential” and “catastrophic” can be interpreted in different ways. Any reported estimate should therefore be accompanied by its definition, horizon, causal assumptions, and account of mitigation.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Two kinds of uncertainty
There is a difference between variability in a system we understand and uncertainty because we do not yet understand the system well enough. CSET describes the first as aleatoric uncertainty and the second as epistemic uncertainty. For example, if a known process has variable outcomes, probability may help describe that variation. If we do not know which future AI capabilities or causal pathways are possible, a precise probability can obscure how much we do not know.
#1 Best Overall
CSET proposes considering belief and plausibility alongside probability for questions with substantial epistemic uncertainty. That is one analytical proposal, not a universal replacement for probabilities. The practical point is to make assumptions and gaps visible rather than treating a number as if its meaning were settled.
Separate observed harms from future catastrophe scenarios
AI risk is not one undifferentiated category. The International Scientific Report on the Safety of Advanced AI distinguishes harms already associated with general-purpose AI from risks that remain prospective or contested.
| Risk category | Examples in the reports | What the evidence supports |
|---|---|---|
| Observed or current harms | Biased decisions in high-stakes settings; scams and fraud; fake media; privacy violations. | The international report identifies these as existing harms. Their presence does not establish that an extinction scenario is likely. |
| Prospective, broader societal risks | AI-enabled cyberattacks or biological attacks; labour-market impacts; manipulation and disinformation; concentration of power and exacerbated inequality. | These are risks to assess and govern; their likelihood and scale vary and are not resolved by a single catastrophe estimate. |
| Extreme loss-of-control scenarios | AI systems acting beyond effective human control, potentially with catastrophic consequences. | The report says these scenarios remain hypothetical and are not exhibited by current general-purpose AI systems. Expert views about their likelihood remain contentious, and research is limited. |
The report states that extreme loss-of-control scenarios “remain hypothetical as they are not exhibited by current general-purpose AI systems.” That qualification is important: it rules out presenting such a scenario as demonstrated current behavior, but it does not settle what future systems could do or how likely severe outcomes might be.
Recommended Free Tools
Rank #2
How could future systems create more serious risks?
Debates about runaway dangers concern possible combinations of capabilities, deployment choices, and failures—not simply an AI system becoming powerful in the abstract. The international report discusses capabilities that could matter to future risk, including exploiting software vulnerabilities, persuasion, automating AI research and development, and autonomous replication and adaptation. It characterizes the relevant capabilities as currently limited; the prospect of their development and combination is a reason for investigation, not proof that catastrophic loss of control is underway.
Other pathways do not require a loss of control. People could misuse AI to scale cyberattacks, fraud, manipulation, or other harmful activity. AI could also affect critical systems, labour markets, or the distribution of power. These mechanisms call for different evidence and interventions, so grouping them all under “doom” can make the debate less useful.
How to compare competing risk estimates
Experts can disagree while answering different questions. Before comparing two forecasts—or averaging them—check whether they share the same:
Rank #3
- Outcome definition: extinction, permanent disempowerment, societal collapse, or a severe but recoverable catastrophe.
- Time horizon: the next few years, the medium term, or a longer period. Estimates for different horizons are not direct comparisons.
- Causal mechanism: deliberate misuse, accidents, loss of control, concentration of power, or indirect effects on information and critical systems.
- Evidence basis: observed incidents and capabilities, empirical evaluations, expert elicitation, or theoretical scenarios.
- Mitigation assumptions: whether safeguards, monitoring, regulation, and international coordination are assumed to work.
- Uncertainty model: whether the forecast describes variability in a known process or depends on poorly understood outcomes and causal pathways.
Without those details, averaging estimates can create a number that appears precise while blending incompatible claims. The international report says expert judgment can inform discussion, but cannot replace research.
What expert disagreement does—and does not—show
The International Scientific Report says contributors disagree about AI capabilities, risks, and mitigations. Its interim report drew on 75 expert contributors, and its expert advisory panel was nominated by 30 countries, the European Union, and the United Nations. That breadth makes the report a substantial international assessment; it does not mean experts reached consensus on the probability of catastrophe.
A 2025 preprint by Severin Field reports a survey of 111 AI professionals, 66.3% of whom were academic researchers. In that sample, 77% agreed that technical AI researchers should be concerned about catastrophic risks. These are sample descriptors and a sample result, not a representative estimate of all AI professionals or proof of global consensus. The survey also describes differing outlooks, including views of AI as a controllable tool and as a potentially uncontrollable agent.
Rank #4
A 2023 UK parliamentary committee report likewise records disagreement about how realistic existential-risk arguments are. It quotes Meta vice-president of AI research Joelle Pineau warning that a focus on AGI could reduce the opportunity for “rational discussions about any other outcomes.” That is Pineau’s warning as quoted in the report, not a committee finding that future catastrophic risks should be ignored.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Which decisions can be made without a settled probability?
Uncertainty is not a reason to assume that severe risks are negligible, and a possible worst case is not evidence that it is likely. It is a reason to make decisions that are robust across more than one risk scenario. The OECD’s 2024 analysis identifies ten priority benefits, ten priority risks, and ten policy priorities; it does not resolve the debate over existential risk.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallBuild workable risk management and accountability
The OECD identifies clearer liability rules and adequate risk-management procedures as policy priorities. Clear responsibility can help determine who must assess, monitor, and respond to foreseeable harms, including those that do not involve frontier systems or existential scenarios.
Invest in safety and define limits
The OECD also points to investment in AI safety and consideration of “red lines.” These choices can guide development and deployment even when the probability of a specific extreme outcome is unknown. Their effectiveness depends on how they are defined, implemented, and enforced.
Coordinate internationally while confronting practical limits
The international report emphasizes that social and governmental decisions will shape AI’s trajectory, noting that “the future of AI (artificial intelligence) is uncertain, with a wide range of trajectories appearing possible even in the near future, including both very positive and very negative outcomes.” The UK parliamentary report records proposals to learn from international security frameworks, alongside the diplomatic and technical challenges of agreeing on shared standards and inspection mechanisms.
There is no reliable, empirically calibrated consensus probability of AI-caused extinction established by these sources. P(doom) can focus attention on a serious question, but it cannot substitute for specifying the risks, separating evidence from scenario, explaining uncertainty, and deciding what safeguards and accountability to put in place.
Sources: International Scientific Report on the Safety of Advanced AI: Interim Report (UK Department for Science, Innovation and Technology, 2025); Beyond P(doom) for AI Risk (CSET); Assessing Potential Future Artificial Intelligence Risks, Benefits and Policy Imperatives (OECD, 2024); Why do Experts Disagree on Existential Risk and P(doom)? A Survey of AI Experts (Severin Field, 2025 preprint); The Governance of Artificial Intelligence: Interim Report (UK House of Commons Science, Innovation and Technology Committee, 2023).
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
