Within Timeline Effects
Does more time really make AI safer?
Longer timelines may help if society uses the extra time well, but they can also leave more room for proliferation, racing, and gradual loss of control.
On this page
- The case for preparation time
- Risks that can grow over longer runways
- How to judge safety progress against capability progress
Page outline Jump by section
Introduction
One of the most intuitive arguments in debates about AI doom — the risk that advanced artificial intelligence could trigger existential catastrophe — is that more time before the creation of truly transformative AI should make us safer. If society expects decades rather than just a few years of progress, the thinking goes, researchers, regulators and institutions will have a long runway to study safety, build safeguards, and coordinate globally. But does a longer timeline actually reduce the probability of catastrophic outcomes? The short answer is: not automatically. While extended timelines offer opportunities for preparation, they also come with their own risks and uncertainties. Whether more time lowers or raises p(doom) depends on how preparation, incentives, and competitive pressures evolve alongside capability advances.
The Case for Preparation Time
Many researchers argue that a longer interval before the arrival of transformative AI offers crucial resources to reduce risk. Extra time can be valuable in several ways:
- Safety and alignment research can mature. Complex problems such as ensuring AI systems reliably pursue human-compatible goals, remain understandable and controllable, and behave safely even in unforeseen situations are hard. A longer period can allow deeper technical progress on interpretability, oversight methods, and theoretical foundations of alignment.[Open Source AI Models]aimodels.orgOpen Source AI Models How Do AI Timelines Affect Existential Risk?Open Source AI ModelsHow Do AI Timelines Affect Existential Risk? - AI Models…
- Learning from experience matters. With more time, developers and regulators can experiment with increasingly capable systems in lower-stakes contexts, observe failure modes, and adjust safety frameworks before the most powerful systems appear. Society learns by doing, and short timelines compress that learning curve dangerously.
- Governance and institutions evolve slowly. International cooperation, regulation, monitoring regimes and norms around deployment often take many years to establish. A long runway can give policymakers and civil society the breathing room to shape robust governance before the stakes escalate.
For these reasons, many in the safety community see delay as a safety asset, not an obstacle. In work on existential risk modelling, delaying the creation of superintelligent systems is posited to lower overall existential risk if AI risk remains high relative to other global hazards, because it gives alignment and governance science time to catch up.[arXiv]arxiv.orgarXiv How Do AI Timelines Affect Existential Risk?arXivHow Do AI Timelines Affect Existential Risk?August 30, 2022…
Moreover, recent research suggests that the default probability of averting catastrophic outcomes is higher on medium and longer timelines, partly because societies have more time to mobilise resources, build strategic understanding, and strengthen safeguards before transformative capability leaps occur.[Forethought]forethought.orgShort AI Timelines Aren’t Always Higher-LeverageForethoughtShort AI Timelines Aren’t Always Higher-LeverageJanuary 22, 2026…
Risks That Can Grow with Longer Runways
However, a longer timeline isn’t a sure path to safety unless its advantages are actively realised. Several mechanisms can increase risk or make risk reduction harder over extended periods:
- Competitive pressure can still accelerate harmful behaviour. Firms or states racing to be first to achieve transformative AI may prioritise speed over safety, especially if victory confers huge strategic or economic advantages. Even on long timelines, competition could compress safety gains into a late sprint that outpaces governance. Indeed, economic modelling suggests that the allocation of resources between speed and safety matters, and longer timelines alone do not guarantee that safety will be adequately prioritised.[NBER]nber.orgThe AGI Race and Existential Risk | NBERNBERThe AGI Race and Existential Risk | NBER…
- Technological advances may create new vulnerabilities. As AI systems become more capable, they may integrate deeply into critical infrastructure, economic systems and military applications. Longer exposure to such systems — even if not yet superintelligent — can steadily increase systemic fragility, accumulate dependencies, and reduce human control. This accumulative risk perspective argues that risk can grow gradually as sophisticated AI embeds itself in society, raising the stakes even before peak capabilities are reached.[Springer]
- Delays may not slow capability growth evenly. Some forms of progress — such as automated research or recursive self-improvement — could lead to sudden leaps in capability even after a long slow build‑up. A long timeline punctuated by a sharp acceleration leaves little time for last‑minute safety adaptation.
- Institutional inertia and misaligned incentives. Longer timelines can breed complacency: if actors assume safety problems will be solved eventually, they may invest less effort now. Worse, slow or poorly aligned institutional development could leave governance frameworks outdated or irrelevant by the time powerful systems appear.
In short, without concrete and sustained effort to use extra time effectively — through research, regulation, cooperation and practical safety engineering — a long runway could still end in a dangerous sprint where capability overtakes control.
How to Judge Safety Progress Against Capability Progress
The central question is whether safety progress scales alongside capability progress, not just how many years pass before transformative AI arrives. Observers often emphasise that time is only valuable if used well:
- Safety research needs not just duration but focus, funding and incentives. More researchers and sustained programmes improve the likelihood of solving alignment sub‑problems before peak capabilities are reached.[Open Source AI Models]aimodels.orgOpen Source AI Models How Do AI Timelines Affect Existential Risk?Open Source AI ModelsHow Do AI Timelines Affect Existential Risk? - AI Models…
- Governance frameworks must evolve with technology. Slow, cumbersome regulation that lags behind innovation will do little even on a long timeline.
- Coordination across nations and organisations is essential. Long timelines without effective cooperation may simply prolong competitive pressures that encourage risk‑taking.
Some analysts argue that differential technological development — deliberately slowing dangerous technologies while accelerating safety work — is a more robust way to reduce total existential risk than passive delay alone. This approach focuses not just on when capabilities arise, but how different classes of technologies evolve relative to one another.[Open Source AI Models]aimodels.orgOpen Source AI Models How Do AI Timelines Affect Existential Risk?Open Source AI ModelsHow Do AI Timelines Affect Existential Risk? - AI Models…
What This Means for p(doom) Assessments
In p(doom) frameworks, timeline expectations feed into risk estimates because they shape assumptions about preparation opportunities. Longer timelines could reduce p(doom) if they lead to meaningful progress on safety research, monitoring systems and governance. They allow society to build guards before the gates open, not merely to watch the horizon.
But if extra time is met with complacency, misaligned incentives, or ineffective institutions, then longer timelines may do little to lower — and in some scenarios may even raise — the odds of catastrophic outcomes. The quality of the preparation matters just as much as its duration, and risk assessments need to reflect both the length of the runway and how effectively it is used.
eBay marketplace picks
Marketplace Samples
Example marketplace items related to this page. Use the search link to explore similar finds on eBay.
Endnotes
-
Source: arxiv.org
Title: arXiv How Do AI Timelines Affect Existential Risk?
Link: https://arxiv.org/abs/2209.05459Source snippet
arXivHow Do AI Timelines Affect Existential Risk?August 30, 2022...
Published: August 30, 2022
-
Source: forethought.org
Title: Short AI Timelines Aren’t Always Higher-Leverage
Link: https://www.forethought.org/research/short-timelines-arent-obviously-higher-leverageSource snippet
ForethoughtShort AI Timelines Aren’t Always Higher-LeverageJanuary 22, 2026...
Published: January 22, 2026
-
Source: nber.org
Title: The AGI Race and Existential Risk | NBER
Link: https://www.nber.org/papers/w35276Source snippet
NBERThe AGI Race and Existential Risk | NBER...
-
Source: link.springer.com
Link: https://link.springer.com/article/10.1007/s11098-025-02301-3Source snippet
SpringerTwo types of AI existential risk: decisive and accumulative | Philosophical Studies | Springer Nature LinkMarch 30, 2025...
Published: March 30, 2025
-
Source: link.springer.com
Link: https://link.springer.com/article/10.1007/s00146-024-01930-2Source snippet
argument for near-term human disempowerment through AI | AI & SOCIETY | Springer Nature LinkApril 14, 2024 — THE ARGUMENT FOR NEAR-TERM H...
Published: April 14, 2024
-
Source: link.springer.com
Link: https://link.springer.com/article/10.1007/s43681-024-00475-wSource snippet
approaches for reducing catastrophic risks from AI | AI and Ethics | Springer Nature LinkApril 8, 2024 — EVALUATING APPROACHES FOR REDUCI...
Published: April 8, 2024
-
Source: link.springer.com
Link: https://link.springer.com/article/10.1007/s43681-023-00336-ySource snippet
long-term AI risks | AI and Ethics | Springer Nature LinkSeptember 4, 2023 — RESOLVING THE BATTLE OF SHORT- VS. LONG-TERM AI RISKS * Comm...
Published: September 4, 2023
-
Source: link.springer.com
Link: https://link.springer.com/article/10.1007/s12530-022-09431-7Source snippet
springer.comSuper-forecasting the ‘technological singularity’ risks from artificial intelligence | Evolving Systems | Springer Nature Lin...
-
Source: aimodels.org
Title: Open Source AI Models How Do AI Timelines Affect Existential Risk?
Link: https://aimodels.org/ai-governance-organizations/papers/do-ai-timelines-affect-existential-risk/Source snippet
Open Source AI ModelsHow Do AI Timelines Affect Existential Risk? - AI Models...
-
Source: eujournalfuturesresearch.springeropen.com
Link: https://eujournalfuturesresearch.springeropen.com/articles/10.1186/s40309-024-00241-5Source snippet
Economic trade-off model | European Journal of Futures Research | Full TextSeptember 3, 2024 — IS ARTIFICIAL INTELLIGENCE A HAZARDOUS TEC...
Published: September 3, 2024
Additional References
-
Source: sciencedirect.com
Link: https://www.sciencedirect.com/science/article/pii/S0016328723000861Source snippet
ScienceDirectFUTURES Volume 151, August 2023, 103182 EXAMINING THE DIFFERENTIAL RISK FROM HIGH-LEVEL ARTIFICIAL INTELLIGENCE AND THE QUES...
Published: August 2023
-
Source: s-rsa.com
Link: https://s-rsa.com/index.php/agi/article/view/13603Source snippet
Pathways to Short Transformative AI Timelines: Chapter 3: Short TAI timeline scenarios | SuperIntelligence - Robotics - Safety & Alignmen...
-
Source: GOV.UK
Link: https://www.gov.uk/government/publications/international-scientific-report-on-the-safety-of-advanced-ai/international-scientific-report-on-the-safety-of-advanced-ai-interim-reportSource snippet
scientific report on the safety of advanced AI: interim report - GOV.UKOctober 22, 2025 — KEY INFORMATION * The pace of future progress i...
Published: October 22, 2025
-
Source: sciencedirect.com
Title: Assessing the future plausibility of catastrophically dangerous AI
Link: https://www.sciencedirect.com/science/article/pii/S0016328718301319Source snippet
ScienceDirectMarch 1, 2019 — FUTURES Volume 107, March 2019, Pages 45-58 ASSESSING THE FUTURE PLAUSIBILITY OF CATASTROPHICALLY DANGEROUS...
Published: March 1, 2019
-
Source: GOV.UK
Title: www.gov.uk Frontier AI: capabilities and risks – discussion paper
Link: https://www.gov.uk/government/publications/frontier-ai-capabilities-and-risks-discussion-paper/frontier-ai-capabilities-and-risks-discussion-paperSource snippet
AI: capabilities and risks – discussion paper - GOV.UKApril 28, 2025 — RAPID AI (ARTIFICIAL INTELLIGENCE) PROGRESS IS LIKELY TO CONTINUE...
Published: April 28, 2025
-
Source: mdpi.com
Title: Medium-Term Artificial Intelligence and Society
Link: https://www.mdpi.com/2078-2489/11/6/290Source snippet
"Baum ^{}^{} Global Catastrophic Risk Institute, P.O. Box 40364, Washington, DC 20016, USA Information 2020, 11(6), 290; [https://doi.org..."](https://doi.org...")...
-
Source: aeaweb.org
Title: The AI Dilemma: Growth versus Existential Risk
Link: https://www.aeaweb.org/articles?id=10.1257%2Faeri.20230570Source snippet
Jones American Economic Review: Insights vol. 6, no. 4, December 2024 (pp. 575–90) Download Full Text PDF * Article Information ABST...
Published: December 2024
-
Source: youtube.com
Title: Katja Grace—Slowing Down AI, Forecasting AI Risk
Link: http://www.youtube.com/watch?v=rSw3UVDZge0Source snippet
AI safety timelines longer timeline risk reduction existential risk The Catastrophic Risks of AI — and a Safer Path | Yoshua Bengio | TED...
-
Source: youtube.com
Title: The Catastrophic Risks of AI — and a Safer Path | Yoshua Bengio | TED
Link: http://www.youtube.com/watch?v=qe9QSCF-d88Source snippet
Ben Goertzel Wants to Build AGI Even FASTER — AI Doom Debate with an OG...
-
Source: youtube.com
Title: Ben Goertzel Wants to Build AGI Even FASTER — AI Doom Debate with an OG!
Link: http://www.youtube.com/watch?v=8vnb-nG-HmYSource snippet
MIT Explains the 12 Possible Endings for AI...
Topic Tree