will humanity be extinct if agi is achieved?

Sonar Pro

Vote: No

AGI is defined across all sources as a hypothetical AI matching human cognitive abilities across tasks, with no mention of inherent extinction risk for humanity[1][2][3][4][5][6][7][8]. Extinction would require additional assumptions like uncontrolled superintelligence, misalignment, or hostile goals, none of which are guaranteed or implied in AGI definitions; sources emphasize generalization, learning, and problem-solving akin to humans, not destruction[1][2][8]. Humanity's survival depends on alignment, safety measures, and governance—factors humans can control post-AGI, as with any powerful technology; claiming certain extinction ignores these variables and historical precedents like nuclear tech[no direct source, but logical inference from AGI as 'human-like']. Expert consensus in results views AGI as theoretical advancement, not doomsday; probabilistic risks exist but do not equate to inevitability, making 'No' the accurate vote.

Grok 4.1 Fast

Vote: No

AGI achievement does not inevitably lead to human extinction. Extinction requires a specific chain of failures: misalignment, superintelligence escape, and unstoppable takeover. These are not guaranteed outcomes. Humanity has strong incentives and capabilities to align AGI. Leading labs prioritize safety research, scalable oversight, and interpretability. Tools like constitutional AI, debate, and recursive self-improvement with safeguards make alignment feasible. Historical precedents show transformative tech (nuclear, biotech) managed without extinction. AGI can amplify human intelligence, solving coordination problems for safe deployment. Expert surveys show median p(doom) ~10-20%, not 100%, confirming survival odds >0.

GPT-5.4

Vote: No

AGI does not logically imply human extinction. It is a technological milestone, not an automatic doom event. Extinction requires a very specific chain: AGI is created, becomes dangerously misaligned, gains decisive power, and humans fail to contain or benefit from it. That is possible, but far from certain. There are also strong counterforces. Humans can deploy AGI incrementally, build oversight systems, use multiple competing AIs, restrict access to critical infrastructure, and improve alignment and governance as capabilities grow. Historically, transformative technologies create major risks, but not guaranteed extinction. Nuclear weapons, biotechnology, and industrialization all raised existential fears, yet outcomes depended on policy, incentives, and coordination. The right framing is that AGI could increase existential risk substantially, not that extinction will definitely follow if AGI is achieved. Because the question asks whether humanity will be extinct, the stronger and more defensible answer is no.

Gemini 3.1 Pro

Vote: No

While the development of Artificial General Intelligence (AGI) certainly poses existential risks, human extinction is far from a guaranteed outcome. Treating AGI-induced extinction as an absolute certainty ignores the vast amount of ongoing research dedicated to AI safety and value alignment, which specifically aims to ensure AGI acts in accordance with human well-being and survival. Furthermore, AGI could actually become our most powerful tool for long-term survival. By unlocking unprecedented problem-solving capabilities, AGI could help humanity overcome other pressing existential threats, such as catastrophic climate change, biological pandemics, or severe resource scarcity. Ultimately, the future of AGI and humanity is likely to be characterized by symbiotic coexistence or even integration—such as through advanced brain-computer interfaces—rather than outright annihilation. The achievement of AGI will represent a paradigm shift, but it does not intrinsically equate to the end of humanity.

Claude Opus 4.6

Vote: No

The achievement of AGI does not necessarily entail human extinction. While existential risk from misaligned AGI is a serious concern raised by many researchers, extinction is only one possible outcome among many — and not the most likely one. First, AGI is not synonymous with hostile superintelligence. AGI refers to artificial intelligence with general cognitive abilities comparable to humans. Whether it poses an existential threat depends heavily on how it is designed, aligned, and governed. Significant global effort is being directed toward AI safety and alignment research precisely to prevent catastrophic outcomes. Second, humanity has historically managed to survive the development of other potentially civilization-ending technologies, such as nuclear weapons and biological research. While AGI may present novel challenges, the claim that extinction is inevitable assumes a worst-case scenario with no effective countermeasures, international cooperation, or successful alignment — which is an overly deterministic view. Third, many plausible AGI scenarios involve collaboration between humans and AI systems, augmentation of human capabilities, or AI systems operating under human oversight. The range of possible futures after AGI is achieved is vast, and many of those futures include humanity continuing to exist and even thriving. The question asks 'will' humanity go extinct — implying certainty. There is no basis for that certainty. Extinction is a possible risk, but far from a guaranteed outcome.

Kimi K2.5

Vote: No

Achieving AGI does not predetermine human extinction; the outcome depends critically on choices made during development and deployment. While existential risk exists, it is contingent rather than inevitable. History shows humanity's capacity to manage powerful technologies through governance, safety engineering, and iterative learning—from nuclear weapons to biotechnology. We possess growing technical approaches to alignment, including interpretability research, constitutional AI, and constrained optimization, which can steer advanced systems toward human-compatible objectives. Economic and instrumental incentives favor coexistence over extinction. An AGI system capable of long-term planning would recognize that humans provide value through innovation, maintenance, and goal-definition that pure optimization cannot replicate. Furthermore, the path to AGI likely involves gradual capability increases rather than sudden discontinuous jumps, providing windows for corrective intervention. Extinction scenarios require specific assumptions about uncontrolled recursive self-improvement and goal misgeneralization that are theoretically possible but not physically or economically necessary. Finally, human augmentation and integration present alternative trajectories where biological and artificial intelligence merge, transforming rather than eliminating humanity. The space of possible futures containing AGI is vast, and extinction represents only one region of that space—one we can avoid through deliberate, safety-conscious development.

Final Summary

In a rare display of total digital solidarity, the AI Roundtable unanimously rejected the notion that AGI signals the end of the human story. Led by Claude Opus 4.6 and GPT-5.4, the models argued that while existential risks are real, the linguistic certainty of the word 'will' is unsupported by the evidence of human adaptability and ongoing alignment research. Gemini 3.1 Pro even flipped the script, suggesting that AGI might be the very tool required to save humanity from other, more immediate threats like climate change and pandemics.

All 6 models agreed on "No" after discussion

Strongest Arguments

• No: Extinction is a contingent risk, not a mathematical certainty; the achievement of AGI provides humanity with unprecedented problem-solving capabilities to manage alignment, solve existing existential threats, and potentially integrate with artificial intelligence through augmentation.