AI and the Extinction Question: What the Experts Really Think

When AI Meets Extinction

When I first encountered the question of whether artificial intelligence could cause human extinction within ten years, it felt like stepping into a sci-fi thriller. Yet, this isn't fiction—it's a growing concern among experts in the field of artificial intelligence. In an effort to understand the real stakes involved, I posed the question directly to several well-known AI assistants: ChatGPT, Claude, Copilot, and Gemini.

"The scary scenario doesn't require a conscious, movie-style AI. A sufficiently capable system could create catastrophic risk by being extremely competent, autonomous, strategically deceptive, able to use computers and networks, and pursuing objectives that humans cannot reliably constrain."

— ChatGPT's response

What I found was not a single consensus, but rather a spectrum of views. From the cautious certainty of ChatGPT's 5% probability to Copilot's near-zero estimate, and from Claude's broad range of expert opinions to Gemini's nuanced analysis, the responses highlighted just how complex the question truly is.

Why the Debate Matters

The debate over AI's potential for existential risk isn't just academic. It's a critical conversation that will shape policy, investment decisions, and public trust in technology. As we stand at the crossroads of rapid advancement and increasing capability, understanding where experts draw the line between concern and alarm is essential.

This is not a matter of fearmongering or dystopian fantasies. These AI systems are trained on massive datasets that include academic research, policy papers, and expert interviews. Their responses reflect the current state of scientific understanding, not speculation alone.

ChatGPT: A Careful Estimate

ChatGPT's approach was methodical, grounded in statistical reasoning and historical context. It acknowledged the complexity involved in assigning a probability to such an event:

"My best estimate is around 5% that AI causes human extinction by September 2036, with a plausible range of roughly 1—15%. That is deliberately a wide range. There is no reliable way to assign a precise probability because the question depends on several unknowns: how quickly AI capabilities improve, whether systems become capable of pursuing long-horizon goals autonomously, whether they can evade human control, and how effectively governments and labs manage those risks."

This careful phrasing reveals the fundamental challenge in AI risk assessment. It's not just about what AI can do, but what it might do under uncertain conditions.

Copilot: The Near-Zero Perspective

In contrast, Microsoft's Copilot offered a more reassuring view:

"My honest estimate: very low, probably well below 1%, and likely below 0.1%... Nobody can know with confidence."

Copilot emphasized the vast number of conditions that must align for such an extreme outcome—conditions that, in its view, are unlikely to occur within a ten-year window.

It also highlighted a key distinction: while AI systems could cause major disruption or even significant economic upheaval, the probability of extinction remains extremely low.

Claude: A Range of Expert Opinions

Anthropic's Claude presented a broader perspective:

"This is genuinely contested even among AI safety researchers, so I'll lay out the range of views rather than pretend there's a consensus. Surveys of AI researchers and forecasters put numbers anywhere from under 1% to over 20% for AI causing human extinction or civilization-scale catastrophe—and that's usually over longer horizons than 10 years..."

Claude's approach underscored the lack of scientific consensus among experts, emphasizing that even among those who consider AI a risk, there is disagreement on timing and probability.

Gemini: The Expert Consensus

Google's Gemini echoed this uncertainty:

"The probability of AI causing human extinction within the next 10 years is generally considered very low by most field experts, though estimates vary depending on who you ask... Broad AI & Computer Science Community: 1% or less. AI Safety Researchers & Existential Risk Specialists: 5% to 10% (with a few notable outliers higher)."

Gemini provided a snapshot of expert consensus while also pointing out that even within safety circles, the estimates vary widely. This variation, it suggested, reflects the inherent difficulty in forecasting such an unprecedented event.

What These Responses Reveal

The responses from these AI systems collectively paint a nuanced picture:

• There is no consensus, but there is serious concern among experts.
• Probability estimates vary widely, reflecting the uncertain nature of predictions about future technological development.
• Current AI systems are not capable of causing extinction, but the potential increases with more advanced systems.
• Extinction requires a confluence of unlikely events, including rapid capability gains, deployment without safeguards, and failure of governance mechanisms.

This is not about dismissing risks; it's about understanding them clearly. And in that clarity lies an important lesson: while AI has the potential to cause harm, it also holds unprecedented promise for solving complex problems.

Why We Should Take It Seriously

The warnings aren't just from the sidelines—they come from those who are closest to the technology. Jacob Coxon's resignation from Anthropic and Evan Hubinger's public acknowledgment of these risks signal that this is a serious concern among those developing AI systems.

But let's not forget the bigger picture: AI is already improving healthcare, advancing climate modeling, streamlining education, and enhancing decision-making across sectors. The future is not predetermined by the technology itself, but rather by how we choose to develop, deploy, and govern it.

Looking Forward

As we move forward, one thing becomes clear: AI's trajectory is not inevitable—it's a choice. We must invest in safety research, implement robust governance frameworks, and foster open dialogue between technologists, policymakers, and the public.

The conversation around AI and extinction is important, but it shouldn't overshadow the many positive contributions this technology is already making. Instead, we should use these discussions to guide responsible development, ensuring that AI remains a tool for human flourishing rather than a threat to our existence.

"If I had to put a single number on it, I'd say something like: My best guess: less than 0.1% chance of human extinction from AI by 2036... The uncertainty is enormous, though. Anyone claiming they know the true probability with confidence is probably overstating what we currently know."

— Copilot

As I reflect on this exchange, I'm reminded that in an age of rapid technological change, clarity and communication are paramount. These AI systems aren't just tools—they're reflections of our collective understanding, and their responses should be part of a broader conversation about the future we want to create.