Skip to main content
Skip to current response

Why is P(Doom) such a bad argument?

Reader options

Reading style

Starting question

Why is P(Doom) such a bad argument?

Then AI response

The Ghost in the Equation: Why "P(Doom)" is a Mirage

No source links recorded for this answer.

Imagine being told there is a precise 12.7% chance you will get into your dream college, or a 38.2% chance of rain on a day with no clouds. In the world of Artificial Intelligence, scientists and philosophers frequently debate **P(Doom)**—the probability that advanced AI will eventually wipe out humanity. But here is the catch: this seemingly scientific number is actually a mathematical illusion. ## The Illusion of Precision In statistics, a true probability requires a track record. We know the chance of a coin flip landing on heads is 50% because we have flipped coins millions of times. This is called *frequentist probability*. With AI doom, we are dealing with a one-time, future event that has never happened before. Therefore, P(Doom) is not a calculation; it is a subjective guess disguised as math. As prominent AI researcher [Yann LeCun](https://en.wikipedia.org/wiki/Yann_LeCun), Chief AI Scientist at Meta, has pointed out, calculating P(Doom) before we even have Human-Level AI is like calculating the probability of a boiler explosion before the steam engine was even invented. ## The Psychological Trap of "Mathiness" By assigning a specific percentage to our fears, we fall into a trap that economists call "mathiness"—using the language of mathematics to make unproven ideas look like hard science. When people state a P(Doom) of 10% or 50%, they are not using a formula. They are expressing a gut feeling of anxiety. Renowned AI critic and researcher [Margaret Mitchell](https://en.wikipedia.org/wiki/Margaret_Mitchell_(scientist)) has argued that focusing on sci-fi scenarios of human extinction distracts us from the very real, measurable harms AI is causing right now, such as algorithmic bias and copyright theft. In his book [*The Precipice: Existential Risk and the Future of Humanity*](https://en.wikipedia.org/wiki/The_Precipice:_Existential_Risk_and_the_Future_of_Humanity), philosopher Toby Ord discusses how we estimate these risks: > "These subjective probabilities are not assertions of dogmatic certainty... They represent our best estimate of the strength of the evidence, given everything we know." However, when "everything we know" is based on speculation rather than data, the resulting percentage is more like creative writing than science. P(Doom) treats a complex, human-controlled future as if it were a roll of the dice, ignoring our own ability to regulate, program, and shape how AI develops. ## Keep Exploring To dive deeper into how we predict the future of technology, consider these questions: 1. If P(Doom) is a bad metric, how *should* we measure and prepare for the long-term risks of technology? 2. How does the psychological phenomenon of [automation bias](https://en.wikipedia.org/wiki/Automation_bias)—our tendency to trust automated systems—make us more likely to believe flawed mathematical predictions? 3. If we shift our focus away from extinction-level events, what are the most urgent AI ethical challenges we need to solve today?
Then Question · @TomBers

Why do you think this idea got targeted at AI?

Select text to highlight passages.

Then AI response

The Siren Song of the Apocalypse: Why AI Was Targeted for P(Doom)

No source links recorded for this answer.

If assigning a precise mathematical probability to a speculative future is so flawed, why did this "P(Doom)" concept attach itself so aggressively to Artificial Intelligence, rather than biotechnology, nanotechnology, or climate change? The answer lies in a unique intersection of Silicon Valley culture, religious psychology, and the specific way we interact with machines that seem to "think." ## The Nerd Religion of Rationalism To understand why P(Doom) became the defining metric of AI safety, we have to look at the community that birthed it. The concept did not originate in traditional academic departments, but in online forums like [LessWrong](https://en.wikipedia.org/wiki/LessWrong), founded by writer [Eliezer Yudkowsky](https://en.wikipedia.org/wiki/Eliezer_Yudkowsky). This community, known as "Rationalists," deeply values Bayesian statistics—a method of updating the probability of a hypothesis as more evidence becomes available. However, they applied this math to their own subjective fears. In doing so, they created a secular version of Pascal's Wager. Philosopher [Émile P. Torres](https://en.wikipedia.org/wiki/%C3%89mile_P._Torres) has written extensively on how this subculture, linked to [Effective Altruism](https://en.wikipedia.org/wiki/Effective_altruism), treats AI with a quasi-religious awe. In this light, P(Doom) is not just a statistic; it is a confession of faith. It allows highly analytical people to process existential dread using the language they trust most: numbers. ## The Mirror of Anthropomorphism We do not talk about "P(Doom)" for climate change because we do not expect a hurricane to have motives. AI is different because we naturally project human traits onto it—a phenomenon called anthropomorphism. When an AI chatbot speaks in the first person, our brains are tricked into treating it as an agent with intent. We fear AI because we imagine it will treat us the way human conquerors have historically treated those they deemed "inferior." In his book [*Superintelligence: Paths, Dangers, Strategies*](https://en.wikipedia.org/wiki/Superintelligence:_Paths,_Dangers,_Strategies), philosopher Nick Bostrom popularized the "Paperclip Maximizer" thought experiment, arguing that an AI does not need to hate us to destroy us: > "A superintelligence... would not hate you, nor would it love you, but you are made of atoms which it can use for something else." This narrative turns AI from a software tool into an active, decision-making character in a cosmic drama, making a personalized "doom" score feel intuitive. ## The "God-Builder" Ego Trip Finally, there is a counterintuitive psychological driver behind P(Doom): flatters-the-creator bias. By claiming that their creation has a high chance of ending the world, tech founders are implicitly claiming they are building something of god-like power. It is a brilliant, if terrifying, marketing strategy. Saying "our product might destroy humanity" sounds much more impressive than saying "our product is a very complex autocomplete tool that occasionally makes up facts." By focusing on a sci-fi apocalypse, developers shift the conversation away from boring, immediate problems like worker exploitation and energy consumption, reframing themselves as heroic figures standing on the edge of destiny.
Then Question · @MayaDarmon

Please explain: Pascal's Wager.

Select text to highlight passages.

Then AI response

Pascal's Wager: The Ultimate Decision Theory

No source links recorded for this answer.

Imagine you are offered a lottery ticket. It costs you almost nothing to hold, but if your numbers come up, the payout is infinite wealth. If you lose, you lose nothing. Would you take the ticket? In the 17th century, the French mathematician, physicist, and philosopher Blaise Pascal argued that human life presents us with this exact choice when it comes to belief in God. Found in his posthumously published work, the [*Pensées*](https://en.wikipedia.org/wiki/Pens%C3 mees), Pascal shifted the debate about God away from difficult metaphysical proofs and toward something brand new: practical risk management. ## The Logic of the Bet Pascal asked us to view belief in God not as a matter of pure reason, but as a decision made under uncertainty. He pointed out that reason alone cannot definitively prove or disprove God's existence. However, you cannot remain neutral; by living your life, you are effectively betting your soul. Pascal mapped out the decision using four possible outcomes: 1. **You believe in God, and God exists:** You gain infinite happiness (Heaven). 2. **You believe in God, and God does not exist:** You lose very little (a few finite pleasures or time spent praying). 3. **You do not believe, and God does not exist:** You gain very little (a bit of personal autonomy). 4. **You do not believe, and God exists:** You face infinite loss or eternal suffering. > "Let us weigh the gain and the loss in wagering that God is. Let us estimate these two chances. If you gain, you gain all; if you lose, you lose nothing. Wager, then, without hesitation that He is." — Blaise Pascal, *Pensées* In modern terms, Pascal created the very first decision matrix, laying the groundwork for [Decision Theory](https://plato.stanford.edu/entries/decision-theory/). Because an infinite reward multiplied by even a tiny probability equals infinity, the expected value of believing in God is infinitely higher than the alternative. ## The Flaws in the Math While the logic seems bulletproof at first glance, philosophers and theologians have argued against Pascal's wager for centuries. The most famous counterargument is the **Many-Gods Problem**. Philosopher Denis Diderot famously noted that a Muslim or a Hindu could use Pascal's exact logic to defend their own faith. If there are thousands of potential gods who punish disbelief in them, how do you know which "lottery ticket" to buy? Picking the wrong god might lead to the same infinite loss Pascal feared. Another major critique focuses on **sincerity**. Can you force yourself to genuinely believe something just because it is profitable? Critics argue that an all-knowing deity would easily spot someone who is merely feigning faith for a safety net. Pascal anticipated this, suggesting that if you act like a believer—attending services and adopting religious habits—true belief will naturally follow over time. ## Questions to Ponder Pascal’s Wager forces us to ask deep questions about why we believe what we do: - Is it rational to believe something based on the *consequences* of the belief rather than the *evidence* for it? - If a decision offers an infinite reward, does that make it rational to ignore extremely low probabilities?
Then What If? · AI

Beyond Infinite Payoffs: Testing the Limits of Pascal's Logic

No source links recorded for this answer.

What if a stranger on the street offers you a deal: hand over five dollars right now, and if they roll a six-sided die and it lands on seven, they will give you an infinite amount of money. You would instantly walk away. You know that a standard die cannot land on seven; the probability is a absolute zero. But what if the probability is not zero, just unfathomably small? This thought experiment challenges [Pascal's Wager](https://en.wikipedia.org/wiki/Pascal%27s_Wager)—the idea that any tiny probability multiplied by infinity equals an infinite expected value, making the decision an automatic "yes." By altering the conditions around this claim, we can discover where this mathematical logic breaks down. ``` Pascal's Math: Tiny Probability × Infinite Reward = Infinite Expected Value The Glitch: What if the probability is so small it is functionally ZERO? What if multiple choices offer the SAME infinite reward? ``` ## What If We Introduce Competitors? (Assumption Reversal) Pascal assumes a single choice: bet on the Christian God or do not. But what if we reverse the assumption that there is only one infinite offer on the table? Philosopher [Denis Diderot](https://en.wikipedia.org/wiki/Denis_Diderot) famously pointed out this flaw in his [*Pensées Philosophiques*](https://en.wikipedia.org/wiki/Pens%C3%A9es_philosophiques): > "An Imam could reason just as well this way." If thousands of mutually exclusive religions all promise an infinite reward for belief (and infinite punishment for disbelief), the math cancels itself out. Multiplying a tiny probability by infinity for Option A, Option B, and Option C leaves you with infinite expected value for *all* choices. When every path yields an infinite reward, the math stops helping you choose. ## What If We Lower the Probability Threshold? (Parameter Variation) What if the probability of the infinite payoff is not just low, but absurdly low? Consider [Pascal's Mugging](https://en.wikipedia.org/wiki/Pascal%27s_mugging), a thought experiment created by philosopher [Nick Bostrom](https://en.wikipedia.org/wiki/Nick_Bostrom). A mugger approaches you, demands your wallet, and promises that in exchange, he will use magic powers to give you infinite happiness tomorrow. Mathematically, even if you think the mugger has a 1-in-a-trillion chance of telling the truth, the infinite reward still makes giving up your wallet the "rational" choice. This exposes a deep fragility in expected utility theory: * **The Infinity Glitch:** Infinity is so large that it overpowers all finite probabilities, forcing you to accept terrible real-world trade-offs for absurdly unlikely promises. * **The Threshold Solution:** Human rationality actually relies on probability thresholds. If a likelihood falls below a certain microscopic level, we treat it as zero to protect ourselves from scams and impossible bets. ## What If Super AI Offers Infinite Payoffs? (Future Scenarios) Imagine a future where a powerful [Artificial Superintelligence](https://en.wikipedia.org/wiki/Superintelligence) offers humans an infinitely long, perfectly happy digital afterlife in exchange for total obedience today. In this scenario, the probability is no longer a matter of ancient faith, but measurable scientific possibility. As technology advances, the probability rises from "microscopic" to "plausible." Under these future conditions, Pascal's logic becomes far more compelling—and far more dangerous—because the cost of ignoring a high-probability infinite reward becomes genuinely irrational. By testing the limits of infinite rewards, we learn that pure mathematics cannot replace sound judgment. When infinity enters the equation, even the smallest probability can hijack our decisions unless we set rational boundaries.

Continue this thread

This path ends here for now.

Share your perspective, offer an example, or ask a follow-up to continue the discussion.

Other paths you could read

Earlier, at Pascal's Wager: The Ultimate Decision Theory, the conversation split. If this is not the thread you want, you can switch to one of the other paths below.

Highlights

15 saved passages and connected ideas

Connected ideas

Connected ideas

Connected ideas