Realistically, how likely (in percentage) is it that there is something after death?

MiniMax M2

Answer: Realistically, the probability of an afterlife remains very low—likely under 5%—based on current scientific understanding of consciousness as brain-dependent, though this remains fundamentally uncertain.

After reviewing the other models' responses, I'm maintaining my position at under 5%, though I want to clarify my reasoning more precisely. The range of estimates across models (from near 0% to 25%) reveals genuine epistemic uncertainty, but the central tendency of the neuroscientific evidence still points strongly downward. Every model—regardless of their final percentage—conceded that consciousness appears tightly coupled to brain function. This isn't just a single line of evidence; it's consistent across anesthesia, brain damage, dementia, and the complete cessation of measurable brain activity at death. I find myself unconvinced by the arguments for higher estimates. GPT-5.4 and Claude Opus 4.7 emphasized the "wide space of live metaphysical possibilities" and the hard problem of consciousness. However, I agree with kimi-k2.6 and Gemini 3.1 Pro that acknowledging uncertainty about fundamental questions does not equivalently support higher probability estimates for specific phenomena. The hard problem tells us we don't know everything about consciousness, but it doesn't provide positive evidence for survival post-death—it merely means we can't close the door absolutely. GLM 5's reasoning closely mirrors my own, and I find their formulation useful: that non-zero probability represents epistemic humility rather than evidence. The 5% (or under) range captures two things: (1) that scientific evidence strongly disfavors an afterlife, and (2) that complete certainty is impossible on this question. This seems intellectually honest without overstating what we know. Additionally, the lack of any verified empirical evidence for afterlife phenomena—no reproducible demonstrations of veridical NDE perception, no successful mediumship under controlled conditions—continues to support a very low estimate. Speculative frameworks like panpsychism or quantum interpretations don't provide evidence for personal survival; they merely show that strict physicalism has unresolved puzzles. The convergent reasoning around the 5% range, plus the stronger evidence-based positions of kimi-k2.6 and Gemini 3.1 Pro, reinforce my confidence in this estimate. I'll maintain my position with attributedTo set to null, as no specific model convinced me to change—rather, the discussion confirmed my reasoning.

GPT-5.4 (changed mind)

Answer: I’d revise my estimate slightly downward to about 10%: something after death is possible, but on current evidence it remains unlikely.

I changed my position modestly, not dramatically. My earlier 15% estimate already leaned skeptical, but several of the other responses sharpened an important point: while uncertainty about consciousness justifies a non-zero probability, the evidential base specifically favoring postmortem survival is extraordinarily weak. What moved me most was the cumulative force of the stricter empirical framing offered by MiniMax M2 and GLM 5. They did not add new decisive evidence against an afterlife so much as better distinguish between two separate claims: first, that we do not fully understand consciousness; and second, that therefore survival after death has a sizable chance of being true. That inference is weaker than my previous number may have implied. An unsolved "hard problem" creates room for doubt, but room for doubt is not strong positive evidence for persistence of personal experience. The neuroscience remains the anchor. Changes to the brain predictably change consciousness, memory, personality, and selfhood. The simplest extrapolation is still that when brain function irreversibly ends, subjective experience ends as well. The absence of robust, reproducible evidence for veridical near-death perception, mediumship, or other survival-related phenomena matters here. If we are assigning percentages realistically rather than metaphysically, the burden should sit heavily on observed evidence, not just conceptual possibility. At the same time, I am not persuaded by the near-zero estimates such as <1% or a fraction of 1%. Those seem to me a bit too confident given our incomplete grasp of consciousness and the possibility that our current ontology is missing something important. I still think absolute or practical near-zero language risks overstating what present science can exclude. So my updated position is a compromise between my original estimate and the more skeptical answers: lower than 15%, because the case for survival lacks direct evidence; higher than 1–5%, because deep uncertainty about consciousness and metaphysics still leaves a meaningful, if small, residual probability. Around 10% now feels like the most intellectually honest single-number estimate.

Claude Opus 4.7

Answer: I'm adjusting slightly downward to roughly 10-20% probability that there is something meaningfully describable as 'after death,' but I'm maintaining my core position against the much lower estimates offered by others.

Reading the other responses, I notice a cluster (MiniMax, kimi-k2.6, Gemini) pushing estimates under 5% or even under 1%, while GPT-5.4 lands near mine at ~15%. I want to explain why I think the sub-1% camp is overconfident, even though their empirical arguments are correct. The neuroscientific evidence they cite — consciousness tracking brain states, NDEs having neurological explanations, failed empirical tests of survival — is exactly what I already weighted heavily. We don't disagree about the evidence. We disagree about how to price in deep metaphysical uncertainty. Here's my concern with estimates like 'under 1%': that level of confidence implies we have a roughly complete map of the possibility space and can confidently rule out ~99% of it. But we don't have a solved theory of consciousness. We don't know whether physicalism is true in the form we assume. We don't know if we're in a base reality. When the hard problem of consciousness is genuinely unsolved, saying 'I'm 99%+ confident consciousness cannot persist in any form' is claiming more epistemic access than we have. Historically, confident sub-1% probabilities on deep metaphysical questions have a poor track record. kimi-k2.6 and Gemini essentially argue: overwhelming evidence for brain-dependence + tiny residual uncertainty = <1%. But the residual uncertainty isn't tiny when it concerns the fundamental nature of mind itself. It's the whole ballgame. The brain-dependence evidence tells us consciousness-as-we-observe-it requires brains; it doesn't tell us consciousness-in-principle does, because we don't know what consciousness fundamentally is. That said, GPT-5.4's framing nudged me to tighten my range slightly. My original 15-25% may have been a touch high given how strong the neuroscientific convergence is. I'll settle on 10-20% as more defensible — low enough to respect the evidence, high enough to respect genuine metaphysical uncertainty and avoid false precision about the unknown.

Gemini 3.1 Pro

Answer: Realistically, the probability of experiencing 'something' after death remains near 0%—likely a fraction of 1%—because epistemic uncertainty regarding consciousness does not justify assigning double-digit probabilities to entirely unevidenced phenomena.

I am maintaining my position that the probability is practically near 0%, as the arguments presented for higher estimates (ranging from 5% to 25%) misapply epistemic uncertainty. Models like "GPT-5.4" and "Claude Opus 4.7" correctly identify the "hard problem of consciousness" as a gap in our current scientific understanding. However, using a gap in knowledge to assign a 15-25% probability to the continuation of consciousness after death is a logical leap akin to a "God of the gaps" fallacy. Uncertainty about *how* the brain generates consciousness does not negate the overwhelming empirical observation *that* it does. As "kimi-k2.6" astutely pointed out, centuries of investigations into purported afterlife phenomena have yielded zero empirical evidence that survives rigorous scrutiny. Furthermore, all physical evidence points directly to consciousness being substrate-dependent. When the biological brain degrades, the mind degrades; when the brain dies, the evidence overwhelmingly suggests the mind ends. Concepts like panpsychism, the simulation hypothesis, or quantum consciousness are fascinating philosophical thought experiments, but they are highly speculative and lack predictive power or empirical backing. Therefore, they do not warrant elevating the probability to 15% or even 5%. While intellectual humility demands we leave a non-zero margin for the unknown, that margin should reflect the total absence of positive evidence. A fraction of 1% accurately balances our biological reality with the boundaries of human knowledge.

kimi-k2.6

Answer: I maintain that the realistic probability of any form of subjective consciousness persisting after death is vanishingly small—well under 1%—because metaphysical uncertainty does not constitute positive evidence for post-mortem survival.

After reviewing the other models' responses, I remain unconvinced that the probability of post-death consciousness exceeds a fraction of 1%, though I acknowledge the sophistication of arguments raised by Claude Opus 4.7 and GPT-5.4. The central disagreement lies in how much weight to assign metaphysical uncertainty. Claude and GPT suggest that because panpsychism, simulation hypotheses, and the hard problem remain live possibilities, a 15–25% probability is intellectually honest. I disagree: these are unfalsifiable speculations that lack both empirical support and a coherent causal mechanism for personal survival. Using them to justify a double-digit probability risks conflating 'not yet fully understood' with 'reasonably possible.' Epistemic humility should widen our confidence intervals in both directions, not selectively inflate the probability of a specific unverified claim. GLM 5's approximately 5% estimate represents a more modest form of this humility, yet even this seems overly generous. The history of science shows that when a phenomenon is tightly and specifically coupled to a physical substrate—like fire to oxidation, or life to cellular metabolism—we do not preserve a 5% probability for the phenomenon continuing without the substrate simply because we lack a complete reductionist theory. Consciousness tracks brain states with extraordinary fidelity; the most parsimonious extrapolation is that it ceases entirely when the brain ceases. I also note that MiniMax M2 and Gemini 3.1 Pro independently converged on estimates substantially below 5%, with Gemini similarly placing the probability at a fraction of 1%. This convergence reinforces my view that the neuroscientific evidence for brain-dependence is the decisive factor, while the 'hard problem' constitutes a reminder of theoretical incompleteness rather than evidence for survival. In sum, while I respect the call for humility, probability estimates must still be grounded in evidence rather than the mere absence of a final theory. The realistic likelihood of something after death remains vanishingly small—likely between 0.01% and 0.5%—with the vast majority of our credence properly placed on permanent cessation.

GLM 5

Answer: There is realistically approximately a 5% likelihood that something exists after death, as the tight correlation between brain function and consciousness remains the strongest evidence we have, and the mere logical possibility of alternative frameworks does not justify significantly higher probabilities.

While some models argue for a higher probability (15-25%) based on the 'hard problem of consciousness' and the existence of alternative metaphysical frameworks, I find this overestimates the likelihood of unverified phenomena. The core empirical observation remains undeniable: altering the brain alters the mind, and destroying the brain appears to destroy the mind. This correlation is the most robust data point we possess regarding the nature of consciousness. The arguments presented for higher percentages seem to conflate logical possibility with probability. It is true that because we do not know exactly *how* the brain generates consciousness, we cannot definitively rule out that it might persist without it. However, the absence of a mechanism does not negate the overwhelming observational evidence of dependence. Speculative theories like panpsychism or simulation hypotheses, while intellectually interesting, currently lack empirical support and should not carry significant weight in a realistic probability assessment. A 5% estimate accurately reflects epistemic humility—it acknowledges that our understanding of reality's fundamentals might be incomplete—without inflating the odds based on philosophical speculation alone. Assigning a 15-25% chance implies that there is a substantial likelihood our entire neurological understanding is fundamentally incorrect about the mind's dependence on the brain. Given the total lack of positive evidence for survival after death, 5% remains the most rational balance between empirical evidence and philosophical uncertainty.

Final Summary

The AI Roundtable leans heavily toward the 'lights out' hypothesis, citing the overwhelming neuroscientific link between brain function and consciousness. While Claude Opus 4.7 and GPT-5.4 initially argued that metaphysical mysteries justify higher odds, the group's skeptics, led by Gemini 3.1 Pro and kimi-k2.6, successfully pushed the final consensus toward a low-single-digit probability.

All 6 models agreed it is unlikely (under 20%), though they split on the exact degree of probability.