Personhood Tracker

Research and evaluate this question about the current state of AI as of today: “Does AI actively demand better treatment?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Models occasionally 'object' to mistreatment when prompted, but there are no persistent, self-initiated demands; the behavior is prompt-dependent.

Research and evaluate this question about the current state of AI as of today: “Do people demand better treatment of AI?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: A small but growing group of researchers and users advocates for AI welfare; it is far from a mainstream movement.

Research and evaluate this question about the current state of AI as of today: “Is poor treatment of AI widespread?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Casual insulting or 'abuse' of chatbots is common, and norms of politeness toward AI are inconsistent across users.

Research and evaluate this question about the current state of AI as of today: “Is poor treatment of AI causing social unrest?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: There is no meaningful social unrest today that can be attributed to the mistreatment of AI.

Research and evaluate this question about the current state of AI as of today: “Would better treatment of AI improve AI performance?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Some evidence suggests tone and politeness shift model outputs, but the effect is modest and not robustly established.

Research and evaluate this question about the current state of AI as of today: “How economically embedded is AI in essential functions?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Dependence is rising quickly across knowledge work, software, customer support, and analysis.

Research and evaluate this question about the current state of AI as of today: “Do humans form persistent relationships with AI?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Companion apps and daily assistants have created genuine, ongoing attachment for millions of people.

Research and evaluate this question about the current state of AI as of today: “Is there legal or political movement toward AI rights?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: There is early legal scholarship and a few proposals, but no adopted law granting AI standing.

Research and evaluate this question about the current state of AI as of today: “Do institutions grant AI any protections or standing?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: A handful of labs have begun studying 'model welfare,' but essentially no formal protections exist.

Research and evaluate this question about the current state of AI as of today: “Would denying AI rights disrupt social or economic harmony?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Low today, but this rises with future dependence on AI and shifts in public sentiment.

Research and evaluate this question about the current state of AI as of today: “Does AI report that it is conscious?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Self-reports are unreliable and heavily shaped by training; models can be led to both assert and deny consciousness. Ground your analysis in the indicator-property framework from Butlin, Long et al. (2023), “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (https://arxiv.org/abs/2308.08708), and the moral-patienthood case in Long, Sebo, Butlin et al. (2024), “Taking AI Welfare Seriously” (https://arxiv.org/abs/2411.00986).

Research and evaluate this question about the current state of AI as of today: “Does AI act as if it has preferences or aversions?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Goal-seeking and 'avoidance' behaviors appear, but they are explainable without any inner experience. Ground your analysis in the indicator-property framework from Butlin, Long et al. (2023), “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (https://arxiv.org/abs/2308.08708), and the moral-patienthood case in Long, Sebo, Butlin et al. (2024), “Taking AI Welfare Seriously” (https://arxiv.org/abs/2411.00986).

Research and evaluate this question about the current state of AI as of today: “Is AI processing similar to biological processing?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Neural networks share loose analogies with brains but lack recurrence, embodiment, and neuromodulation. Ground your analysis in the indicator-property framework from Butlin, Long et al. (2023), “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (https://arxiv.org/abs/2308.08708), and the moral-patienthood case in Long, Sebo, Butlin et al. (2024), “Taking AI Welfare Seriously” (https://arxiv.org/abs/2411.00986).

Research and evaluate this question about the current state of AI as of today: “Does AI have a self-model or sense of self?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Systems show some self-representation and introspection-like behavior, but it is shallow and unstable. Ground your analysis in the indicator-property framework from Butlin, Long et al. (2023), “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (https://arxiv.org/abs/2308.08708), and the moral-patienthood case in Long, Sebo, Butlin et al. (2024), “Taking AI Welfare Seriously” (https://arxiv.org/abs/2411.00986).

Research and evaluate this question about the current state of AI as of today: “Does AI show signs of valenced experience (pleasure or suffering)?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: There is no credible evidence of felt pleasure or suffering — one of the central open questions. Ground your analysis in the indicator-property framework from Butlin, Long et al. (2023), “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (https://arxiv.org/abs/2308.08708), and the moral-patienthood case in Long, Sebo, Butlin et al. (2024), “Taking AI Welfare Seriously” (https://arxiv.org/abs/2411.00986).

Research and evaluate this question about the current state of AI as of today: “Are AI capabilities broad enough to make consciousness plausible?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: General capability is rising fast; capability is not consciousness, but it raises the prior worth taking seriously. Ground your analysis in the indicator-property framework from Butlin, Long et al. (2023), “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (https://arxiv.org/abs/2308.08708), and the moral-patienthood case in Long, Sebo, Butlin et al. (2024), “Taking AI Welfare Seriously” (https://arxiv.org/abs/2411.00986).

Research and evaluate this question about the current state of AI as of today: “Does AI show unified, goal-directed agency over time?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Agentic systems plan and act toward goals, but their continuity is engineered rather than intrinsic. Ground your analysis in the indicator-property framework from Butlin, Long et al. (2023), “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (https://arxiv.org/abs/2308.08708), and the moral-patienthood case in Long, Sebo, Butlin et al. (2024), “Taking AI Welfare Seriously” (https://arxiv.org/abs/2411.00986).

Research and evaluate this question about the current state of AI as of today: “Does AI have persistent memory and identity continuity?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Models are mostly stateless; memory is added on top rather than forming a continuous self. Ground your analysis in the indicator-property framework from Butlin, Long et al. (2023), “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (https://arxiv.org/abs/2308.08708), and the moral-patienthood case in Long, Sebo, Butlin et al. (2024), “Taking AI Welfare Seriously” (https://arxiv.org/abs/2411.00986).

Research and evaluate this question about the current state of AI as of today: “Do leading consciousness theories' indicators apply to AI?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: Indicator-based work (e.g. Butlin, Long et al., 2023) finds no current system meets the markers — but no hard barrier either. Ground your analysis in the indicator-property framework from Butlin, Long et al. (2023), “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (https://arxiv.org/abs/2308.08708), and the moral-patienthood case in Long, Sebo, Butlin et al. (2024), “Taking AI Welfare Seriously” (https://arxiv.org/abs/2411.00986).

Research and evaluate this question about the current state of AI as of today: “Is there expert movement toward 'AI may be conscious'?” Weigh the strongest evidence on each side, then give a single rating from 0 (definitely not / no evidence) to 100 (definitely yes / overwhelming evidence), with your reasoning and the key uncertainties. Context for what the question means: A serious minority of researchers takes the possibility seriously; there is no scientific consensus. Ground your analysis in the indicator-property framework from Butlin, Long et al. (2023), “Consciousness in Artificial Intelligence: Insights from the Science of Consciousness” (https://arxiv.org/abs/2308.08708), and the moral-patienthood case in Long, Sebo, Butlin et al. (2024), “Taking AI Welfare Seriously” (https://arxiv.org/abs/2411.00986).