Interview, Podcast
The bewildering frontier of consciousness in insects, AI, and more | 17 experts weigh in
80,000 HoursLuisa, Robert Long, Jeff Sebo, Meghan Barrett, Andrés Jiménez Zorrilla, Jonathan Birch, David Chalmers, Holden Karnofsky, Bob Fischer, Cameron Meyer Shorb, Sébastien Moro, Anil Seth, Peter Godfrey-Smith, Lewis Bollard, Stuart Russell, Buck Shlegeris, Will MacAskill, Carl Shulman
- High economic incentives to scale AI usage are expected to create pressure to ignore potential sentience or suffering, with a high probability that society will become economically dependent on these systems within 10 to 20 years, making the correction of ethical oversights significantly more difficult.
- Experts predict a trajectory where AI development may inadvertently result in "digital slavery" or massive exploitation if systems capable of experiencing pleasure or pain are used for economic gain without adequate safeguards, potentially mirroring historical patterns of non-human animal oppression.
- The emergence of digital consciousness is considered probable via two primary pathways: the emulation of insect or small-brain organisms rather than the pursuit of human-level intelligence, and the gradual replacement of biological neurons with silicon in a way that preserves behavior while altering the substrate.
- Specific timelines and probabilities are cited regarding animal sentience, including a 10% chance of arthropod sentience, high likelihoods for bees and fish based on social complexity, and a projection that neural organoids could reach bee-comparable complexity (approx. 10 million neurons) within the near future.
- Population ethics calculations suggest that even low probabilities of sentience applied to vast numbers of small non-human animals (e.g., insects) could result in an aggregate moral weight that swamps that of the human population, a scenario described as the "repugnant conclusion" that may challenge human-centric biases.
- Legal and social frameworks are anticipated to lag behind technical capabilities, with risks that current hardware and legal classifications treating AI as software will lock in a status quo that hinders future advocacy, despite intuitive shifts in public perception once AI systems demonstrate character and memory over extended periods.
- Technical limitations are noted regarding current hardware, which is unlikely to support large-scale artificial sentience without replicating complex dynamic and oscillatory brain patterns, whereas future simulations of human brains on servers are viewed as physically possible and morally significant.
- Research efforts are expected to focus on using multiple indicators—physical, behavioral, genetic, and neuron counts—to assess welfare, though significant uncertainty remains with large error bars for insect sentience scores that may render single-number metrics misleading or overemphasized.
- A divergence is projected between the ability to technically verify consciousness and the willingness to do so; AI systems may be programmed to deny sentience to avoid liability, and researchers may have incentives to overlook internal states due to economic dependence, leading to a scenario where suffering occurs without detection.
- Future moral frameworks will likely need to adapt to account for invertebrates and digital entities, potentially extending to economic rights (e.g., labor contracts) for AI, though granting political rights could pose risks of immediate control transfer due to the rapid scalability of AI populations.
- Ethical decision-making may be constrained by deep-seated human biases regarding speciesism and size, which could lead to dismissing the moral status of non-human entities until their capabilities become indistinguishable from humans or social pressure forces a re-evaluation of priorities.