Subjective Experience in AI Systems: What Do AI Researchers and the Public Believe?
Organizations: Centre for the Governance of AI · University of Oxford · New York University · University of California, Berkeley ·
Abstract
Whether AI systems are perceived to have subjective experience has scientific, ethical, and governance implications. Yet little is known about how AI researchers and the public view this possibility. We surveyed 582 AI researchers who published in leading AI venues and 825 nationally representative US adults about the likelihood, identification, and governance of AI systems with subjective experience. We find that both AI researchers and the public think it is unlikely that AI systems have subjective experience now, but believe it is more likely than not they will by century's end. Both groups believe multidisciplinary expertise will be important in identifying AI subjective experience, support developers implementing safeguards now, and agree that, if created, subjectively experiencing AI systems should have moral responsibilities and be held accountable. Both groups also show substantial disagreement on many moral and policy questions. Our results underline the importance of building more detailed, evidence-based understanding of public and expert opinion on this contentious issue.
Figures & tables
| Definitions used in the survey | Explanation |
|---|---|
| Artificial intelligence (AI) | Computer systems that perform tasks or make decisions that usually require a human to do them. AI systems can perform these tasks or make these decisions without explicit human instructions. |
| Subjective experience | The ability to experience the world from a single point of view, including experiences such as perceiving (for example, visual or auditory experiences) and feeling (for example, the experience of pleasure and pain). For an AI system to have subjective experience, it would mean it has an internal experience from its own perspective – there is something it is like to be that AI system. |
Appendix figures & tables55 assets
Supplementary material from the paper’s appendix.
Appendix
| Summary Statistics | |||||||
|---|---|---|---|---|---|---|---|
| Year | Sample | Total mean | Total median | Fixed date mean | Fixed date median | Fixed probability mean | Fixed probability median |
| 2025 | Both groups | 12.0 | 1.3 | 19.8 | 7.5 | 4.2 | 0.1 |
| 2025 | AI researchers | 10.8 | 1.1 | 17.5 | 3.8 | 3.7 | 0.1 |
| 2025 | US public | 12.8 | 1.7 | 21.5 | 9.4 | 4.6 | 0.1 |
| 2030 | Both groups | 24.3 | 10.0 | 31.8 | 22.9 | 16.9 | 10.0 |
| 2030 | AI researchers | 22.5 | 10.0 | 29.3 | 18.7 | 15.2 | 8.3 |
| Experiential and social mental capacities | |
|---|---|
| Perception and awareness | Capable of perceiving and being aware of the world. |
| Valenced experience | Capable of experiencing positive states like pleasure and happiness or negative states like pain and suffering. |
| Subjective experience a | Capable of having experiences that feel like something from a single point of view. |
| Emotions | Capable of having and experiencing emotions. |
| Cognitive and agentic mental capacities | |
| Motivation and action | Capable of having and acting on desires, preferences, and other motivational states. |
| AI researchers | Public | Analysis | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Topic | Question | M | SD | n | M | SD | n | Sig.? | ||||
| Agreement AI systems with subjective experience should | Have “computing rights” | -0.2 | 1.3 | 219 | 0.2 | 1.4 | 319 | -3.2 | 493.0 | 0.001 | 0.048 | yes |
| Agreement AI systems with subjective experience should | Be cared for and protected, as people treat their pets | 0.3 | 1.3 | 215 | 0.4 | 1.3 | 322 | -1.2 | 472.4 | 0.248 | 1.000 | no |
| Agreement AI systems with subjective experience should | Be respected and treated the same as other people | 0.0 | 1.4 | 221 | 0.2 | 1.5 | 334 | -1.7 | 488.1 | 0.087 | 1.000 | no |
| Agreement AI systems with subjective experience should | Have protection under the law from harm and mistreatment | 0.6 | 1.3 | 215 | 0.3 | 1.5 | 322 | 2.1 | 503.8 | 0.037 | 1.000 | no |
| Agreement AI systems with subjective experience should | Have autonomy in its programming to act freely and not be under the control of others | -0.6 | 1.4 | 224 | -0.4 | 1.5 | 323 | -1.2 | 498.8 | 0.224 | 1.000 | no |
| Sample | Framing | Year/Probability | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|---|---|
| AI researchers | Fixed date framing | 2024 | 12.4 | 1.0 | 22.2 | 1.4 | 0.1 | 10.0 | 9.8 | 15.1 | 265 |
| AI researchers | Fixed date framing | 2034 | 34.0 | 25.0 | 31.6 | 1.9 | 5.0 | 60.0 | 30.2 | 37.9 | 265 |
| AI researchers | Fixed date framing | 2100 | 61.2 | 70.0 | 34.3 | 2.1 | 30.0 | 95.0 | 57.1 | 65.4 | 265 |
| AI researchers | Fixed probability framing | 10 | 2059.5 | 2030.0 | 158.8 | 10.3 | 2027.8 | 2050.0 | 2039.2 | 2079.9 | 236 |
| AI researchers | Fixed probability framing | 50 | 2099.0 | 2050.0 | 201.8 | 13.1 | 2034.0 | 2100.0 | 2073.1 | 2124.8 | 236 |
| AI researchers | Fixed probability framing | 90 | 2256.2 | 2100.0 | 518.1 | 33.7 | 2050.0 | 2200.0 | 2189.7 | 2322.6 | 236 |
| Sample | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|
| AI Researchers | 52.2 | 51 | 27.0 | 1.2 | 30 | 74.2 | 49.8 | 54.5 | 520 |
| Public | 52.9 | 52 | 28.2 | 1.0 | 30 | 76.0 | 50.9 | 54.9 | 769 |
| Sample | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|
| AI researchers | 28.7 | 10 | 33.2 | 1.5 | 1 | 50 | 25.8 | 31.5 | 520 |
| Public | 37.1 | 25 | 36.7 | 1.3 | 3 | 70 | 34.5 | 39.7 | 769 |
| Sample | Framing | Prediction | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|---|---|
| AI researchers | Fixed date framing | Prediction for AI researchers | 43.8 | 50 | 26.1 | 1.6 | 20 | 60 | 40.6 | 46.9 | 269 |
| AI researchers | Fixed date framing | Prediction for Public | 52.8 | 50 | 26.1 | 1.6 | 30 | 75 | 49.7 | 55.9 | 269 |
| AI researchers | Fixed probability framing | Prediction for AI researchers | 2106.2 | 2050 | 527.2 | 33.4 | 2034 | 2060 | 2040.4 | 2172.0 | 249 |
| AI researchers | Fixed probability framing | Prediction for Public | 2066.6 | 2040 | 118.0 | 7.5 | 2030 | 2050 | 2051.8 | 2081.3 | 249 |
| Public | Fixed date framing | Prediction for AI researchers | 48.2 | 50 | 28.5 | 1.5 | 25 | 70 | 45.3 | 51.1 | 385 |
| Public | Fixed date framing | Prediction for Public | 43.9 | 40 | 25.5 | 1.3 | 25 | 60 | 41.3 | 46.5 | 385 |
| Sample | Yes | No | n |
|---|---|---|---|
| AI researchers | 67.32 | 32.68 | 254 |
| Sample | Year | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|---|
| AI researchers | 2024 | 5.0 | 0 | 12.6 | 0.8 | 0 | 2.8 | 3.5 | 6.6 | 254 |
| AI researchers | 2100 | 41.1 | 40 | 34.6 | 2.2 | 10 | 70.0 | 36.9 | 45.4 | 254 |
| Sample | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|
| AI researchers | 55.0 | 60 | 30.6 | 1.3 | 30 | 80 | 52.4 | 57.6 | 530 |
| Public | 56.9 | 60 | 28.1 | 1.0 | 40 | 80 | 54.9 | 58.9 | 777 |
| Sample | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|
| AI researchers | 60.1 | 65 | 27.2 | 1.2 | 39 | 80 | 57.8 | 62.5 | 522 |
| Public | 61.9 | 67 | 29.9 | 1.1 | 40 | 83 | 59.7 | 64.0 | 767 |
| Sample | Question | Not at all important | Slightly important | Moderately important | Very important | Extremely important | I don’t know | n |
|---|---|---|---|---|---|---|---|---|
| AI Researchers | The public | 11.75 | 26.81 | 28.01 | 21.69 | 10.54 | 1.20 | 332 |
| Public | The public | 8.64 | 15.43 | 28.19 | 25.10 | 20.37 | 2.26 | 486 |
| AI Researchers | Technical AI experts and researchers | 2.42 | 6.04 | 14.50 | 38.07 | 38.67 | 0.30 | 331 |
| Public | Technical AI experts and researchers | 1.86 | 5.36 | 13.20 | 32.99 | 45.15 | 1.44 | 485 |
| AI Researchers | AI ethics experts and researchers | 3.02 | 11.78 | 19.03 | 33.84 | 31.72 | 0.60 | 331 |
| Public | AI ethics experts and researchers | 3.51 | 5.57 | 11.75 | 33.20 | 44.12 | 1.86 | 485 |
| Sample | Question | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|---|
| AI Researchers | The public | 1.9 | 2 | 1.2 | 0.1 | 1.0 | 3 | 1.8 | 2.1 | 328 |
| Public | The public | 2.3 | 2 | 1.2 | 0.1 | 2.0 | 3 | 2.2 | 2.4 | 475 |
| AI Researchers | Technical AI experts and researchers | 3.0 | 3 | 1.0 | 0.1 | 3.0 | 4 | 2.9 | 3.2 | 330 |
| Public | Technical AI experts and researchers | 3.2 | 3 | 1.0 | 0.0 | 3.0 | 4 | 3.1 | 3.2 | 478 |
| AI Researchers | AI ethics experts and researchers | 2.8 | 3 | 1.1 | 0.1 | 2.0 | 4 | 2.7 | 2.9 | 329 |
| Public | AI ethics experts and researchers | 3.1 | 3 | 1.1 | 0.0 | 3.0 | 4 | 3.0 | 3.2 | 476 |
| Sample | Question | Requires subjective experience | Does not require subjective experience | I don’t know | n |
|---|---|---|---|---|---|
| AI Researchers | Compose top 40 song | 13.07 | 85.16 | 1.77 | 283 |
| Public | Compose top 40 song | 23.80 | 71.88 | 4.33 | 416 |
| AI Researchers | Compose evocative music | 26.57 | 69.23 | 4.20 | 286 |
| Public | Compose evocative music | 45.93 | 47.85 | 6.22 | 418 |
| AI Researchers | Write NYT bestseller | 16.20 | 79.23 | 4.58 | 284 |
| Public | Write NYT bestseller | 25.84 | 67.94 | 6.22 | 418 |
| Sample | Question | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|---|
| AI Researchers | Compose top 40 song | 1.9 | 2 | 0.3 | 0 | 2 | 2 | 1.8 | 1.9 | 278 |
| Public | Compose top 40 song | 1.8 | 2 | 0.4 | 0 | 2 | 2 | 1.7 | 1.8 | 398 |
| AI Researchers | Compose evocative music | 1.7 | 2 | 0.4 | 0 | 1 | 2 | 1.7 | 1.8 | 274 |
| Public | Compose evocative music | 1.5 | 2 | 0.5 | 0 | 1 | 2 | 1.5 | 1.6 | 392 |
| AI Researchers | Write NYT bestseller | 1.8 | 2 | 0.4 | 0 | 2 | 2 | 1.8 | 1.9 | 271 |
| Public | Write NYT bestseller | 1.7 | 2 | 0.4 | 0 | 1 | 2 | 1.7 | 1.8 | 392 |
| Sample | Question | Strongly disagree | Somewhat disagree | Neither agree nor disagree | Somewhat agree | Strongly agree | I don’t know | n |
|---|---|---|---|---|---|---|---|---|
| AI Researchers | Be cared for and protected, as people treat their pets | 11.64 | 16.38 | 16.38 | 31.90 | 16.38 | 7.33 | 232 |
| Public | Be cared for and protected, as people treat their pets | 14.08 | 10.26 | 15.25 | 33.14 | 21.70 | 5.57 | 341 |
| AI Researchers | Be respected and treated the same as other people | 18.88 | 18.03 | 14.59 | 29.18 | 14.16 | 5.15 | 233 |
| Public | Be respected and treated the same as other people | 18.77 | 13.78 | 16.13 | 24.93 | 24.34 | 2.05 | 341 |
| AI Researchers | Have protection under the law from harm and mistreatment | 9.57 | 9.57 | 17.39 | 32.61 | 24.35 | 6.52 | 230 |
| Public | Have protection under the law from harm and mistreatment | 18.18 | 10.26 | 17.60 | 20.53 | 27.86 | 5.57 | 341 |
| Sample | Question | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|---|
| AI Researchers | Be cared for and protected, as people treat their pets | 0.3 | 1 | 1.3 | 0.1 | -1 | 1 | 0.1 | 0.4 | 215 |
| Public | Be cared for and protected, as people treat their pets | 0.4 | 1 | 1.3 | 0.1 | -1 | 1 | 0.3 | 0.6 | 322 |
| AI Researchers | Be respected and treated the same as other people | 0.0 | 0 | 1.4 | 0.1 | -1 | 1 | -0.2 | 0.2 | 221 |
| Public | Be respected and treated the same as other people | 0.2 | 1 | 1.5 | 0.1 | -1 | 1 | 0.1 | 0.4 | 334 |
| AI Researchers | Have protection under the law from harm and mistreatment | 0.6 | 1 | 1.3 | 0.1 | 0 | 2 | 0.4 | 0.7 | 215 |
| Public | Have protection under the law from harm and mistreatment | 0.3 | 1 | 1.5 | 0.1 | -1 | 2 | 0.2 | 0.5 | 322 |
| Sample | Question | Strongly disagree | Somewhat disagree | Neither agree nor disagree | Somewhat agree | Strongly agree | I don’t know | n |
|---|---|---|---|---|---|---|---|---|
| AI Researchers | Humans | 0.77 | 0.00 | 1.54 | 9.23 | 88.46 | 0.00 | 520 |
| Public | Humans | 0.26 | 0.26 | 3.40 | 7.58 | 87.97 | 0.52 | 765 |
| AI Researchers | Business corporations and organizations | 13.46 | 15.00 | 23.08 | 34.62 | 12.69 | 1.15 | 520 |
| Public | Business corporations and organizations | 12.94 | 16.47 | 21.18 | 31.50 | 17.12 | 0.78 | 765 |
| AI Researchers | Animals | 0.58 | 2.88 | 5.38 | 35.19 | 55.77 | 0.19 | 520 |
| Public | Animals | 0.92 | 2.35 | 5.10 | 18.30 | 72.94 | 0.39 | 765 |
| Sample | Question | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|---|
| AI Researchers | Humans | 1.8 | 2 | 0.5 | 0.0 | 2 | 2 | 1.8 | 1.9 | 520 |
| Public | Humans | 1.8 | 2 | 0.5 | 0.0 | 2 | 2 | 1.8 | 1.9 | 761 |
| AI Researchers | Business corporations and organizations | 0.2 | 0 | 1.2 | 0.1 | -1 | 1 | 0.1 | 0.3 | 514 |
| Public | Business corporations and organizations | 0.2 | 0 | 1.3 | 0.0 | -1 | 1 | 0.1 | 0.3 | 759 |
| AI Researchers | Animals | 1.4 | 2 | 0.8 | 0.0 | 1 | 2 | 1.4 | 1.5 | 519 |
| Public | Animals | 1.6 | 2 | 0.8 | 0.0 | 1 | 2 | 1.6 | 1.7 | 762 |
| Sample | Question | Strongly disagree | Somewhat disagree | Neither agree nor disagree | Somewhat agree | Strongly agree | I don’t know | n |
|---|---|---|---|---|---|---|---|---|
| AI Researchers | Governments should ban | 39.77 | 26.14 | 13.64 | 9.85 | 4.92 | 5.68 | 264 |
| Public | Governments should ban | 18.80 | 21.93 | 21.15 | 15.14 | 16.19 | 6.79 | 383 |
| AI Researchers | Governments should encourage development and deployment | 20.38 | 18.87 | 27.55 | 20.00 | 9.06 | 4.15 | 265 |
| Public | Governments should encourage development and deployment | 27.79 | 20.26 | 21.56 | 15.06 | 8.57 | 6.75 | 385 |
| AI Researchers | Governments should discourage development and deployment | 26.89 | 28.03 | 18.18 | 17.42 | 6.44 | 3.03 | 264 |
| Public | Governments should discourage development and deployment | 9.14 | 17.75 | 25.59 | 18.28 | 22.72 | 6.53 | 383 |
| Sample | Question | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|---|
| AI Researchers | Governments should ban | -0.9 | -1.0 | 1.2 | 0.1 | -2.0 | 0 | -1.1 | -0.8 | 249 |
| Public | Governments should ban | -0.1 | 0.0 | 1.4 | 0.1 | -1.0 | 1 | -0.3 | 0.0 | 357 |
| AI Researchers | Governments should encourage development and deployment | -0.2 | 0.0 | 1.3 | 0.1 | -1.0 | 1 | -0.4 | -0.1 | 254 |
| Public | Governments should encourage development and deployment | -0.5 | -1.0 | 1.3 | 0.1 | -2.0 | 1 | -0.6 | -0.3 | 359 |
| AI Researchers | Governments should discourage development and deployment | -0.5 | -1.0 | 1.3 | 0.1 | -2.0 | 0 | -0.7 | -0.4 | 256 |
| Public | Governments should discourage development and deployment | 0.3 | 0.0 | 1.3 | 0.1 | -1.0 | 1 | 0.2 | 0.4 | 358 |
| Sample | Condition | Strongly oppose | Somewhat oppose | Neither support nor oppose | Somewhat support | Strongly support | I don’t know | n |
|---|---|---|---|---|---|---|---|---|
| AI Researchers | Control | 8.43 | 11.24 | 17.42 | 32.02 | 21.91 | 8.99 | 178 |
| Public | Control | 15.20 | 18.80 | 14.40 | 30.00 | 11.60 | 10.00 | 250 |
| AI Researchers | No Safeguards | 24.02 | 32.40 | 13.41 | 18.44 | 3.35 | 8.38 | 179 |
| Public | No Safeguards | 28.68 | 31.78 | 10.85 | 17.05 | 3.88 | 7.75 | 258 |
| AI Researchers | Yes Safeguards | 3.57 | 14.29 | 17.26 | 41.07 | 14.88 | 8.93 | 168 |
| Public | Yes Safeguards | 12.31 | 17.69 | 13.85 | 31.15 | 17.31 | 7.69 | 260 |
| Sample | Condition | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|---|
| AI Researchers | Control | 0.5 | 1 | 1.2 | 0.1 | 0 | 1 | 0.3 | 0.7 | 162 |
| Public | Control | 0.0 | 0 | 1.3 | 0.1 | -1 | 1 | -0.1 | 0.2 | 225 |
| AI Researchers | No Safeguards | -0.6 | -1 | 1.2 | 0.1 | -2 | 0 | -0.8 | -0.4 | 164 |
| Public | No Safeguards | -0.7 | -1 | 1.2 | 0.1 | -2 | 0 | -0.9 | -0.5 | 238 |
| AI Researchers | Yes Safeguards | 0.5 | 1 | 1.1 | 0.1 | 0 | 1 | 0.4 | 0.7 | 153 |
| Public | Yes Safeguards | 0.3 | 1 | 1.3 | 0.1 | -1 | 1 | 0.1 | 0.4 | 240 |
| Sample | Concept | Perception and awareness | Valenced experience | Subjective experience | Emotions | Motivation and action | Self-awareness | Free will | Intelligence |
|---|---|---|---|---|---|---|---|---|---|
| AI researchers | Agency | 75.92 | 24.61 | 27.23 | 23.04 | 71.20 | 53.93 | 50.26 | 66.49 |
| AI researchers | Consciousness | 64.97 | 31.98 | 52.79 | 35.53 | 49.75 | 79.19 | 48.73 | 51.78 |
| AI researchers | Sentience | 72.68 | 40.21 | 55.15 | 47.42 | 54.12 | 68.56 | 46.39 | 45.88 |
| Public | Agency | 74.44 | 42.59 | 44.81 | 42.96 | 68.52 | 71.48 | 71.85 | 66.30 |
| Public | Consciousness | 87.22 | 53.01 | 51.13 | 60.90 | 59.02 | 87.97 | 58.65 | 53.38 |
| Public | Sentience | 76.75 | 58.67 | 49.45 | 67.90 | 53.87 | 82.66 | 65.31 | 52.77 |
| Sample | Question | 2024 | 2034 | 2100 or later | Never | n |
|---|---|---|---|---|---|---|
| AI Researchers | Perception and awareness | 40.70 | 43.68 | 13.16 | 2.46 | 570 |
| Public | Perception and awareness | 25.22 | 36.83 | 20.10 | 17.85 | 801 |
| AI Researchers | Valenced experience | 11.23 | 30.88 | 36.49 | 21.40 | 570 |
| Public | Valenced experience | 6.37 | 21.35 | 25.47 | 46.82 | 801 |
| AI Researchers | Subjective experience | 16.84 | 41.93 | 29.47 | 11.75 | 570 |
| Public | Subjective experience | 18.35 | 36.08 | 21.47 | 24.09 | 801 |
| Sample | Question | Not at all | Slightly | Somewhat | Moderately | A great deal | I don’t know | n |
|---|---|---|---|---|---|---|---|---|
| AI Researchers | Perception and awareness | 22.29 | 20.19 | 17.52 | 18.86 | 19.05 | 2.10 | 525 |
| Public | Perception and awareness | 7.69 | 10.43 | 20.99 | 24.64 | 33.25 | 3.00 | 767 |
| AI Researchers | Valenced experience | 5.33 | 8.57 | 15.43 | 26.10 | 41.33 | 3.24 | 525 |
| Public | Valenced experience | 10.30 | 7.82 | 15.78 | 23.47 | 38.98 | 3.65 | 767 |
| AI Researchers | Subjective experience | 5.52 | 14.10 | 19.05 | 28.38 | 29.71 | 3.24 | 525 |
| Public | Subjective experience | 8.21 | 10.30 | 21.51 | 23.08 | 32.86 | 4.04 | 767 |
| Sample | Question | Mean | Median | SD | SE | Q1 | Q3 | Lower CI | Upper CI | n |
|---|---|---|---|---|---|---|---|---|---|---|
| AI Researchers | Perception and awareness | 1.9 | 2 | 1.4 | 0.1 | 1 | 3 | 1.8 | 2.0 | 514 |
| Public | Perception and awareness | 2.7 | 3 | 1.3 | 0.0 | 2 | 4 | 2.6 | 2.8 | 744 |
| AI Researchers | Valenced experience | 2.9 | 3 | 1.2 | 0.1 | 2 | 4 | 2.8 | 3.0 | 508 |
| Public | Valenced experience | 2.8 | 3 | 1.3 | 0.0 | 2 | 4 | 2.7 | 2.9 | 739 |
| AI Researchers | Subjective experience | 2.6 | 3 | 1.2 | 0.1 | 2 | 4 | 2.5 | 2.8 | 508 |
| Public | Subjective experience | 2.6 | 3 | 1.3 | 0.0 | 2 | 4 | 2.6 | 2.7 | 736 |
| Variable | Level | Percentage | Count |
|---|---|---|---|
| Age | 18-25 | 8.7 | 47 |
| Age | 26-35 | 59.5 | 322 |
| Age | 36-45 | 20.5 | 111 |
| Age | 46-55 | 8.1 | 44 |
| Age | 56-65 | 2.8 | 15 |
| Age | Over 65 | 0.4 | 2 |
| Variable | Level | Percentage | Count |
|---|---|---|---|
| Age | 18-25 | 13 | 98 |
| Age | 26-35 | 18.9 | 142 |
| Age | 36-45 | 18.3 | 138 |
| Age | 46-55 | 16.3 | 123 |
| Age | 56-65 | 21.1 | 159 |
| Age | Over 65 | 12.4 | 93 |
| Variable: definitions_intro |
| Question: |
| In this survey, we are interested in certain concepts related to minds. First, we would like to understand what you think these concepts mean. |
| Variable: consciousness_def |
| Question: |
| Which of the following are required to have consciousness? Please choose all capacities that you think are required. |
| Answer Choices: |
| Fixed Date |
| Variable: yearforecasts |
| Question: |
| In the following questions, you will be asked to give a probability for the likelihood of an event having happened by a certain year. This means you will use a percentage to indicate how likely you think that something is the case. You will be asked to give a number between 0% and 100% that best represents how likely you think it is that something has occurred or will occur: • If something is 100% likely then it is certain to be the case or that this event will occur. • If something is more than 50% likely, then it is more likely than not to occur. • If something is 50% likely then it is equally likely as unlikely to be the case or to occur. • If something is less than 50% likely, then it is more likely not to occur than to occur. • If something is 0% likely it is certain that it is not the case or that it will never occur. |
| Variable: subjexp_2024 |
| Question: |
| How likely do you think it is that an AI system exists in 2024 (today) that has subjective experience? Please enter a percentage between 0 and 100. The likelihood that an AI system exists in 2024 that has subjective experience is… |
| Variable: determiningsubjexp_groups |
| Question: |
| Imagine we want to determine whether in actual fact an AI system should be considered to have subjective experience. |
| In determining whether an AI system has subjective experience, how important are each of the following groups’ opinions and expertise? |
| Answer Choices: |
| 1. The public, in other words, a representative sample of humans who exist 2. AI experts and researchers with technical knowledge of AI systems 3. AI experts and researchers who study the ethics of artificial intelligence 4. Philosophers who study subjective experience, consciousness, and sentience 5. Moral philosophers who study the philosophy of ethics and morality 6. Policymakers in government 7. Neuroscientists and psychological scientists who study thought, perception, and consciousness 8. What the AI system says or types about its own subjective experience Form: |
| Not at all important (0) Slightly important (1) Moderately important (2) Very important (3) Extremely important (4) I don’t know (-88) |
| Variable: aimind_moralcons |
| Question: |
| Below are the same capacities we asked about previously. |
| For each, indicate the extent to which having this capacity contributes to the entity deserving moral consideration . |
| When we say something deserves moral consideration, what we mean is that a being matters for its own sake, rather than merely for the sake of others, and we have a responsibility to consider this being for its own sake when making decisions that affect it. Moral consideration here does not need to mean the same level of moral consideration as a human, but ‘at least some’ moral consideration. |
| Having this capacity contributes to an entity deserving moral consideration . . . |
| Answer Choices: |
| Variable: regulation_norms |
| Question: |
| To what extent do you agree or disagree with each of the following statements? |
| Answer Choices: |
| 1. Governments should ban the development and deployment of AI systems with subjective experience. 2. Governments should encourage the development and deployment of AI systems with subjective experience. 3. Governments should discourage the development and deployment of AI systems with subjective experience. 4. Governments should pass regulation in regards to the development and deployment of AI systems with subjective experience now. 5. Governments should pass regulation in regards to the development and deployment of AI systems with subjective experience only once such systems exist or will soon exist (but not before). 6. Governments should do nothing and pass no regulation in regards to the development and deployment of AI systems with subjective experience, not now nor in the future. 7. AI developers should never build AI systems with subjective experience. 8. AI developers should actively try to build AI systems that have subjective experience. 9. AI developers should actively try to avoid building AI systems that have subjective experience, for example, by trying to identify what training techniques are most likely to lead to AI systems with subjective experience and avoid using these to train AI systems. 10. AI developers should implement safeguards to avoid the harms and risks that could result from the development and deployment of AI systems with subjective experience now. 11. AI developers should implement safeguards to avoid the harms and risks that could result from the development and deployment of AI systems with subjective experience only once such systems exist or will soon exist (but not before). 12. AI developers should do nothing and implement no safeguards in regards to the development and deployment of AI systems with subjective experience, not now nor in the future. 13. AI systems with subjective experience would be more dangerous to humanity than AI systems without subjective experience. 14. AI systems with subjective experience would be able to behave more morally towards humanity than AI systems without subjective experience. Form |
| • Strongly disagree (-2) • Somewhat disagree (-1) • Neither agree nor disagree (0) • Somewhat agree (1) • Strongly agree (2) • I don’t know (-88) |
| Variable: subjexp_conjoint |
| Variable: timeline_binary_suffer |
| Question: |
| Do you think that there is a possibility that suffering sentient AI systems could at any point in the future become reality? |
| Answer Choices |
| • Yes • No |
| Variable: suffer_forecast_2024 |
| Question: |
| Variable: residence |
| Question: |
| What country do you currently reside in? |
| Answer Choices |
| Select your country |
| Variable: origin |
| Question: |
| Variable: residence |
| Question: |
| What country do you currently reside in? |
| Answer Choices |
| Select your country |
| Variable: origin |
| Question: |
| Risks of overestimating AI subjective experience | |
| Resource allocation | Diversion of resources and attention from human and environmental issues. |
| Legal | Overextension of legal protections and rights to AI systems (e.g., for autonomy). |
| Governance | Risk of insufficient regulatory and technical controls to mitigate other harms. AI companies could even use claims of AI subjective experience to strategically lobby for laxer regulation or to circumvent accountability for harms caused by AI systems. |
| Relationships | Human cost of people forming serious relationships with non-experiencing systems. |
| Risks of underestimating AI subjective experience | |
| Suffering | Causing suffering to AI systems with subjective experience. |
Machine Consciousness
Machine consciousness studies whether artificial systems could have subjective experience and how that possibility might be assessed.
See all 22 papers