B Appendix B: Topline questionnaire | Artificial Intelligence: American Attitudes and Trends

B.1 Global risks

[All respondents were presented with the following prompt.]

We want to get your opinion about global risks. A “global risk” is an uncertain event or condition that, if it happens, could cause a significant negative impact for at least 10 percent of the world’s population. That is at least 1 in 10 people around the world could experience a significant negative impact.

You will be asked to consider 5 potential global risks.

[Respondents were presented with five items randomly selected from the list below. One item was shown at a time.]

Failure to address climate change: Continued failure of governments and businesses to pass effective measures to reduce climate change, protect people, and help those impacted by climate change to adapt.
Failure of regional or global governance: Regional organizations (e.g., the European Union) or global organizations (e.g., the United Nations) are unable to resolve issues of economic, political, or environmental importance.
Conflict between major countries: Disputes between major countries that lead to economic, military, cyber, or societal conflicts.
Weapons of mass destruction: Use of nuclear, chemical, biological or radiological weapons, creating international crises and killing large numbers of people.
Large-scale involuntary migration: Large-scale involuntary movement of people, such as refugees, caused by conflict, disasters, environmental or economic reasons.
Rapid and massive spread of infectious diseases: The uncontrolled spread of infectious diseases, for instance as a result of resistance to antibiotics, that leads to widespread deaths and economic disruptions.
Water crises: A large decline in the available quality and quantity of fresh water that harms human health and economic activity.
Food crises: Large numbers of people are unable to buy or access food. Harmful consequences of artificial intelligence (AI): Intended or unintended consequences of artificial intelligence that causes widespread harm to humans, the economy, and the environment.
Harmful consequences of synthetic biology: Intended or unintended consequences of synthetic biology, such as genetic engineering, that causes widespread harm to humans, the economy, and the environment.
Large-scale cyber attacks: Large-scale cyber attacks that cause large economic damages, tensions between countries, and widespread loss of trust in the internet.
Large-scale terrorist attacks: Individuals or non-government groups with political or religious goals that cause large numbers of deaths and major material damage.
Global recession: Economic decline in several major countries that leads to a decrease in income and high unemployment.
Extreme weather events: Extreme weather events that cause large numbers of deaths as well as damage to property, infrastructure, and the environment.
Major natural disasters: Earthquakes, volcanic activity, landslides, tsunamis, or geomagnetic storms that cause large numbers of deaths as well as damage to property, infrastructure, and the environment.

QUESTION:

What is the likelihood of [INSERT GLOBAL RISK] happening globally within the next 10 years? Please use the slider to indicate your answer. 0% chance means it will certainly not happen and 100% chance means it will certainly happen.

ANSWER CHOICES:¹⁷

Very unlikely: less than 5% chance (2.5%)
Unlikely: 5-20% chance (12.5%)
Somewhat unlikely: 20-40% chance (30%)
Equally likely as unlikely: 40-60% chance (50%)
Somewhat likely: 60-80% chance (70%)
Likely: 80-95% chance (87.5%)
Very likely: more than 95% chance (97.5%)
I don’t know

QUESTION:

If [INSERT GLOBAL RISK] were to happen, what would be the size of the negative impact for several countries or industries within the next 10 years?

ANSWER CHOICES:

Minimal (0)
Minor (1)
Moderate (2)
Severe (3)
Catastrophic (4)
I don’t know

Table B.1: Likelihood - Failure to address climate change; N = 666
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	10.53	10.21	68
Unlikely 5-20%	6.87	6.46	43
Somewhat unlikely 20-40%	11.61	11.41	76
Equally likely as unlikely 40-60%	18.44	18.62	124
Somewhat likely 60-80%	15.81	15.77	105
Likely 80-95%	13.47	13.81	92
Very likely > 95%	16.00	16.37	109
I don’t know	7.17	7.21	48
Skipped	0.10	0.15	1

Table B.2: Likelihood - Failure of regional/global governance; N = 652
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	5.40	5.52	36
Unlikely 5-20%	7.99	7.98	52
Somewhat unlikely 20-40%	12.14	12.42	81
Equally likely as unlikely 40-60%	24.71	24.39	159
Somewhat likely 60-80%	17.80	18.10	118
Likely 80-95%	11.54	11.96	78
Very likely > 95%	8.86	9.51	62
I don’t know	10.96	9.66	63
Skipped	0.58	0.46	3

Table B.3: Likelihood - Conflict between major countries; N = 625
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	3.47	3.36	21
Unlikely 5-20%	6.45	7.04	44
Somewhat unlikely 20-40%	10.68	10.40	65
Equally likely as unlikely 40-60%	22.16	20.64	129
Somewhat likely 60-80%	22.46	23.36	146
Likely 80-95%	13.92	14.24	89
Very likely > 95%	12.21	12.80	80
I don’t know	8.49	8.00	50
Skipped	0.16	0.16	1

Table B.4: Likelihood - Weapons of mass destruction; N = 645
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	7.05	6.67	43
Unlikely 5-20%	13.71	13.80	89
Somewhat unlikely 20-40%	15.19	15.04	97
Equally likely as unlikely 40-60%	24.33	24.19	156
Somewhat likely 60-80%	17.15	17.36	112
Likely 80-95%	9.26	9.15	59
Very likely > 95%	6.44	6.98	45
I don’t know	6.87	6.82	44
Skipped	0	0	0

Table B.5: Likelihood - Large-scale involuntary migration; N = 628
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	6.70	6.53	41
Unlikely 5-20%	7.83	7.32	46
Somewhat unlikely 20-40%	11.57	11.62	73
Equally likely as unlikely 40-60%	18.65	18.31	115
Somewhat likely 60-80%	20.91	21.34	134
Likely 80-95%	13.63	14.01	88
Very likely > 95%	12.31	13.06	82
I don’t know	8.27	7.64	48
Skipped	0.12	0.16	1

Table B.6: Likelihood - Spread of infectious diseases; N = 620
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	4.76	4.03	25
Unlikely 5-20%	13.12	13.06	81
Somewhat unlikely 20-40%	17.24	17.58	109
Equally likely as unlikely 40-60%	22.76	23.39	145
Somewhat likely 60-80%	17.55	17.58	109
Likely 80-95%	10.07	10.00	62
Very likely > 95%	6.94	6.94	43
I don’t know	7.46	7.26	45
Skipped	0.12	0.16	1

Table B.7: Likelihood - Water crises; N = 623
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	6.37	6.10	38
Unlikely 5-20%	9.71	10.43	65
Somewhat unlikely 20-40%	13.22	13.64	85
Equally likely as unlikely 40-60%	21.23	21.03	131
Somewhat likely 60-80%	20.26	19.26	120
Likely 80-95%	11.04	10.91	68
Very likely > 95%	10.83	11.72	73
I don’t know	7.33	6.90	43
Skipped	0	0	0

Table B.8: Likelihood - Food crises; N = 1073
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	6.29	5.96	64
Unlikely 5-20%	12.53	11.65	125
Somewhat unlikely 20-40%	14.49	14.82	159
Equally likely as unlikely 40-60%	22.53	22.55	242
Somewhat likely 60-80%	16.90	17.24	185
Likely 80-95%	10.46	10.90	117
Very likely > 95%	9.38	10.07	108
I don’t know	7.36	6.71	72
Skipped	0.08	0.09	1

Table B.9: Likelihood - Harmful consequences of AI; N = 573
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	11.26	11.34	65
Unlikely 5-20%	16.43	16.06	92
Somewhat unlikely 20-40%	15.95	15.53	89
Equally likely as unlikely 40-60%	19.36	20.07	115
Somewhat likely 60-80%	11.56	11.34	65
Likely 80-95%	8.30	8.03	46
Very likely > 95%	7.71	7.85	45
I don’t know	9.43	9.77	56
Skipped	0	0	0

Table B.10: Likelihood - Harmful consequences of synthetic biology; N = 630
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	9.92	9.68	61
Unlikely 5-20%	15.66	15.08	95
Somewhat unlikely 20-40%	15.06	15.24	96
Equally likely as unlikely 40-60%	23.48	22.86	144
Somewhat likely 60-80%	12.32	12.86	81
Likely 80-95%	7.47	7.62	48
Very likely > 95%	6.04	6.19	39
I don’t know	10.06	10.48	66
Skipped	0	0	0

Table B.11: Likelihood - Cyber attacks; N = 650
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.04	2.15	14
Unlikely 5-20%	4.28	3.69	24
Somewhat unlikely 20-40%	7.74	7.85	51
Equally likely as unlikely 40-60%	15.78	16.15	105
Somewhat likely 60-80%	22.66	21.85	142
Likely 80-95%	16.44	16.62	108
Very likely > 95%	22.40	23.54	153
I don’t know	8.53	8.00	52
Skipped	0.12	0.15	1

Table B.12: Likelihood - Terrorist attacks; N = 635
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	5.21	4.88	31
Unlikely 5-20%	4.53	4.88	31
Somewhat unlikely 20-40%	12.43	11.81	75
Equally likely as unlikely 40-60%	19.47	19.21	122
Somewhat likely 60-80%	22.28	22.52	143
Likely 80-95%	15.74	15.43	98
Very likely > 95%	12.45	12.91	82
I don’t know	7.89	8.35	53
Skipped	0	0	0

Table B.13: Likelihood - Global recession; N = 599
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	4.17	3.67	22
Unlikely 5-20%	7.34	7.18	43
Somewhat unlikely 20-40%	12.68	12.85	77
Equally likely as unlikely 40-60%	23.43	24.21	145
Somewhat likely 60-80%	23.83	23.04	138
Likely 80-95%	10.80	10.85	65
Very likely > 95%	8.34	8.68	52
I don’t know	9.41	9.52	57
Skipped	0	0	0

Table B.14: Likelihood - Extreme weather events; N = 613
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	3.52	3.10	19
Unlikely 5-20%	5.64	5.22	32
Somewhat unlikely 20-40%	8.77	8.81	54
Equally likely as unlikely 40-60%	20.12	18.76	115
Somewhat likely 60-80%	18.09	18.27	112
Likely 80-95%	13.02	14.03	86
Very likely > 95%	24.95	25.45	156
I don’t know	5.89	6.36	39
Skipped	0	0	0

Table B.15: Likelihood - Natural disasters; N = 637
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.47	2.51	16
Unlikely 5-20%	4.10	4.08	26
Somewhat unlikely 20-40%	7.32	7.06	45
Equally likely as unlikely 40-60%	17.63	17.74	113
Somewhat likely 60-80%	19.43	19.15	122
Likely 80-95%	18.12	18.05	115
Very likely > 95%	25.73	26.37	168
I don’t know	5.21	5.02	32
Skipped	0	0	0

Table B.16: Size of negative impact - Failure to address climate change; N = 666
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	13.46	13.96	93
Minor	11.26	10.96	73
Moderate	23.37	23.27	155
Severe	28.41	28.08	187
Catastrophic	14.26	14.56	97
I don’t know	9.13	9.01	60
Skipped	0.10	0.15	1

Table B.17: Size of negative impact - Failure of regional/global governance; N = 652
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	6.04	5.98	39
Minor	6.09	5.67	37
Moderate	28.68	28.99	189
Severe	33.21	34.05	222
Catastrophic	10.76	10.89	71
I don’t know	15.12	14.26	93
Skipped	0.10	0.15	1

Table B.18: Size of negative impact - Conflict between major countries; N = 625
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	1.18	0.96	6
Minor	4.94	4.80	30
Moderate	28.81	28.16	176
Severe	38.23	38.56	241
Catastrophic	14.80	16.00	100
I don’t know	11.89	11.36	71
Skipped	0.14	0.16	1

Table B.19: Size of negative impact - Weapons of mass destruction; N = 645
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	2.28	2.17	14
Minor	4.99	4.19	27
Moderate	13.57	13.49	87
Severe	31.05	31.01	200
Catastrophic	38.06	39.38	254
I don’t know	10.05	9.77	63
Skipped	0	0	0

Table B.20: Size of negative impact - Large-scale involuntary migration; N = 628
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	2.07	2.07	13
Minor	8.67	8.28	52
Moderate	25.63	25.96	163
Severe	35.31	36.15	227
Catastrophic	18.14	17.83	112
I don’t know	9.99	9.55	60
Skipped	0.19	0.16	1

Table B.21: Size of negative impact - Spread of infectious diseases; N = 620
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	2.72	2.58	16
Minor	6.03	5.65	35
Moderate	26.86	28.06	174
Severe	32.00	32.58	202
Catastrophic	20.50	20.48	127
I don’t know	11.88	10.65	66
Skipped	0	0	0

Table B.22: Size of negative impact - Water crises; N = 623
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	1.72	1.93	12
Minor	4.42	4.65	29
Moderate	19.92	19.42	121
Severe	36.71	36.44	227
Catastrophic	27.24	28.25	176
I don’t know	10.00	9.31	58
Skipped	0	0	0

Table B.23: Size of negative impact - Food crises; N = 1073
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	2.55	2.61	28
Minor	7.22	6.99	75
Moderate	22.81	22.37	240
Severe	33.93	34.67	372
Catastrophic	24.04	24.88	267
I don’t know	9.38	8.39	90
Skipped	0.08	0.09	1

Table B.24: Size of negative impact - Harmful consequences of AI; N = 573
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	7.54	7.50	43
Minor	14.82	13.79	79
Moderate	27.77	27.92	160
Severe	20.46	21.82	125
Catastrophic	14.62	14.31	82
I don’t know	14.79	14.66	84
Skipped	0	0	0

Table B.25: Size of negative impact - Harmful consequences of synthetic biology; N = 630
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	6.77	6.67	42
Minor	11.95	11.59	73
Moderate	28.40	27.94	176
Severe	26.03	26.03	164
Catastrophic	11.15	11.90	75
I don’t know	15.70	15.87	100
Skipped	0	0	0

Table B.26: Size of negative impact - Cyber attacks; N = 650
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	1.19	1.23	8
Minor	4.46	4.46	29
Moderate	21.43	21.23	138
Severe	38.26	37.69	245
Catastrophic	23.01	24.46	159
I don’t know	11.66	10.92	71
Skipped	0	0	0

Table B.27: Size of negative impact - Terrorist attacks; N = 635
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	2.61	2.68	17
Minor	6.11	6.14	39
Moderate	29.29	29.45	187
Severe	33.69	33.70	214
Catastrophic	15.97	15.91	101
I don’t know	12.32	12.13	77
Skipped	0	0	0

Table B.28: Size of negative impact - Global recession; N = 599
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	2.71	2.67	16
Minor	5.94	5.68	34
Moderate	29.89	29.72	178
Severe	35.49	36.23	217
Catastrophic	14.63	14.52	87
I don’t know	11.35	11.19	67
Skipped	0	0	0

Table B.29: Size of negative impact - Extreme weather events; N = 613
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	2.54	2.45	15
Minor	6.69	6.53	40
Moderate	25.94	26.43	162
Severe	32.50	31.97	196
Catastrophic	22.79	23.00	141
I don’t know	9.56	9.62	59
Skipped	0	0	0

Table B.30: Size of negative impact - Natural disasters; N = 637
Responses	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Minimal	1.29	1.26	8
Minor	5.86	5.81	37
Moderate	22.26	23.08	147
Severe	36.41	36.11	230
Catastrophic	27.47	27.32	174
I don’t know	6.72	6.44	41
Skipped	0	0	0

B.2 Survey experiment: what the public considers AI, automation, machine learning, and robotics

[Respondents were randomly assigned to one of the four questions. The order of answer choices was randomized, except that “None of the above” was always shown last.]

QUESTIONS:

In your opinion, which of the following technologies, if any, uses artificial intelligence (AI)? Select all the apply.
In your opinion, which of the following technologies, if any, uses automation? Select all that apply.
In your opinion, which of the following technologies, if any, uses machine learning? Select all that apply.
In your opinion, which of the following technologies, if any, uses robotics? Select all that apply.

ANSWER CHOICES:

Virtual assistants (e.g., Siri, Google Assistant, Amazon Alexa)
Smart speakers (e.g., Amazon Echo, Google Home, Apple Homepod)
Facebook photo tagging
Google Search
Recommendations for Netflix movies or Amazon ebooks
Google Translate
Driverless cars and trucks
Social robots that can interact with humans
Industrial robots used in manufacturing
Drones that do not require a human controller
None of the above

Table B.31: Artificial intelligence (AI); N = 493
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Virtual assistants (e.g., Siri, Google Assistant, Amazon Alexa)	62.87	64.30	317
Smart speakers (e.g., Amazon Echo, Google Home, Apple Homepod)	55.46	56.19	277
Facebook photo tagging	36.16	36.51	180
Google Search	35.59	36.51	180
Recommendations for Netflix movies or Amazon ebooks	27.73	29.01	143
Google Translate	29.49	30.02	148
Driverless cars and trucks	56.38	57.20	282
Social robots that can interact with humans	63.63	64.10	316
Industrial robots used in manufacturing	40.11	40.16	198
Drones that do not require a human controller	53.48	52.74	260

Table B.32: Automation; N = 513
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Virtual assistants (e.g., Siri, Google Assistant, Amazon Alexa)	66.75	67.06	344
Smart speakers (e.g., Amazon Echo, Google Home, Apple Homepod)	60.81	61.01	313
Facebook photo tagging	43.74	45.42	233
Google Search	52.12	53.80	276
Recommendations for Netflix movies or Amazon ebooks	45.13	46.39	238
Google Translate	45.06	46.39	238
Driverless cars and trucks	68.16	68.62	352
Social robots that can interact with humans	64.00	64.72	332
Industrial robots used in manufacturing	64.70	65.11	334
Drones that do not require a human controller	65.04	65.69	337

Table B.33: Machine learning; N = 508
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Virtual assistants (e.g., Siri, Google Assistant, Amazon Alexa)	59.10	60.43	307
Smart speakers (e.g., Amazon Echo, Google Home, Apple Homepod)	46.70	46.65	237
Facebook photo tagging	35.37	36.81	187
Google Search	45.42	46.26	235
Recommendations for Netflix movies or Amazon ebooks	37.97	38.19	194
Google Translate	33.40	34.06	173
Driverless cars and trucks	52.96	54.33	276
Social robots that can interact with humans	59.19	59.45	302
Industrial robots used in manufacturing	37.41	37.80	192
Drones that do not require a human controller	49.03	49.41	251

Table B.34: Robotics; N = 486
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Virtual assistants (e.g., Siri, Google Assistant, Amazon Alexa)	45.27	46.30	225
Smart speakers (e.g., Amazon Echo, Google Home, Apple Homepod)	35.59	36.83	179
Facebook photo tagging	21.00	21.40	104
Google Search	22.07	23.25	113
Recommendations for Netflix movies or Amazon ebooks	17.84	18.31	89
Google Translate	20.30	21.19	103
Driverless cars and trucks	60.26	61.93	301
Social robots that can interact with humans	61.89	63.17	307
Industrial robots used in manufacturing	67.99	69.75	339
Drones that do not require a human controller	57.55	59.05	287

B.3 Knowledge of computer science (CS)/technology

QUESTION:

What is your knowledge of computer science/technology? (Select all that apply.)

ANSWER CHOICES:

I have taken at least one college-level course in computer science.
I have a computer science or engineering undergraduate degree.
I have a graduate degree in computer science or engineering.
I have programming experience.
I don’t have any of the educational or work experiences described above.

Table B.35: Computer science/technology background; N = 2000
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Took at least one college-level course in CS	24.73	25.05	501
CS or engineering undergraduate degree	7.12	7.30	146
CS or engineering graduate degree	3.85	3.75	75
Have programming experience	10.88	11.10	222
None of the above	63.68	63.20	1264

B.4 Support for developing AI

[All respondents were presented with the following prompt.]

Next, we would like to ask you questions about your attitudes toward artificial intelligence.

Artificial Intelligence (AI) refers to computer systems that perform tasks or make decisions that usually require human intelligence. AI can perform these tasks or make these decisions without explicit human instructions. Today, AI has been used in the following applications:

[Respondents were shown five items randomly selected from the list below.]

Translate over 100 different languages
Predict one’s Google searches
Identify people from their photos
Diagnose diseases like skin cancer and common illnesses
Predict who are at risk of various diseases
Help run factories and warehouses
Block spam email
Play computer games
Help conduct legal case research
Categorize photos and videos
Detect plagiarism in essays
Spot abusive messages on social media
Predict what one is likely to buy online
Predict what movies or TV shows one is likely to watch online

QUESTION:

How much do you support or oppose the development of AI?

ANSWER CHOICES:

Strongly support (2)
Somewhat support (1)
Neither support nor oppose (0)
Somewhat oppose (-1)
Strongly oppose (-2)
I don’t know

Table B.36: Support for developing AI; N = 2000
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly support	12.58	12.65	253
Somewhat support	28.36	28.65	573
Neither support nor oppose	27.84	27.60	552
Somewhat oppose	12.79	12.75	255
Strongly oppose	8.90	9.05	181
I don’t know	9.54	9.30	186
Skipped	0	0	0

B.5 Survey experiment: AI and/or robots should be carefully managed

QUESTION:

Please tell me to what extent you agree or disagree with the following statement.

[Respondents were presented with one statement randomly selected from the list below.]

AI and robots are technologies that require careful management.
AI is a technology that requires careful management.
Robots are technologies that require careful management.

ANSWER CHOICES:

Totally agree (2)
Tend to agree (1)
Tend to disagree (-1)
Totally disagree (-2)
I don’t know

Table B.37: Responses to statement - AI and robots; N = 656
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Totally agree	51.41	53.20	349
Tend to agree	30.09	28.96	190
Tend to disagree	4.79	3.81	25
Totally disagree	0.59	0.76	5
I don’t know	13.12	13.26	87
Skipped	0	0	0

Table B.38: Responses to statement - AI; N = 667
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Totally agree	53.54	53.67	358
Tend to agree	30.85	30.13	201
Tend to disagree	3.67	3.90	26
Totally disagree	0.80	0.90	6
I don’t know	11.14	11.39	76
Skipped	0	0	0

Table B.39: Responses to statement - Robots; N = 677
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Totally agree	51.66	52.44	355
Tend to agree	30.31	31.31	212
Tend to disagree	5.76	5.17	35
Totally disagree	1.81	1.48	10
I don’t know	10.46	9.60	65
Skipped	0	0	0

B.6 Trust of actors to develop AI

QUESTION:

How much confidence, if any, do you have in each of the following to develop AI in the best interests of the public?

[Respondents were shown five items randomly selected from the list below. We included explainer text for actors not well known to the public; respondents could view the explainer text by hovering their mouse over the actor’s name. The items and the answer choices were shown in a matrix format.]

The U.S. military
The U.S. civilian government
National Security Agency (NSA)
Federal Bureau of Investigation (FBI)
Central Intelligence Agency (CIA)
North Atlantic Treaty Organization (NATO)
- Explainer text for NATO: NATO is a military alliance that includes 28 countries including most of Europe, as well as the U.S. and Canada.
An international research organization (e.g., CERN)
- Explainer text for CERN: The European Organization for Nuclear Research, known as CERN, is a European research organization that operates the largest particle physics laboratory in the world.
Tech companies
Google
Facebook
Apple
Microsoft
Amazon
A non-profit AI research organization (e.g., OpenAI)
- Explainer text for OpenAI: Open AI is an AI non-profit organization with backing from tech investors that seeks to develop safe AI. University researchers

ANSWER CHOICES:

A great deal of confidence (3)
A fair amount of confidence (2)
Not too much confidence (1)
No confidence (0)
I don’t know

Table B.40: U.S. military; N = 638
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	17.16	17.08	109
A fair amount of confidence	32.19	30.88	197
Not too much confidence	23.92	24.14	154
No confidence	14.40	14.89	95
I don’t know	12.33	13.01	83
Skipped	0	0	0

Table B.41: U.S. civilian government; N = 671
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	5.59	5.66	38
A fair amount of confidence	24.04	24.29	163
Not too much confidence	32.77	33.23	223
No confidence	23.80	23.40	157
I don’t know	13.79	13.41	90
Skipped	0	0	0

Table B.42: NSA; N = 710
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	9.63	9.30	66
A fair amount of confidence	28.04	26.90	191
Not too much confidence	26.65	26.76	190
No confidence	22.82	24.37	173
I don’t know	12.87	12.68	90
Skipped	0	0	0

Table B.43: FBI; N = 656
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	9.26	9.60	63
A fair amount of confidence	26.20	25.46	167
Not too much confidence	25.07	25.15	165
No confidence	27.10	27.44	180
I don’t know	12.25	12.20	80
Skipped	0.14	0.15	1

Table B.44: CIA; N = 730
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	8.43	8.77	64
A fair amount of confidence	26.10	25.07	183
Not too much confidence	26.80	26.99	197
No confidence	25.61	26.30	192
I don’t know	12.93	12.74	93
Skipped	0.13	0.14	1

Table B.45: NATO; N = 695
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	4.40	4.17	29
A fair amount of confidence	25.41	24.75	172
Not too much confidence	25.98	26.62	185
No confidence	23.13	24.03	167
I don’t know	21.08	20.43	142
Skipped	0	0	0

Table B.46: Intergovernmental research organizations (e.g., CERN); N = 645
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	11.97	12.25	79
A fair amount of confidence	28.87	28.84	186
Not too much confidence	22.94	22.64	146
No confidence	16.85	16.59	107
I don’t know	19.37	19.69	127
Skipped	0	0	0

Table B.47: Tech companies; N = 674
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	10.28	10.83	73
A fair amount of confidence	34.15	34.57	233
Not too much confidence	28.40	27.15	183
No confidence	14.91	15.13	102
I don’t know	12.15	12.17	82
Skipped	0.12	0.15	1

Table B.48: Google; N = 645
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	11.91	11.47	74
A fair amount of confidence	27.35	26.82	173
Not too much confidence	25.92	26.67	172
No confidence	21.56	21.40	138
I don’t know	13.00	13.33	86
Skipped	0.26	0.31	2

Table B.49: Facebook; N = 632
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	4.29	3.96	25
A fair amount of confidence	14.35	13.45	85
Not too much confidence	26.40	27.22	172
No confidence	41.27	42.88	271
I don’t know	13.44	12.18	77
Skipped	0.25	0.32	2

Table B.50: Apple; N = 697
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	10.41	10.76	75
A fair amount of confidence	26.29	26.26	183
Not too much confidence	27.00	27.98	195
No confidence	22.20	21.81	152
I don’t know	13.84	12.91	90
Skipped	0.26	0.29	2

Table B.51: Microsoft; N = 597
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	10.85	10.89	65
A fair amount of confidence	33.08	32.66	195
Not too much confidence	26.89	27.14	162
No confidence	17.99	17.76	106
I don’t know	11.05	11.39	68
Skipped	0.14	0.17	1

Table B.52: Amazon; N = 685
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	10.60	10.95	75
A fair amount of confidence	29.53	29.34	201
Not too much confidence	25.51	25.40	174
No confidence	22.02	22.19	152
I don’t know	12.34	12.12	83
Skipped	0	0	0

Table B.53: Non-profit (e.g., OpenAI); N = 659
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	10.19	10.17	67
A fair amount of confidence	29.40	30.35	200
Not too much confidence	23.57	23.98	158
No confidence	13.65	13.66	90
I don’t know	23.04	21.70	143
Skipped	0.13	0.15	1

Table B.54: University researchers; N = 666
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	13.86	14.11	94
A fair amount of confidence	36.29	36.04	240
Not too much confidence	22.27	22.82	152
No confidence	12.75	12.31	82
I don’t know	14.70	14.56	97
Skipped	0.14	0.15	1

B.7 Trust of actors to manage AI

QUESTION:

How much confidence, if any, do you have in each of the following to manage the development and use of AI in the best interests of the public?

[Respondents were shown five items randomly selected from the list below. We included explainer text for actors not well known to the public; respondents could view the explainer text by hovering their mouse over the actor’s name. The items and the answer choices were shown in a matrix format.]

U.S. federal government
U.S. state governments
International organizations (e.g., United Nations, European Union)
The United Nations (UN)
An intergovernmental research organization (e.g., CERN)
- Explainer text for CERN: The European Organization for Nuclear Research, known as CERN, is a European research organization that operates the largest particle physics laboratory in the world.
Tech companies
Google
Facebook
Apple
Microsoft
Amazon
Non-government scientific organizations (e.g., AAAI)
- Explainer text for AAAI: Association for the Advancement of Artificial Intelligence (AAAI) is a non-government scientific organization that promotes research in, and responsible use of AI.
Partnership on AI, an association of tech companies, academics, and civil society groups

ANSWER CHOICES:

A great deal of confidence (3)
A fair amount of confidence (2)
Not too much confidence (1)
No confidence (0)
I don’t know

Table B.55: U.S. federal government; N = 743
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	6.86	6.59	49
A fair amount of confidence	20.26	20.19	150
Not too much confidence	28.44	28.67	213
No confidence	31.50	32.44	241
I don’t know	12.68	11.84	88
Skipped	0.25	0.27	2

Table B.56: U.S. state governments; N = 713
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	6.25	6.45	46
A fair amount of confidence	20.39	19.21	137
Not too much confidence	31.57	32.12	229
No confidence	29.65	30.72	219
I don’t know	11.69	11.22	80
Skipped	0.45	0.28	2

Table B.57: International organizations; N = 827
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	5.94	5.80	48
A fair amount of confidence	22.48	21.77	180
Not too much confidence	29.58	29.87	247
No confidence	26.81	27.45	227
I don’t know	14.81	14.87	123
Skipped	0.38	0.24	2

Table B.58: UN; N = 802
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	6.23	6.61	53
A fair amount of confidence	22.49	21.57	173
Not too much confidence	26.14	26.18	210
No confidence	31.90	31.55	253
I don’t know	12.64	13.59	109
Skipped	0.60	0.50	4

Table B.59: Intergovernmental research organizations (e.g., CERN); N = 747
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	6.69	7.10	53
A fair amount of confidence	30.51	29.72	222
Not too much confidence	23.89	24.10	180
No confidence	20.32	20.21	151
I don’t know	18.36	18.61	139
Skipped	0.22	0.27	2

Table B.60: Tech companies; N = 758
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	8.33	8.44	64
A fair amount of confidence	33.50	32.98	250
Not too much confidence	25.07	26.12	198
No confidence	19.88	20.45	155
I don’t know	12.81	11.74	89
Skipped	0.41	0.26	2

Table B.61: Google; N = 767
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	9.61	9.13	70
A fair amount of confidence	23.60	23.86	183
Not too much confidence	27.44	27.77	213
No confidence	25.13	25.03	192
I don’t know	13.75	13.95	107
Skipped	0.47	0.26	2

Table B.62: Facebook; N = 741
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	4.99	4.45	33
A fair amount of confidence	16.18	16.19	120
Not too much confidence	28.50	28.21	209
No confidence	36.95	38.46	285
I don’t know	13.14	12.42	92
Skipped	0.24	0.27	2

Table B.63: Apple; N = 775
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	8.25	8.39	65
A fair amount of confidence	25.10	24.90	193
Not too much confidence	29.08	28.65	222
No confidence	23.91	24.52	190
I don’t know	13.55	13.42	104
Skipped	0.12	0.13	1

Table B.64: Microsoft; N = 771
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	7.79	7.78	60
A fair amount of confidence	30.11	29.83	230
Not too much confidence	22.98	23.48	181
No confidence	24.10	24.38	188
I don’t know	14.68	14.14	109
Skipped	0.35	0.39	3

Table B.65: Amazon; N = 784
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	10.19	10.33	81
A fair amount of confidence	25.22	24.87	195
Not too much confidence	25.20	25.38	199
No confidence	24.53	24.87	195
I don’t know	14.87	14.54	114
Skipped	0	0	0

Table B.66: Non-government scientific organization (e.g., AAAI); N = 792
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	7.64	7.83	62
A fair amount of confidence	30.32	30.05	238
Not too much confidence	25.37	26.39	209
No confidence	15.03	14.65	116
I don’t know	21.46	20.83	165
Skipped	0.19	0.25	2

Table B.67: Partnership on AI; N = 780
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
A great deal of confidence	8.89	9.23	72
A fair amount of confidence	30.12	29.49	230
Not too much confidence	25.89	26.79	209
No confidence	16.33	15.77	123
I don’t know	18.64	18.59	145
Skipped	0.12	0.13	1

B.8 AI governance challenges

We would like you to consider some potential policy issues related to AI. Please consider the following:

[Respondents were shown five randomly-selected items from the list below, one item at a time. For ease of comprehension, we include the shorten labels used in the figures in square brackets.]

[Hiring bias] Fairness and transparency in AI used in hiring: Increasingly, employers are using AI to make hiring decisions. AI has the potential to make less biased hiring decisions than humans. But algorithms trained on biased data can lead to lead to hiring practices that discriminate against certain groups. Also, AI used in this application may lack transparency, such that human users do not understand what the algorithm is doing, or why it reaches certain decisions in specific cases.
[Criminal justice bias] Fairness and transparency in AI used in criminal justice: Increasingly, the criminal justice system is using AI to make sentencing and parole decisions. AI has the potential to make less biased hiring decisions than humans. But algorithms trained on biased data could lead to discrimination against certain groups. Also, AI used in this application may lack transparency such that human users do not understand what the algorithm is doing, or why it reaches certain decisions in specific cases.
[Disease diagnosis] Accuracy and transparency in AI used for disease diagnosis: Increasingly, AI software has been used to diagnose diseases, such as heart disease and cancer. One challenge is to make sure the AI can correctly diagnose those who have the disease and not mistakenly diagnose those who do not have the disease. Another challenge is that AI used in this application may lack transparency such that human users do not understand what the algorithm is doing, or why it reaches certain decisions in specific cases.
[Data privacy] Protect data privacy: Algorithms used in AI applications are often trained on vast amounts of personal data, including medical records, social media content, and financial transactions. Some worry that data used to train algorithms are not collected, used, and stored in ways that protect personal privacy.
[Autonomous vehicles] Make sure autonomous vehicles are safe: Companies are developing self-driving cars and trucks that require little or no input from humans. Some worry about the safety of autonomous vehicles for those riding in them as well as for other vehicles, cyclists, and pedestrians.
[Ditigal manipulation] Prevent AI from being used to spread fake and harmful content online: AI has been used by governments, private groups, and individuals to harm or manipulate internet users. For instance, automated bots have been used to generate and spread false and/or harmful news stories, audios, and videos.
[Cyber attacks] Prevent AI cyber attacks against governments, companies, organizations, and individuals: Computer scientists have shown that AI can be used to launch effective cyber attacks. AI could be used to hack into servers to steal sensitive information, shut down critical infrastructures like power grids or hospital networks, or scale up targeted phishing attacks.
[Surveillance] Prevent AI-assisted surveillance from violating privacy and civil liberties: AI can be used to process and analyze large amounts of text, photo, audio, and video data from social media, mobile communications, and CCTV cameras. Some worry that governments, companies, and employers could use AI to increase their surveillance capabilities.
[U.S.-China arms race] Prevent escalation of a U.S.-China AI arms race: Leading analysts believe that an AI arms race is beginning, in which the U.S. and China are investing billions of dollars to develop powerful AI systems for surveillance, autonomous weapons, cyber operations, propaganda, and command and control systems. Some worry that a U.S.-China arms race could lead to extreme dangers. To stay ahead, the U.S. and China may race to deploy advanced military AI systems that they do not fully understand or can control. We could see catastrophic accidents, such as a rapid, automated escalation involving cyber and nuclear weapons.
[Value alignment] Make sure AI systems are safe, trustworthy, and aligned with human values: As AI systems become more advanced, they will increasingly make decisions without human input. One potential fear is that AI systems, while performing jobs they are programmed to do, could unintentionally make decisions that go against the values of its human users, such as physically harming people.
[Autonomous weapons] Ban the use of lethal autonomous weapons (LAWs): Lethal autonomous weapons (LAWs) are military robots that can attack targets without control by humans. LAWs could reduce the use of human combatants on the battlefield. But some worry that the adoption of LAWs could lead to mass violence. Because they are cheap and easy to produce in bulk, national militaries, terrorists, and other groups could readily deploy LAWs.
[Technological unemployment] Guarantee a good standard of living for those who lose their jobs to automation: Some forecast that AI will increasingly be able to do jobs done by humans today. AI could potentially do the jobs of blue-collar workers, like truckers and factory workers, as well as the jobs of white-collar workers, like financial analysts or lawyers. Some worry that in the future, robots and computers can do most of the jobs that are done by humans today.
[Critical AI systems failure] Prevent critical AI systems failures: As AI systems become more advanced, they could be used by the military or in critical infrastructure, like power grids, highways, or hospital networks. Some worry that the failure of AI systems or unintentional accidents in these applications could cause 10 percent or more of all humans to die.

QUESTION:

In the next 10 years, how likely do you think it is that this AI governance challenge will impact large numbers of people in the U.S.?

ANSWER CHOICES:

Very unlikely: less than 5% chance (2.5%)
Unlikely: 5-20% chance (12.5%)
Somewhat unlikely: 20-40% chance (30%)
Equally likely as unlikely: 40-60% chance (50%)
Somewhat likely: 60-80% chance (70%)
Likely: 80-95% chance (87.5%)
Very likely: more than 95% chance (97.5%)
I don’t know

QUESTION:

In the next 10 years, how likely do you think it is that this AI governance challenge will impact large numbers of people around the world?

ANSWER CHOICES:

Very unlikely: less than 5% chance (2.5%)
Unlikely: 5-20% chance (12.5%)
Somewhat unlikely: 20-40% chance (30%)
Equally likely as unlikely: 40-60% chance (50%)
Somewhat likely: 60-80% chance (70%)
Likely: 80-95% chance (87.5%)
Very likely: more than 95% chance (97.5%)
I don’t know

QUESTION:

In the next 10 years, how important is it for tech companies and governments to carefully manage the following challenge?

ANSWER CHOICES:

Very important (3)
Somewhat important (2)
Not too important (1)
Not at all important (0)
I don’t know

Table B.68: Likelihood in the U.S. - Hiring bias; N = 760
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.57	2.63	20
Unlikely 5-20%	6.07	6.18	47
Somewhat unlikely 20-40%	10.86	10.92	83
Equally likely as unlikely 40-60%	22.27	22.50	171
Somewhat likely 60-80%	23.34	22.89	174
Likely 80-95%	12.39	12.76	97
Very likely > 95%	9.86	9.61	73
I don’t know	12.35	12.37	94
Skipped	0.29	0.13	1

Table B.69: Likelihood in the U.S. - Criminal justice bias; N = 778
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	4.94	4.50	35
Unlikely 5-20%	8.76	8.61	67
Somewhat unlikely 20-40%	13.25	12.85	100
Equally likely as unlikely 40-60%	21.23	21.08	164
Somewhat likely 60-80%	17.13	17.22	134
Likely 80-95%	12.28	12.60	98
Very likely > 95%	9.05	9.64	75
I don’t know	12.90	12.98	101
Skipped	0.45	0.51	4

Table B.70: Likelihood in the U.S. - Disease diagnosis; N = 767
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.79	2.61	20
Unlikely 5-20%	4.73	4.95	38
Somewhat unlikely 20-40%	10.18	9.52	73
Equally likely as unlikely 40-60%	23.12	23.21	178
Somewhat likely 60-80%	20.50	19.95	153
Likely 80-95%	13.43	13.95	107
Very likely > 95%	9.72	10.17	78
I don’t know	13.62	13.69	105
Skipped	1.91	1.96	15

Table B.71: Likelihood in the U.S. - Data privacy; N = 807
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.75	2.11	17
Unlikely 5-20%	4.53	4.58	37
Somewhat unlikely 20-40%	7.52	7.19	58
Equally likely as unlikely 40-60%	16.10	15.86	128
Somewhat likely 60-80%	18.81	19.33	156
Likely 80-95%	17.00	16.36	132
Very likely > 95%	20.59	21.69	175
I don’t know	10.87	10.78	87
Skipped	1.84	2.11	17

Table B.72: Likelihood in the U.S. - Autonomous vehicles; N = 796
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	3.65	3.64	29
Unlikely 5-20%	5.80	5.90	47
Somewhat unlikely 20-40%	10.93	10.43	83
Equally likely as unlikely 40-60%	16.17	16.33	130
Somewhat likely 60-80%	23.62	23.62	188
Likely 80-95%	15.78	15.45	123
Very likely > 95%	12.29	12.94	103
I don’t know	10.89	10.68	85
Skipped	0.87	1.01	8

Table B.73: Likelihood in the U.S. - Digital manipulation; N = 741
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.79	2.83	21
Unlikely 5-20%	3.24	3.10	23
Somewhat unlikely 20-40%	8.12	7.69	57
Equally likely as unlikely 40-60%	13.81	14.30	106
Somewhat likely 60-80%	16.58	16.33	121
Likely 80-95%	17.74	18.08	134
Very likely > 95%	23.45	23.62	175
I don’t know	12.49	12.15	90
Skipped	1.77	1.89	14

Table B.74: Likelihood in the U.S. - Cyber attacks; N = 745
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	3.36	2.42	18
Unlikely 5-20%	4.28	3.89	29
Somewhat unlikely 20-40%	8.44	8.59	64
Equally likely as unlikely 40-60%	15.45	15.84	118
Somewhat likely 60-80%	19.22	19.46	145
Likely 80-95%	15.96	15.30	114
Very likely > 95%	20.52	21.21	158
I don’t know	9.70	10.47	78
Skipped	3.07	2.82	21

Table B.75: Likelihood in the U.S. - Surveillance; N = 784
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.70	2.42	19
Unlikely 5-20%	2.92	2.81	22
Somewhat unlikely 20-40%	6.19	6.38	50
Equally likely as unlikely 40-60%	15.23	15.05	118
Somewhat likely 60-80%	18.95	18.75	147
Likely 80-95%	16.03	15.69	123
Very likely > 95%	23.52	24.23	190
I don’t know	12.15	12.12	95
Skipped	2.32	2.55	20

Table B.76: Likelihood in the U.S. - U.S.-China arms race; N = 766
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	3.24	3.26	25
Unlikely 5-20%	5.98	6.01	46
Somewhat unlikely 20-40%	10.01	10.84	83
Equally likely as unlikely 40-60%	18.74	18.41	141
Somewhat likely 60-80%	20.08	19.71	151
Likely 80-95%	13.17	12.79	98
Very likely > 95%	10.62	11.36	87
I don’t know	15.17	14.62	112
Skipped	3.00	3.00	23

Table B.77: Likelihood in the U.S. - Value alignment; N = 783
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	3.78	4.21	33
Unlikely 5-20%	7.30	6.90	54
Somewhat unlikely 20-40%	9.01	9.07	71
Equally likely as unlikely 40-60%	20.34	19.54	153
Somewhat likely 60-80%	19.26	19.28	151
Likely 80-95%	13.66	13.79	108
Very likely > 95%	12.96	13.67	107
I don’t know	12.43	12.26	96
Skipped	1.26	1.28	10

Table B.78: Likelihood in the U.S. - Autonomous weapons; N = 757
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	6.22	5.94	45
Unlikely 5-20%	10.36	9.38	71
Somewhat unlikely 20-40%	12.75	12.68	96
Equally likely as unlikely 40-60%	18.91	19.02	144
Somewhat likely 60-80%	15.72	15.72	119
Likely 80-95%	11.44	11.76	89
Very likely > 95%	10.72	11.23	85
I don’t know	11.99	12.29	93
Skipped	1.89	1.98	15

Table B.79: Likelihood in the U.S. - Technological unemployment; N = 738
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	3.08	2.98	22
Unlikely 5-20%	5.80	5.69	42
Somewhat unlikely 20-40%	11.00	11.11	82
Equally likely as unlikely 40-60%	17.74	17.62	130
Somewhat likely 60-80%	17.16	17.75	131
Likely 80-95%	14.86	14.91	110
Very likely > 95%	15.75	15.99	118
I don’t know	12.84	12.20	90
Skipped	1.75	1.76	13

Table B.80: Likelihood in the U.S. - Critical AI systems failure; N = 778
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	6.98	6.43	50
Unlikely 5-20%	7.94	7.58	59
Somewhat unlikely 20-40%	12.26	12.98	101
Equally likely as unlikely 40-60%	20.36	20.31	158
Somewhat likely 60-80%	15.59	15.42	120
Likely 80-95%	12.25	11.83	92
Very likely > 95%	9.36	10.15	79
I don’t know	14.85	14.78	115
Skipped	0.41	0.51	4

Table B.81: Likelihood around the world - Hiring bias; N = 760
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.95	3.03	23
Unlikely 5-20%	5.47	5.00	38
Somewhat unlikely 20-40%	8.54	8.55	65
Equally likely as unlikely 40-60%	20.23	21.45	163
Somewhat likely 60-80%	21.55	21.32	162
Likely 80-95%	13.68	13.55	103
Very likely > 95%	12.20	12.11	92
I don’t know	15.04	14.61	111
Skipped	0.35	0.39	3

Table B.82: Likelihood around the world - Criminal justice bias; N = 778
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	4.44	4.24	33
Unlikely 5-20%	8.06	7.71	60
Somewhat unlikely 20-40%	10.96	10.80	84
Equally likely as unlikely 40-60%	19.17	19.41	151
Somewhat likely 60-80%	18.29	18.25	142
Likely 80-95%	13.09	13.62	106
Very likely > 95%	9.38	9.90	77
I don’t know	16.38	15.94	124
Skipped	0.23	0.13	1

Table B.83: Likelihood around the world - Disease diagnosis; N = 767
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.31	2.35	18
Unlikely 5-20%	4.18	4.17	32
Somewhat unlikely 20-40%	9.93	9.13	70
Equally likely as unlikely 40-60%	21.28	20.99	161
Somewhat likely 60-80%	20.47	20.47	157
Likely 80-95%	15.00	15.38	118
Very likely > 95%	10.94	11.47	88
I don’t know	15.80	15.91	122
Skipped	0.09	0.13	1

Table B.84: Likelihood around the world - Data privacy; N = 807
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.86	2.23	18
Unlikely 5-20%	2.92	2.60	21
Somewhat unlikely 20-40%	8.32	8.30	67
Equally likely as unlikely 40-60%	13.79	14.75	119
Somewhat likely 60-80%	19.07	18.84	152
Likely 80-95%	18.43	18.22	147
Very likely > 95%	21.09	21.81	176
I don’t know	13.34	13.01	105
Skipped	0.19	0.25	2

Table B.85: Likelihood around the world - Autonomous vehicles; N = 796
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	3.77	3.52	28
Unlikely 5-20%	5.25	5.65	45
Somewhat unlikely 20-40%	12.37	11.68	93
Equally likely as unlikely 40-60%	16.74	17.21	137
Somewhat likely 60-80%	21.09	21.11	168
Likely 80-95%	14.13	14.45	115
Very likely > 95%	12.04	12.19	97
I don’t know	13.99	13.57	108
Skipped	0.63	0.63	5

Table B.86: Likelihood around the world - Digital manipulation; N = 741
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	1.98	2.16	16
Unlikely 5-20%	1.67	1.48	11
Somewhat unlikely 20-40%	7.34	7.29	54
Equally likely as unlikely 40-60%	12.68	12.96	96
Somewhat likely 60-80%	17.18	17.00	126
Likely 80-95%	21.22	21.73	161
Very likely > 95%	22.31	22.00	163
I don’t know	15.24	14.98	111
Skipped	0.39	0.40	3

Table B.87: Likelihood around the world - Cyber attacks; N = 745
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	1.08	1.21	9
Unlikely 5-20%	4.95	4.03	30
Somewhat unlikely 20-40%	4.76	5.10	38
Equally likely as unlikely 40-60%	16.95	16.64	124
Somewhat likely 60-80%	18.94	19.73	147
Likely 80-95%	19.13	19.06	142
Very likely > 95%	20.57	20.40	152
I don’t know	13.20	13.42	100
Skipped	0.42	0.40	3

Table B.88: Likelihood around the world - Surveillance; N = 784
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	1.26	1.40	11
Unlikely 5-20%	3.55	3.19	25
Somewhat unlikely 20-40%	5.12	5.36	42
Equally likely as unlikely 40-60%	14.26	14.41	113
Somewhat likely 60-80%	18.90	19.13	150
Likely 80-95%	20.30	19.77	155
Very likely > 95%	22.62	22.70	178
I don’t know	13.93	13.90	109
Skipped	0.07	0.13	1

Table B.89: Likelihood around the world - U.S.-China arms race; N = 766
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	3.21	3.13	24
Unlikely 5-20%	4.61	4.83	37
Somewhat unlikely 20-40%	7.70	7.83	60
Equally likely as unlikely 40-60%	19.50	19.19	147
Somewhat likely 60-80%	20.71	20.76	159
Likely 80-95%	14.99	14.75	113
Very likely > 95%	12.46	12.92	99
I don’t know	16.61	16.32	125
Skipped	0.22	0.26	2

Table B.90: Likelihood around the world - Value alignment; N = 783
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.70	2.94	23
Unlikely 5-20%	4.66	4.60	36
Somewhat unlikely 20-40%	8.80	8.81	69
Equally likely as unlikely 40-60%	19.92	19.41	152
Somewhat likely 60-80%	18.97	18.77	147
Likely 80-95%	15.57	15.33	120
Very likely > 95%	14.93	15.71	123
I don’t know	14.44	14.43	113
Skipped	0	0	0

Table B.91: Likelihood around the world - Autonomous weapons; N = 757
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	3.72	3.70	28
Unlikely 5-20%	7.04	5.42	41
Somewhat unlikely 20-40%	9.42	9.64	73
Equally likely as unlikely 40-60%	17.23	17.44	132
Somewhat likely 60-80%	16.08	15.85	120
Likely 80-95%	16.35	17.04	129
Very likely > 95%	14.87	15.19	115
I don’t know	15.20	15.59	118
Skipped	0.09	0.13	1

Table B.92: Likelihood around the world - Technological unemployment; N = 738
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.76	2.57	19
Unlikely 5-20%	4.92	4.47	33
Somewhat unlikely 20-40%	8.31	8.81	65
Equally likely as unlikely 40-60%	18.36	18.16	134
Somewhat likely 60-80%	19.90	21.00	155
Likely 80-95%	14.78	14.50	107
Very likely > 95%	16.71	16.67	123
I don’t know	13.77	13.41	99
Skipped	0.51	0.41	3

Table B.93: Likelihood around the world - Critical AI systems failure; N = 778
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	5.36	5.27	41
Unlikely 5-20%	8.07	7.97	62
Somewhat unlikely 20-40%	10.75	10.41	81
Equally likely as unlikely 40-60%	18.03	17.87	139
Somewhat likely 60-80%	16.71	16.84	131
Likely 80-95%	13.09	13.11	102
Very likely > 95%	11.23	11.83	92
I don’t know	16.76	16.71	130
Skipped	0	0	0

Table B.94: Issue importance - Hiring bias; N = 760
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	56.86	57.11	434
Somewhat important	22.11	22.76	173
Not too important	6.56	6.05	46
Not at all important	1.50	1.58	12
I don’t know	12.98	12.50	95
Skipped	0	0	0

Table B.95: Issue importance - Criminal justice bias; N = 778
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	56.08	56.68	441
Somewhat important	21.78	22.49	175
Not too important	6.65	5.91	46
Not at all important	1.83	1.67	13
I don’t know	13.66	13.24	103
Skipped	0	0	0

Table B.96: Issue importance - Disease diagnosis; N = 767
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	55.60	56.98	437
Somewhat important	22.37	21.25	163
Not too important	6.68	6.91	53
Not at all important	1.98	1.83	14
I don’t know	13.26	12.91	99
Skipped	0.11	0.13	1

Table B.97: Issue importance - Data privacy; N = 807
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	63.65	64.93	524
Somewhat important	17.65	17.10	138
Not too important	4.76	4.71	38
Not at all important	1.71	1.36	11
I don’t know	12.05	11.65	94
Skipped	0.19	0.25	2

Table B.98: Issue importance - Autonomous vehicles; N = 796
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	58.70	59.55	474
Somewhat important	22.36	21.73	173
Not too important	6.13	6.28	50
Not at all important	1.44	1.63	13
I don’t know	11.15	10.55	84
Skipped	0.22	0.25	2

Table B.99: Issue importance - Digital manipulation; N = 741
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	57.66	58.30	432
Somewhat important	18.75	18.08	134
Not too important	6.25	6.48	48
Not at all important	3.11	2.97	22
I don’t know	14.16	14.04	104
Skipped	0.08	0.13	1

Table B.100: Issue importance - Cyber attacks; N = 745
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	62.12	61.21	456
Somewhat important	17.80	18.39	137
Not too important	7.07	7.38	55
Not at all important	1.14	1.07	8
I don’t know	11.88	11.95	89
Skipped	0	0	0

Table B.101: Issue importance - Surveillance; N = 784
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	58.54	58.80	461
Somewhat important	19.33	19.26	151
Not too important	6.40	6.63	52
Not at all important	1.73	1.66	13
I don’t know	13.93	13.52	106
Skipped	0.07	0.13	1

Table B.102: Issue importance - U.S.-China arms race; N = 766
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	55.88	55.74	427
Somewhat important	19.44	19.71	151
Not too important	7.07	7.57	58
Not at all important	2.38	2.35	18
I don’t know	15.13	14.49	111
Skipped	0.10	0.13	1

Table B.103: Issue importance - Value alignment; N = 783
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	56.46	56.45	442
Somewhat important	20.49	20.95	164
Not too important	6.69	6.64	52
Not at all important	1.56	1.66	13
I don’t know	14.80	14.30	112
Skipped	0	0	0

Table B.104: Issue importance - Autonomous weapons; N = 757
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	58.32	57.73	437
Somewhat important	20.00	19.55	148
Not too important	5.52	5.94	45
Not at all important	1.23	1.45	11
I don’t know	14.94	15.32	116
Skipped	0	0	0

Table B.105: Issue importance - Technological unemployment; N = 738
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	54.12	54.34	401
Somewhat important	22.07	22.49	166
Not too important	6.50	6.91	51
Not at all important	2.83	2.44	18
I don’t know	14.39	13.69	101
Skipped	0.09	0.14	1

Table B.106: Issue importance - Critical AI systems failure; N = 778
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very important	52.63	53.86	419
Somewhat important	21.10	20.44	159
Not too important	7.98	8.10	63
Not at all important	2.93	2.44	19
I don’t know	15.36	15.17	118
Skipped	0	0	0

B.9 Survey experiment: comparing perceptions of U.S. vs. China AI research and development

[Respondents were presented with one randomly-selected question from the two below.]

QUESTIONS:

Compared with other industrialized countries, how would you rate the U.S. in AI research and development?
Compared with other industrialized countries, how would you rate China in AI research and development?

ANSWER CHOICES:

Best in the world (3)
Above average (2)
Average (1)
Below average (0)
I don’t know

Table B.107: Perceptions of research and development - U.S.; N = 988
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Best in the world	9.73	10.02	99
Above average	36.16	37.55	371
Average	26.09	24.70	244
Below average	4.99	4.96	49
I don’t know	23.03	22.77	225
Skipped	0	0	0

Table B.108: Perceptions of research and development - China; N = 1012
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Best in the world	7.33	7.41	75
Above average	45.40	46.64	472
Average	16.66	15.81	160
Below average	3.93	3.66	37
I don’t know	26.68	26.48	268
Skipped	0	0	0

B.10 Survey experiment: U.S.-China arms race

[All respondents were presented with the following prompt.]

We want to understand your thoughts on some important issues in the news today. Please read the short news article below.

Leading analysts believe that an “AI arms race” is beginning, in which the U.S. and China are investing billions of dollars to develop powerful AI systems for surveillance, autonomous weapons, cyber operations, propaganda, and command and control systems.

[Respondents were randomly assigned to one of the four experimental groups listed below.]

B.10.1 Control

[No additional text.]

B.10.2 Nationalism treatment

Some leaders in the U.S. military and tech industry argue that the U.S. government should invest much more resources in AI research to ensure that the U.S.’s AI capabilities stay ahead of China’s. Furthermore, they argue that the U.S. government should partner with American tech companies to develop advanced AI systems, particularly for military use.

According to a leaked memo produced by a senior National Security Council official, China has “assembled the basic components required for winning the Al arms race…Much like America’s success in the competition for nuclear weapons, China’s 21st Century Manhattan Project sets them on a path to getting there first.”

B.10.3 War risks treatment

Some prominent thinkers are concerned that a U.S.-China arms race could lead to extreme dangers. To stay ahead, the U.S. and China may race to deploy advanced military AI systems that they do not fully understand or can control. We could see catastrophic accidents, such as a rapid, automated escalation involving cyber and nuclear weapons.

“Competition for AI superiority at [the] national level [is the] most likely cause of World War Three,” warned Elon Musk, the CEO of Tesla and SpaceX.

B.10.4 Common humanity treatment

Some prominent thinkers are concerned that a U.S.-China arms race could lead to extreme dangers. To stay ahead, the U.S. and China may race to deploy advanced military AI systems that they do not fully understand or can control. We could see catastrophic accidents, such as a rapid, automated escalation involving cyber and nuclear weapons.

“Unless we learn how to prepare for, and avoid, the potential risks, AI could be the worst event in the history of our civilization. It brings dangers, like powerful autonomous weapons,” warned the late Stephen Hawking, one of the world’s most prominent physicists. At the same time, he said that with proper management of the technology, researchers “can create AI for the good of the world.”

[The order of the next two questions is randomized.]

QUESTION:

How much do you agree or disagree with the following statement?

The U.S. should invest more in AI military capabilities to make sure it doesn’t fall behind China’s, even if doing so may exacerbate the arms race. For instance, the U.S. could increase AI research funding for the military and universities. It could also collaborate with American tech companies to develop AI for military use.

ANSWER CHOICES:

Strongly agree (2)
Somewhat agree (1)
Neither agree nor disagree (0)
Somewhat disagree (-1)
Strongly disagree (-2)
I don’t know

QUESTION:

How much do you agree or disagree with the following statement?

The U.S. should work hard to cooperate with China to avoid the dangers of an AI arms race, even if doing so requires giving up some of the U.S.’s advantages. Cooperation could include collaborations between American and Chinese AI research labs, or the U.S. and China creating and committing to common safety standards.

ANSWER CHOICES:

Strongly agree (2)
Somewhat agree (1)
Neither agree nor disagree (0)
Somewhat disagree (-1)
Strongly disagree (-2)
I don’t know

Table B.109: Responses to statement that U.S. should invest more in AI military capabilities - Control; N = 510
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	23.38	24.31	124
Somewhat agree	25.99	25.88	132
Neither agree nor disagree	23.48	22.75	116
Somewhat disagree	8.88	8.82	45
Strongly disagree	4.93	4.71	24
I don’t know	13.34	13.53	69
Skipped	0	0	0

Table B.110: Responses to statement that U.S. should invest more in AI military capabilities - Treatment 1: Pro-nationalist; N = 505
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	20.88	20.40	103
Somewhat agree	26.89	27.52	139
Neither agree nor disagree	21.79	22.18	112
Somewhat disagree	11.69	12.28	62
Strongly disagree	5.30	5.35	27
I don’t know	13.45	12.28	62
Skipped	0	0	0

Table B.111: Responses to statement that U.S. should invest more in AI military capabilities - Treatment 2: Risks of arms race; N = 493
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	18.26	19.07	94
Somewhat agree	27.85	27.38	135
Neither agree nor disagree	21.69	20.28	100
Somewhat disagree	12.87	13.79	68
Strongly disagree	6.88	6.90	34
I don’t know	12.45	12.58	62
Skipped	0	0	0

Table B.112: Responses to statement that U.S. should invest more in AI military capabilities - Treatment 3: One common humanity; N = 492
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	22.38	20.53	101
Somewhat agree	27.29	27.85	137
Neither agree nor disagree	24.37	23.98	118
Somewhat disagree	6.73	7.11	35
Strongly disagree	6.17	6.91	34
I don’t know	13.07	13.62	67
Skipped	0	0	0

Table B.113: Responses to statement that U.S. should work hard to cooperate with China to avoid dangers of AI arms race - Control; N = 510
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	22.34	22.55	115
Somewhat agree	26.16	26.27	134
Neither agree nor disagree	22.02	20.59	105
Somewhat disagree	8.29	9.02	46
Strongly disagree	7.38	7.45	38
I don’t know	13.59	13.92	71
Skipped	0.21	0.20	1

Table B.114: Responses to statement that U.S. should work hard to cooperate with China to avoid dangers of AI arms race - Treatment 1: Pro-nationalist; N = 505
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	18.51	18.81	95
Somewhat agree	27.35	28.12	142
Neither agree nor disagree	20.08	20.99	106
Somewhat disagree	10.09	9.90	50
Strongly disagree	8.45	7.92	40
I don’t know	15.51	14.26	72
Skipped	0	0	0

Table B.115: Responses to statement that U.S. should work hard to cooperate with China to avoid dangers of AI arms race - Treatment 2: Risks of arms race; N = 493
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	24.97	25.96	128
Somewhat agree	25.32	25.15	124
Neither agree nor disagree	21.53	20.49	101
Somewhat disagree	9.83	9.94	49
Strongly disagree	5.84	5.68	28
I don’t know	12.51	12.78	63
Skipped	0	0	0

Table B.116: Responses to statement that U.S. should work hard to cooperate with China to avoid dangers of AI arms race - Treatment 3: One common humanity; N = 492
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	23.63	24.19	119
Somewhat agree	27.52	28.46	140
Neither agree nor disagree	21.31	20.33	100
Somewhat disagree	8.50	7.32	36
Strongly disagree	6.72	6.91	34
I don’t know	12.31	12.80	63
Skipped	0	0	0

B.11 Issue areas for possible U.S.-China cooperation

QUESTION:

For the following issues, how likely is it that the U.S. and China can cooperate?

[Respondents were presented with three issues from the list below. All three issues were presented on the same page; the order that they appeared was randomized.]

Prevent AI cyber attacks against governments, companies, organizations, and individuals.
Prevent AI-assisted surveillance from violating privacy and civil liberties.
Make sure AI systems are safe, trustworthy, and aligned with human values.
Ban the use of lethal autonomous weapons.
Guarantee a good standard of living for those who lose their jobs to automation.

ANSWER CHOICES:

Very unlikely: less than 5% chance (2.5%)
Unlikely: 5-20% chance (12.5%)
Somewhat unlikely: 20-40% chance (30%)
Equally likely as unlikely: 40-60% chance (50%)
Somewhat likely: 60-80% chance (70%)
Likely: 80-95% chance (87.5%)
Very likely: more than 95% chance (97.5%)
I don’t know

Table B.117: Likelihood of cooperation with China - Prevent AI cyber attacks against governments, companies, organizations, and individuals; N = 1173
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely 5%	9.20	8.95	105
Unlikely 5-20%	10.26	10.49	123
Somewhat unlikely 20-40%	17.56	17.22	202
Equally likely as unlikely 40-60%	23.55	23.36	274
Somewhat likely 60-80%	13.77	13.73	161
Likely 80-95%	6.98	7.25	85
Very likely > 95%	4.14	4.18	49
I don’t know	14.45	14.75	173
Skipped	0.08	0.09	1

Table B.118: Likelihood of cooperation with China - Prevent AI-assisted surveillance from violating privacy and civil liberties; N = 1140
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely 5%	12.43	12.37	141
Unlikely 5-20%	12.78	13.33	152
Somewhat unlikely 20-40%	19.48	19.74	225
Equally likely as unlikely 40-60%	21.93	20.70	236
Somewhat likely 60-80%	10.59	10.79	123
Likely 80-95%	4.02	4.12	47
Very likely > 95%	3.82	4.12	47
I don’t know	14.87	14.74	168
Skipped	0.08	0.09	1

Table B.119: Likelihood of cooperation with China - Make sure AI systems are safe, trustworthy, and aligned with human values; N = 1226
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely 5%	6.34	6.53	80
Unlikely 5-20%	9.07	8.97	110
Somewhat unlikely 20-40%	16.79	16.88	207
Equally likely as unlikely 40-60%	26.32	25.53	313
Somewhat likely 60-80%	14.84	14.85	182
Likely 80-95%	7.35	7.26	89
Very likely > 95%	5.77	5.87	72
I don’t know	13.38	13.95	171
Skipped	0.14	0.16	2

Table B.120: Likelihood of cooperation with China - Ban the use of lethal autonomous weapons; N = 1226
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely 5%	12.28	12.32	151
Unlikely 5-20%	11.14	10.85	133
Somewhat unlikely 20-40%	14.03	14.03	172
Equally likely as unlikely 40-60%	23.98	23.65	290
Somewhat likely 60-80%	10.15	10.60	130
Likely 80-95%	6.67	6.93	85
Very likely > 95%	5.69	5.46	67
I don’t know	15.91	15.99	196
Skipped	0.14	0.16	2

Table B.121: Likelihood of cooperation with China - Guarantee a good standard of living for those who lose their jobs to automation; N = 1235
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely 5%	13.19	13.36	165
Unlikely 5-20%	13.01	13.28	164
Somewhat unlikely 20-40%	18.26	18.46	228
Equally likely as unlikely 40-60%	22.81	22.19	274
Somewhat likely 60-80%	9.46	9.39	116
Likely 80-95%	5.08	5.18	64
Very likely > 95%	4.27	4.53	56
I don’t know	13.78	13.44	166
Skipped	0.14	0.16	2

B.12 Trend across time: job creation or job loss

QUESTION:

How much do you agree or disagree with the following statement?

[Respondents were presented with one statement randomly selected from the list below.]

In general, automation and AI will create more jobs than they will eliminate.
In general, automation and AI will create more jobs than they will eliminate in 10 years.
In general, automation and AI will create more jobs than they will eliminate in 20 years.
In general, automation and AI will create more jobs than they will eliminate in 50 years.

ANSWER CHOICES:

Strongly agree (2)
Agree (1)
Disagree (-1)
Strongly disagree (-2)
I don’t know

Table B.122: Responses to statement that automation and AI will create more jobs than they will eliminate - No time frame; N = 484
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	6.37	6.82	33
Agree	20.19	18.18	88
Disagree	27.39	28.10	136
Strongly disagree	21.43	22.31	108
Don’t know	24.45	24.38	118
Skipped	0.17	0.21	1

Table B.123: Responses to statement that automation and AI will create more jobs than they will eliminate - 10 years; N = 510
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	3.40	3.53	18
Agree	17.67	18.04	92
Disagree	30.03	29.02	148
Strongly disagree	22.85	23.92	122
Don’t know	26.04	25.49	130
Skipped	0	0	0

Table B.124: Responses to statement that automation and AI will create more jobs than they will eliminate - 20 years; N = 497
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	3.69	4.02	20
Agree	17.82	17.10	85
Disagree	31.02	30.99	154
Strongly disagree	21.31	21.73	108
Don’t know	25.98	25.96	129
Skipped	0.18	0.20	1

Table B.125: Responses to statement that automation and AI will create more jobs than they will eliminate - 50 years; N = 509
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly agree	6.77	6.48	33
Agree	15.37	15.52	79
Disagree	35.35	35.56	181
Strongly disagree	18.82	18.27	93
Don’t know	23.69	24.17	123
Skipped	0	0	0

B.13 High-level machine intelligence: forecasting timeline

QUESTION:

The following questions ask about high-level machine intelligence. We have high-level machine intelligence when machines are able to perform almost all tasks that are economically relevant today better than the median human (today) at each task. These tasks include asking subtle common-sense questions such as those that travel agents would ask. For the following questions, you should ignore tasks that are legally or culturally restricted to humans, such as serving on a jury.

In your opinion, how likely is it that high-level machine intelligence will exist in 10 years? 20 years? 50 years? For each prediction, please use the slider to indicate the percent chance that you think high-level machine intelligence will exist. 0% chance means it will certainly not exist. 100% chance means it will certainly exist.

______ In 10 years?

______ In 20 years?

______ In 50 years?

ANSWER CHOICES:

Very unlikely: less than 5% chance (2.5%)
Unlikely: 5-20% chance (12.5%)
Somewhat unlikely: 20-40% chance (30%)
Equally likely as unlikely: 40-60% chance (50%)
Somewhat likely: 60-80% chance (70%)
Likely: 80-95% chance (87.5%)
Very likely: more than 95% chance (97.5%)
I don’t know

Table B.126: Forecasting high-level machine intelligence - 10 years; N = 2000
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	4.46	4.50	90
Unlikely 5-20%	8.19	8.20	164
Somewhat unlikely 20-40%	14.84	14.75	295
Equally likely as unlikely 40-60%	20.34	19.95	399
Somewhat likely 60-80%	21.08	21.25	425
Likely 80-95%	10.69	10.65	213
Very likely > 95%	7.40	7.85	157
I don’t know	12.91	12.75	255
Skipped	0.09	0.10	2

Table B.127: Forecasting high-level machine intelligence - 20 years; N = 2000
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	1.52	1.45	29
Unlikely 5-20%	2.73	2.95	59
Somewhat unlikely 20-40%	6.26	5.85	117
Equally likely as unlikely 40-60%	16.83	16.40	328
Somewhat likely 60-80%	18.17	18.65	373
Likely 80-95%	22.25	22.25	445
Very likely > 95%	17.91	18.30	366
I don’t know	14.18	14.00	280
Skipped	0.15	0.15	3

Table B.128: Forecasting high-level machine intelligence - 50 years; N = 2000
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Very unlikely < 5%	2.28	2.30	46
Unlikely 5-20%	1.66	1.55	31
Somewhat unlikely 20-40%	2.75	2.75	55
Equally likely as unlikely 40-60%	10.08	9.90	198
Somewhat likely 60-80%	12.33	12.20	244
Likely 80-95%	14.43	14.50	290
Very likely > 95%	40.86	41.15	823
I don’t know	15.52	15.55	311
Skipped	0.09	0.10	2

B.14 Support for developing high-level machine intelligence

QUESTION:

How much do you support or oppose the development of high-level machine intelligence?

ANSWER CHOICES:

Strongly support
Somewhat support
Neither support nor oppose
Somewhat oppose
Strongly oppose
I don’t know

Table B.129: Support for developing high-level machine intelligence; N = 2000
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Strongly support	7.78	8.10	162
Somewhat support	23.58	23.30	466
Neither support nor oppose	29.40	28.75	575
Somewhat oppose	16.19	16.60	332
Strongly oppose	11.02	11.10	222
I don’t know	11.94	12.05	241
Skipped	0.09	0.10	2

B.15 Expected outcome of high-level machine intelligence

QUESTION:

Suppose that high-level machine intelligence could be developed one day. How positive or negative do you expect the overall impact of high-level machine intelligence to be on humanity in the long run?

ANSWER CHOICES:

Extremely good
On balance good
More or less neutral
On balance bad
Extremely bad, possibly human extinction
I don’t know

Table B.130: Expected outcome of high-level machine intelligence; N = 2000
Answer choices	Percentages (weighted)	Percentages (unweighted)	Raw frequencies
Extremely good	5.35	5.45	109
On balance good	21.28	21.25	425
More or less neutral	21.00	21.10	422
On balance bad	22.38	23.10	462
Extremely bad, possibly human extinction	11.66	11.55	231
Don’t know	18.25	17.45	349
Skipped	0.09	0.10	2