Warmth, Competence, and the Determinants of Trust in Artificial Intelligence: A Cross-Sectional Survey from China

Figures & data

Table 1. Factors potentially related to trust in artificial intelligence (AI).

Dimension	Factor	Construct definition
Trustor	Agency	People’s sense of control of AI (High-Level Expert Group on Artificial Intelligence [AI HLEG], 2019)
	Emotional experience	The sense of security and comfort derived from using AI (Komiak & Benbasat, 2004)
	Familiarity	An individual’s knowledge of AI based on past interactive experiences and learning (Gefen et al., 2003)
	Hedonic motivation	A form of intrinsic motivation that refers to the pleasure or enjoyment derived from using AI (Gursoy et al., 2019)
	Technology trust	An individual’s innate tendency to rely on AI without requiring exhaustive proof of its reliability (Delgosha & Hajiheydari, 2021)
	Trust propensity	An individual’s stable, internal inclination to trust others (McKnight et al., 1998)
Trustee	Accountability	The extent to which AI can take responsibility for its actions and decisions (AI HLEG, 2019)
	Anthropomorphism	The degree to which AI resembles human forms, characteristics, or behaviors (Gursoy et al., 2019)
	Competence	An individual’s perception regarding the ability and performance of AI to accomplish tasks as required (Harris-Watson et al., 2023)
	Fairness	The principle of fair and equal treatment for different users, groups, and individuals (AI HLEG, 2019)
	Privacy protection	The degree to which individuals perceive that their personal information and data are safe on AI systems (Salisbury et al., 2001)
	Robustness	The reliability of the operation of AI and the repeatability of results (AI HLEG, 2019)
	Transparency	The degree to which the inner workings and decision-making processes of an AI system are understandable by individuals (AI HLEG, 2019)
	Warmth	The extent to which users perceive that AI represents their interests (Harris-Watson et al., 2023)
Context	Facilitating conditions	The extent to which individuals perceive environmental objective factors that facilitate the use of AI (Venkatesh et al., 2003)
	Perceived benefits	The degree to which individuals believe that they will become better off from using AI (Kim et al., 2008)
	Safeguards	Laws or regulations to protect the responsible application of AI (Lockey et al., 2020)
	Social influence	The degree to which individuals perceive that influential people in their network support AI adoption (Venkatesh et al., 2003)
	Uncertainties	The perception and evaluation of uncertainties or doubts individuals have regarding AI (Khan et al., 2014)

Table 2. Key demographic characteristics of participants.

Variable	Category	Northeastern	Eastern	Central	Western	All sample
		139 (0.06)	851 (0.39)	589 (0.27)	608 (0.28)
Sex	Male	69 (0.50)	394 (0.46)	275 (0.47)	272 (0.45)	1176 (0.54)
	Female	70 (0.50)	457 (0.54)	314 (0.53)	336 (0.55)	1011 (0.46)
Age	18–29	36 (0.26)	236 (0.28)	161 (0.27)	143 (0.24)	576 (0.26)
	30–39	37 (0.27)	209 (0.25)	142 (0.24)	181 (0.30)	569 (0.26)
	40–60	66 (0.47)	406 (0.47)	286 (0.49)	274 (0.46)	1042 (0.48)
Education	High school or below	20 (0.14)	98 (0.12)	112 (0.19)	110 (0.18)	340 (0.16)
	Three-year college	33 (0.24)	208 (0.24)	147 (0.25)	162 (0.27)	550 (0.25)
	Bachelor	73 (0.53)	458 (0.54)	292 (0.50)	303 (0.50)	1126 (0.51)
	Master or above	13 (0.09)	87 (0.10)	38 (0.06)	33 (0.05)	171 (0.08)
Subjective social status	Low (≤ 5)	93 (0.67)	431 (0.51)	359 (0.61)	374 (0.62)	1257 (0.57)
	High (> 5)	46 (0.33)	420 (0.49)	230 (0.39)	234 (0.38)	930 (0.43)
	Mean	4.68 (1.67)	5.35 (1.56)	5.11 (1.58)	4.97 (1.63)	5.13 (1.60)

Note. The numbers shown in the table are the numbers of participants of a certain category in a region or the whole sample with the proportion in parentheses, except for the last row, in which “Mean” indicates the average score of reported subjective social status with the standard deviation in parentheses.

Table 3. Descriptive statistics and reliability and validity indices of 22 key measures.

Measure	Mean	SD	Cronbach’s α	Composite reliability	AVE
Trust in AI	3.153	0.687	0.869	0.898	0.527
Factors related to the trustor
Technology trust	3.711	0.728	0.803	0.884	0.718
Emotional experience	3.644	0.722	0.807	0.886	0.722
Hedonic motivation	3.642	0.854	0.893	0.933	0.824
Agency	3.570	0.746	0.786	0.861	0.609
Trust propensity	3.569	0.754	0.838	0.892	0.676
Familiarity	3.448	0.575	0.805	0.865	0.561
Factors related to the trustee
Fairness	3.779	0.866	0.907	0.941	0.843
Competence	3.720	0.633	0.738	0.851	0.655
Transparency	3.598	0.750	0.776	0.856	0.598
Robustness	3.330	0.669	0.753	0.840	0.570
Warmth	3.312	0.731	0.864	0.898	0.596
Accountability	3.291	0.853	0.811	0.875	0.637
Anthropomorphism	2.582	1.051	0.920	0.944	0.808
Privacy protection	2.117	0.775	0.885	0.906	0.616
Factors related to the context
Perceived benefits	4.214	0.590	0.838	0.902	0.755
Uncertainties	4.155	0.665	0.846	0.883	0.561
Facilitating conditions	3.641	0.687	0.773	0.855	0.597
Social influence	3.258	0.799	0.874	0.915	0.729
Safeguard	3.128	0.899	0.904	0.929	0.722
Behavioral intention
Intention to share private data	3.116	0.887	0.874	0.914	0.726
Support for AI in HRM^a	2.899	0.880	0.793	0.866	0.618

Note. Scores of each measure ranged from 1 to 5. The factors are those identified as potentially related to trust in artificial intelligence (AI), ordered by their mean scores in each dimension. SD: standard deviation; AVE: average variance extracted; HRM: human resources management. ^aOnly data from participants who reported being currently employed at the time of the study (N = 1701) were included.

Figure 1. Trust in AI by participant group. (A) by three age groups, (B) by two levels of subjective social status, (C) by two levels of education, (D) by sex, and (E) by four economic regions in China. Error bars indicate one standard deviation.

Table 4. Correlation coefficients among the 22 key variables.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21
1 Trust in AI	–
2 Warmth	0.63	–
3 Competence	0.60	0.58	–
4 Robustness	0.57	0.59	0.47	–
5 Emotional experience	0.56	0.58	0.52	0.55	–
6 Safeguard	0.53	0.66	0.46	0.57	0.59	–
7 Facilitating conditions	0.53	0.54	0.52	0.50	0.59	0.50	–
8 Social influence	0.52	0.52	0.44	0.46	0.57	0.52	0.66	–
9 Hedonic motivation	0.50	0.51	0.42	0.45	0.71	0.45	0.49	0.50	–
10 Privacy protection	0.47	0.50	0.34	0.42	0.38	0.52	0.32	0.39	0.30	–
11 Agency	0.47	0.58	0.43	0.52	0.54	0.59	0.51	0.48	0.49	0.38	–
12 Transparency	0.45	0.54	0.42	0.49	0.54	0.56	0.50	0.48	0.48	0.37	0.72	–
13 Technology trust	0.44	0.41	0.38	0.44	0.42	0.37	0.42	0.38	0.38	0.26	0.39	0.36	–
14 Accountability	0.44	0.58	0.36	0.51	0.51	0.63	0.48	0.46	0.46	0.41	0.69	0.70	0.32	–
15 Perceived benefits	0.44	0.43	0.51	0.42	0.55	0.38	0.55	0.42	0.49	0.19	0.43	0.43	0.47	0.34	–
16 Fairness	0.43	0.56	0.42	0.53	0.49	0.48	0.46	0.38	0.40	0.29	0.52	0.48	0.39	0.50	0.45	–
17 Familiarity	0.31	0.27	0.24	0.25	0.33	0.28	0.40	0.32	0.27	0.18	0.33	0.24	0.23	0.23	0.26	0.21	–
18 Anthropomorphism	0.28	0.27	0.15	0.26	0.19	0.22	0.16	0.22	0.23	0.20	0.18	0.24	0.13	0.22	0.07**	0.11	0.07**	–
19 Trust propensity	0.28	0.30	0.24	0.31	0.27	0.33	0.28	0.29	0.23	0.18	0.30	0.25	0.49	0.27	0.25	0.27	0.15	0.12	–
20 Uncertainties	−.16	−.19	−.08	−.19	−.12	−.22	−.12	−.16	−.07**	−.41	−.11	−.11	−.03^	−.17	0.01^	−.09	−.1	−.04*	−.07**	–
21 Intention to share private data	0.60	0.56	0.45	0.47	0.45	0.51	0.45	0.45	0.39	0.59	0.45	0.43	0.37	0.44	0.34	0.38	0.25	0.23	0.22	−.19	–
22 Support for AI in HRM^a	0.47	0.45	0.38	0.38	0.43	0.41	0.41	0.45	0.38	0.35	0.38	0.36	0.28	0.37	0.33	0.32	0.21	0.29	0.19	−.17	0.42

Note. Apart from the two behavioral intention measures (i.e., intention to share private data and support for AI in HRM), the other 19 measures are ordered by the absolute values of their correlations with trust in AI (second column). The correlation coefficients without symbols are significant with ps < 0.001. For those with symbols, **, *, and ^ indicate statistical significance at the p < 0.01, p < 0.05, and p > 0.05 levels, respectively. The correlations between support for AI in HRM and other measures were based on only participants who reported being employed at the time of the study (N = 1701). AI: Artificial Intelligence; HRM: human resources management.

Figure 2. Cumulative adjusted R² in a stepwise regression analysis of trust in artificial intelligence. Each bar represents the cumulative R² contribution of a factor that was entered into the regression model. Nine factors were selected, and they are ordered by their selection sequence. The bars are color-coded to indicate the trust dimension to which each selected factor belongs.

Figure 3. Interactions among warmth, competence, and three other factors related to trust in artificial intelligence (AI). (A) The interaction between warmth and competence. High and low levels of warmth and competence correspond to one standard deviation above and below the mean, respectively. Values of trust in AI were the mean residual values after controlling for demographic and individual variables. (B), (C), and (D) The standardized direct and indirect effects of three factors on trust in AI with warmth and competence as the mediators. Δb is the mediation effect of warmth minus that of competence. Path coefficients in each model were significant with ps < 0.001. CI: confidence interval.

Figure 4. The direct and indirect effects of warmth and competence on two behavioral intention measures with trust in artificial intelligence (AI) as the mediator. (A) and (B) Intention to share private data with AI as the intention measure. (C) and (D) Support for using AI in HRM as the intention measure. The path coefficients in each model were significant with ps < 0.001. The results for support for using AI in HRM were based on data from participants who reported being currently employed (N = 1701). CI: confidence interval; HRM: human resources management.

High-Level Expert Group on Artificial Intelligence [AI HLEG]. (2019, April). Ethics guidelines for trustworthy AI [European Commission report]. https://op.europa.eu/en/publication-detail/-/publication/d3988569-0434-11ea-8c1f-01aa 75ed71a1

 Google Scholar

Komiak, S. X., & Benbasat, I. (2004). Understanding customer trust in agent-mediated electronic commerce, web-mediated electronic commerce, and traditional commerce. Information Technology and Management, 5(1/2), 181–207. https://doi.org/10.1023/B:ITEM.0000008081.55563.d4

 Google Scholar

Gefen, D., Karahanna, E., & Straub, D. W. (2003). Trust and TAM in online shopping: An integrated model. MIS Quarterly, 27(1), 51–90. https://doi.org/10.2307/30036519

 Web of Science ®Google Scholar

Gursoy, D., Chi, O. H., Lu, L., & Nunkoo, R. (2019). Consumers acceptance of artificially intelligent (AI) device use in service delivery. International Journal of Information Management, 49, 157–169. https://doi.org/10.1016/j.ijinfomgt.2019.03.008

 Web of Science ®Google Scholar

Delgosha, M. S., & Hajiheydari, N. (2021). How human users engage with consumer robots? A dual model of psychological ownership and trust to explain post-adoption behaviours. Computers in Human Behavior, 117, 106660. https://doi.org/10.1016/j.chb.2020.106660

 Web of Science ®Google Scholar

McKnight, D. H., Cummings, L. L., & Chervany, N. L. (1998). Initial trust formation in new organizational relationships. Academy of Management Review, 23(3), 473–490. https://doi.org/10.5465/amr.1998.926622

 Web of Science ®Google Scholar

Harris-Watson, A. M., Larson, L. E., Lauharatanahirun, N., DeChurch, L. A., & Contractor, N. S. (2023). Social perception in human-AI teams: Warmth and competence predict receptivity to AI teammates. Computers in Human Behavior, 145, 107765. https://doi.org/10.1016/j.chb.2023.107765

 Web of Science ®Google Scholar

Salisbury, W. D., Pearson, R. A., Pearson, A. W., & Miller, D. W. (2001). Perceived security and World Wide Web purchase intention. Industrial Management & Data Systems, 101(4), 165–177. https://doi.org/10.1108/02635570110390071

 Web of Science ®Google Scholar

Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information technology: Toward a unified view. MIS Quarterly, 3, 27, 425–478. https://doi.org/10.2307/30036540

 Web of Science ®Google Scholar

Kim, D. J., Ferrin, D. L., & Rao, H. R. (2008). A trust-based consumer decision-making model in electronic commerce: The role of trust, perceived risk, and their antecedents. Decision Support Systems, 44(2), 544–564. https://doi.org/10.1016/j.dss.2007.07.001

 Web of Science ®Google Scholar

Lockey, S., Gillespie, N., & Curtis, C. (2020). Trust in artificial intelligence: Australian insights. The University of Queensland and KPMG. https://doi.org/10.14264/b32f129

 Google Scholar

Khan, G. F., Swar, B., & Lee, S. K. (2014). Social media risks and benefits: A public sector perspective. Social Science Computer Review, 32(5), 606–627. https://doi.org/10.1177/0894439314524701

 Web of Science ®Google Scholar

Data availability statement

Data pertaining to the analyses and results of this study can be accessed without restrictions at https://osf.io/z4qr2.