DBA Dataset The Impact of Artificial Intelligence on the Consumer Behavior of Greeks: Understanding the Awareness of AI Benefits in Their Consumer Journey and How Lack of Knowledge of the Concept of Artificial Intelligence Can Lead to Ignorance, Anxiety, and Fear.

Results Descriptive Statistics Sociodemographic characteristics Table 1 presents the Sociodemographic characteristics of the survey’s participants. The sample was comprised of 318 participants with a total average age of 37.52 (S.D = 13.83) and a range of 14-79 years. 308 (96.9%) of the participants were Greek and 224 (70.4%) were Christian Orthodox. Table 1. Sociodemographic Characteristics of Participants n % Gender Female 157 49.4 Male 153 48.1 Other 8 2.5 I like shopping Clothing - Fashion 103 32.4 Electronics 64 20.1 Food - Refreshments 44 13.8 Nutrition - Health 31 9.7 Decoration 11 3.5 Car Accessories 10 3.1 Other 55 17.3 I shop Occasionally 92 28.9 After research 125 39.3 When sth appears as an offer 20 6.3 Often 81 25.5 Shopping budget A few 64 20.1 Normal 180 56.6 A lot 74 23.3 My online presence is I'm only doing my research and having fun 199 62.6 I simply follow influencers 28 8.8 I purchase things and credit offers 91 28.6 I get influenced by Offers 49 15.4 Personal research 138 43.4 Online reviews 61 19.2 Friends and their reviews 56 17.6 My Tech Savviness is Very low 3 .9 2 15 4.7 3 65 20.4 4 114 35.8 Very high 121 38.1 My Shopping Platform is Shops 47 14.8 Online shops 19 6.0 Both 246 77.4 My Lifestyle is Technology 74 23.3 Art 55 17.3 Travelling 67 21.1 Sports 32 10.1 Nature 35 11.0 Fashion 15 4.7 Religion and Patriotism 17 5.3 Tradition 23 7.2 Place of residence Rural 167 52.5 Urban 151 47.5 In figure can be seen that 135 (73.9%) had at least high tech Savviness. Figure 1. Tech Savviness barchart Principal Component Analysis on the items of Artificial Intelligence Principal components analysis (PCA) was run on the items of Artificial Intelligence in order to define the significant factors arising from them. The assumptions of performing the PCA were met prior to the analysis with Kaiser-Meyer-Olkin value of .90 and a statistically significant Bartlett's test of sphericity was (p < .001). PCA produced five components that had initial eigenvalues total greater or near one explaining 38.32%, 10.81%, 6.32%, 5.78 and 4.87% respectively, of the total variance. Scree plot (Figure 2) also suggested five components as well, according to Cattell (1966). However, component number 4 had low Cronbach a levels (α = .40) therefore it was rejected. Finally, three components were used that explained a total of 55.44% of the total variance. Figure 2. Scree plot suggesting the number of components used A Varimax orthogonal rotation was used for the interpretation of the data. As seen in Table 2, Component 1 was consisted of 6. I’m worried that AI will collect too much of my personal information, 5. I’m afraid that artificial intelligence (AI) will monitor consumer behavior, 7. The forecasting of my preferences by AI makes me feel as my privacy is in danger, 4. I'm worried that AI with forecast and affect purchasing decisions, 8. AI may harm humans to achieve a goal, which gives me anxiety and 9. I worry that the control of AI/ Super AI by a few individuals will introduce great risks to the entire society. Component 2 was consisted of 17. Understanding AI algorithms requires a high level of talent, which is difficult for me, 16. I do not think I would be able to perform well in professional courses in AI, 18. AI technology updates too quickly and is very difficult to learn, 1. How well do you understand the concepts of artificial intelligence? and2. Do you believe that media and government agencies provide sufficient knowledge and information to citizens-consumers in order to feel secure about the topic of artificial intelligence and its capabilities? Finally, Component 3 was consisted of 19. I worry that people cannot figure out how AI makes decisions during when they want to purchase a good or a service, 21. I worry that AI will attain the same level of consciousness as humans, 20. I worry that humans have special feelings (such as love) for super AI and all of these is against in my religion NEW input from the researcher, 14. I am worried that an AI machine may be brighter than me, 15. I’m worried that AI will replace many people’s work and 13. I am worried that AI will replace my work in the future. According to the aforementioned groups of question the final components were named into «Implications, Surveillance, and Ethical Issues», «Concerns about lack of knowledge about AI» and «Consumer and Social Concerns». Table 2. Rotated Structure Matrix for PCA Items Rotated Component Coefficients Comp. 1 Comp. 2 Comp. 3 6. I’m worried that AI will collect too much of my personal information .891 5. I’m afraid that artificial intelligence (AI) will monitor consumer behavior .804 7. The forecasting of my preferences by AI makes me feel as my privacy is in danger .760 4. I'm worried that AI with forecast and affect purchasing decisions .724 8. AI may harm humans to achieve a goal, which gives me anxiety .628 9. I worry that the control of AI/ Super AI by a few individuals will introduce great risks to the entire society. .567 17. Understanding AI algorithms requires a high level of talent, which is difficult for me .916 16. I do not think I would be able to perform well in professional courses in AI .781 18. AI technology updates too quickly and is very difficult to learn .558 1. How well do you understand the concepts of artificial intelligence? -.535 2. Do you believe that media and government agencies provide sufficient knowledge and information to citizens-consumers in order to feel secure about the topic of artificial intelligence and its capabilities? .392 19. I worry that people cannot figure out how AI makes decisions during when they want to purchase a good or a service -.924 21. I worry that AI will attain the same level of consciousness as humans -.870 20. I worry that humans have special feelings (such as love) for super AI and all of these is against in my religion NEW input from the researcher -.585 14. I am worried that an AI machine may be brighter than me -.524 15. I’m worried that AI will replace many people’s work -.442 13. I am worried that AI will replace my work in the future -.403 Scales of AI Mean scores and standard deviations for the scales created by the component analyses are presented in table 3 below. Also, it is noted that participants wish schools/ government/ media to train them more and in a more regular basis around AI in the future (M = 4.25, S.D = .99). Table 3.Means, St.Ds., Min. and Max. scores for the Scales of AI Variable Min. Max. Mean St.D. Implications, Surveillance, and Ethical Issues (α = .90) 6 32 22.19 6.18 4. I'm worried that AI with forecast and affect purchasing decisions 1 5 3.56 1.24 5. I’m afraid that artificial intelligence (AI) will monitor consumer behavior 1 5 3.44 1.37 6. I’m worried that AI will collect too much of my personal information 1 5 3.77 1.26 7. Forecasting of my preferences makes me feel as my privacy is in danger 1 7 3.84 1.38 8. AI may harm humans to achieve a goal, which gives me anxiety 1 5 3.76 1.18 9. I worry that the control of AI/ Super AI by a few individuals will introduce great risks to the entire society. 1 5 3.82 1.15 Concerns about lack of knowledge about AI (α = .56) 11 29 18.95 3.91 1R. How well do you understand the concepts of artificial intelligence? 3 7 4.84 1.23 2R. Do you believe that media and government agencies provide sufficient knowledge and information to citizens-consumers in order to feel secure about the topic of artificial intelligence and its capabilities? 3 7 5.87 1.24 16. I don’t think I would be able to perform well in professional courses in AI 1 5 2.76 1.44 17. Understanding AI algorithms requires a high level of talent, which is difficult for me 1 5 2.44 1.33 18. AI technology updates too quickly and is very difficult to learn 1 5 3.03 1.24 Consumer and Social Concerns (α = .90) 6 30 19.11 6.70 13. I am worried that AI will replace my work in the future 1 5 3.14 1.38 14. I am worried that an AI machine may be brighter than me 1 5 3.01 1.44 15. I’m worried that AI will replace many people’s work 1 5 3.74 1.19 19. I worry that people cannot figure out how AI makes decisions during when they want to purchase a good or a service 1 5 3.00 1.38 20. I worry that humans have special feelings (such as love) for super AI and all of these is against in my religion NEW input from the researcher 1 5 3.25 1.36 21. I worry that AI will attain the same level of consciousness as humans 1 5 2.97 1.46 Normality Testing Normality tests were conducted in order to examine whether the data collected are normally distributed according to the results of the Kolmogorov – Smirnov test. The modalities of «Implications, Surveillance, and Ethical Issues», «Concerns about lack of knowledge about AI» and «Consumer and Social Concerns» were non - normally distributed (p < .001). Inferential Statistics Relation between AI scales and age For the purpose of examining the correlation between the AI’s variables and Age the Pearson Correlation was applied, after confirming all the necessary prerequisites (Table 4). Table 4. Descriptive Statistics and Pearson Correlations Variables n 1 2 3 1. Implications, Surveillance, and Ethical Issues 318 - 2. Concerns about lack of knowledge about AI 318 .39*** - 3. Consumer and Social Concerns 318 66.*** .53*** - 4. Age 318 .03 .09 -.03 ***p<.001. The correlation showed that Age was not related with «Implications, Surveillance, and Ethical Issues» (p = .559) nor «Concerns about lack of knowledge about AI» (p = .110) nor «Consumer and Social Concerns» (p = .672). However, it is noteworthy that «Implications, Surveillance, and Ethical Issues» was related with «Concerns about lack of knowledge about AI» [r(3,314) = .39, p < .001] and «Consumer and Social Concerns» [r(3,314) = .66, p < .001] and that «Concerns about lack of knowledge about AI» was related with «Consumer and Social Concerns» [r(3,314) = .53, p < .001]. Relation between AI scales and Location An Independent samples T- test was performed, in order to examine the hypothesis that there are differences between those living in urban and those living in rural areas according to the AI scales’ scores (Table 5). Table 5. Independent samples t-test scores and values of descriptive statistics for the effect of Location on AI scales Location Measure Rural Urban 95% CI for the mean difference t df M SD M SD Implications, Surveillance, and Ethical Issues 22.59 6.170 21.74 6.19 -.51 2.22 1.24 316 Concerns about lack of knowledge about AI 19.33 4.143 18.52 3.61 -.05 1.66 1.85 315.59 Consumer and Social Concerns 19.92 6.774 18.22 6.53 .23 3.17 2.27* 316 *p < .05. The results of the t-test independent samples show that the Consumer and Social Concerns differs between people living in Rural and Urban areas [t(316) = 2.27, p =.024]. Particularly, people in rural areas have higher consumer and social concerns regarding AI compared to people in urban areas (Figure 3). Finally, there were no differences between Urban and Rural areas in Implications, Surveillance, and Ethical Issues (p = .217) and Concerns about lack of knowledge about AI (p = .065). Figure 3. Confidence Interval Error bar for Consumer and Social Concerns according to Location Relation between shopping preferences and AI scales Three one – way independent samples analysis of variance were performed in order to evaluate the differences between the AI scales as regards the shopping preferences (Table 6). Table 6. One-way ANOVA for the AI scales as regards the shopping preferences Implications, Surveillance, and Ethical Issues Concerns about lack of knowledge about AI Consumer and Social Concerns M SD M SD M SD Shopping Preferences Clothing - Fashion 21.88 6.02 18.63 3.80 18.75 6.23 Electronics 19.56 6.68 17.64 3.36 16.86 6.46 Food - Refreshments 22.41 5.73 19.34 4.36 19.64 7.09 Nutrition - Health 22.84 5.13 19.00 3.70 19.48 5.73 Decoration 25.82 7.07 22.36 5.90 24.18 7.78 Car Accessories 26.40 3.20 21.90 3.14 24.60 3.78 The results show that there are statistically significant differences between the several Shopping Preferences as regards the Implications, Surveillance, and Ethical Issues [F(6,311) = 4.28, p < .001], the Concerns about lack of knowledge about AI [F(6,311) = 4.13, p < .001] and Consumer and Social Concerns [F(6,311) = 3.76, p < .001]. The post-hoc comparison, using Tukey method, showed that those that preferred Electronics had lower Implications, Surveillance, and Ethical Issues compared to those that preferred Decoration (p = .026) and Car Accessories (p = .016). Similarly, those that preferred Electronics had lower Concerns about lack of knowledge about AI compared to those that preferred Decoration (p = .003) and Car Accessories (p = .019) and, again, those that preferred Electronics had lower Consumer and Social Concerns compared to those that preferred Decoration (p = .012) and Car Accessories (p = .010) (Figures 4,5 and 6). Figures 4, 5 and 6. Mean of AI scales according to shopping preferences Relation between shopping platforms and AI scales Three one – way independent samples analysis of variance were performed in order to evaluate the differences between the AI scales as regards the shopping platforms (Table 7). Table 7. One-way ANOVA for the AI scales as regards the shopping platforms Implications, Surveillance, and Ethical Issues Concerns about lack of knowledge about AI Consumer and Social Concerns M SD M SD M SD Shopping Platforms Shops 24.77 5.14 20.40 4.24 21.32 6.52 Online shops 20.05 6.08 17.63 4.04 19.11 6.59 Both 21.92 6.29 18.81 3.80 18.69 6.72 The results show that there are statistically significant differences between the several Shopping Platforms as regards the Implications, Surveillance, and Ethical Issues [F(6,311) = 5.56, p = .004], the Concerns about lack of knowledge about AI [F(6,311) = 4.53, p = .011] and Consumer and Social Concerns [F(6,311) = 3.07, p = .048]. The post-hoc comparison, using Tukey method, showed that those that preferred Shops had higher Implications, Surveillance, and Ethical Issues compared to those that preferred Online Shops (p = .014) or Both (p = .010). Similarly, those that preferred Shops had higher Concerns about lack of knowledge about AI compared to those that preferred Online Shops (p = .025) or Both (p = .028). Finally, those that preferred Shops had higher Consumer and Social Concerns compared to those that preferred Both (p = .037) (Figure 7, 8 and 9). Figures 7, 8 and 9. Mean of AI scales according to shopping platforms Relation between religion and AI scales Three one – way independent samples analysis of variance were performed in order to evaluate the differences between the AI scales as regards the religion (Table 8). Table 8. One-way ANOVA for the AI scales as regards the religion Implications, Surveillance, and Ethical Issues Concerns about lack of knowledge about AI Consumer and Social Concerns M SD M SD M SD Religion Christian Orthodox 22.58 6.22 18.86 3.97 19.71 6.59 Non religious 21.54 5.76 18.94 3.77 17.43 6.34 Other 20.55 6.64 19.62 3.86 18.28 7.78 The results show that there are statistically significant differences between the several Religion as regards the Consumer and Social Concerns [F(2,315) = 3.19, p = .042], whereas there are no differences in Implications, Surveillance, and Ethical Issues (p = .160) and the Concerns about lack of knowledge about (p = .618). The post-hoc comparison, using Tukey method, showed that Orthodox Christians showed higher Consumer and Social Concerns regarding AI compared to non-religious people (p = .042) (Figure 10). Figure 10. Mean of Consumer and Social Concerns in AI according to religion Bibliography Cattell, R. B. (1966). The scree test for the number of factors. Multivariate Behavioral Research, 1, 245-276.