DOI: https://doi.org/10.1038/s41598-024-53335-2
PMID: https://pubmed.ncbi.nlm.nih.gov/38316898
تاريخ النشر: 2024-02-05
المؤلف: Jan‐Philipp Stein وآخرون
الموضوع الرئيسي: المعلومات المضللة وتأثيراتها
نظرة عامة
تقدم ورقة البحث فحصًا شاملاً للمواقف العامة تجاه الذكاء الاصطناعي (AI)، مع معالجة الفجوات الكبيرة في الأدبيات الحالية. يقدم المؤلفون ATTARI-12، وهو استبيان جديد مصمم لقياس المواقف تجاه الذكاء الاصطناعي ككيان موحد، مستقل عن التطبيقات المحددة. يظهر هذا المقياس موثوقية وصلاحية جيدة عبر دراستين شملتا 640 مشاركًا. في دراسة لاحقة مع 298 مشاركًا، يستكشف المؤلفون العلاقة بين مواقف الذكاء الاصطناعي وسمات الشخصية المختلفة، بما في ذلك الخمسة الكبار، والثالوث المظلم، وعقلية المؤامرة. تشير النتائج إلى أن ارتفاع القابلية للتوافق والعمر الأصغر يرتبطان بمواقف أكثر إيجابية تجاه الذكاء الاصطناعي، بينما يرتبط الميل نحو معتقدات المؤامرة بتصورات سلبية.
تؤكد الورقة على أهمية فهم هذه المواقف، خاصة مع تزايد دمج تقنيات الذكاء الاصطناعي في الحياة اليومية. يجادل المؤلفون بأن المواقف تجاه الذكاء الاصطناعي تتشكل ليس فقط من خلال الميزات المحددة للتكنولوجيا ولكن أيضًا من خلال عوامل نفسية أوسع. يبرزون الحاجة إلى الشفافية في تطوير الذكاء الاصطناعي للتخفيف من التصورات السلبية، خاصة بين الأفراد المعرضين للتفكير المؤامراتي. يدعو المؤلفون إلى مزيد من البحث لتوضيح فهم مواقف الذكاء الاصطناعي، مقترحين تمييزًا محتملاً بين تصورات قدرات الذكاء الاصطناعي المتعلقة بالوكالة مقابل الخبرة، مما قد يوفر رؤى أعمق حول تفاعلات المستخدمين مع تقنيات الذكاء الاصطناعي.
النتائج
ركزت نتائج الدراسة على الصلاحية العاملية لمقياس ATTARI-12، الذي يقيس المواقف تجاه الذكاء الاصطناعي. تم إجراء تحليل عامل تأكيدي، بدءًا من نموذج عامل واحد يفترض وجود بنية موقف عامة. وُجد أن هذا النموذج له ملاءمة أقل مقارنة بنموذج ثنائي العامل S-1، الذي شمل عاملين محددين متعامدين للعناصر المعرفية والعاطفية. من الجدير بالذكر أن النموذج الثنائي العامل الذي أخذ في الاعتبار العناصر ذات الصياغة السلبية أظهر ملاءمة قوية مع مؤشر الملاءمة المقارن (CFI) قدره 0.98، وخطأ الجذر المتوسط المربع (RMSEA) قدره 0.03، وخطأ الجذر المتوسط المعياري (SRMR) قدره 0.03. كشفت التحليلات أنه بينما تم تفسير 79% من التباين بواسطة العامل العام، كان عامل الصياغة المحددة مسؤولًا عن 21%. كانت الاتساق الداخلي للمقياس ممتازة، مع قيمة ألفا كرونباخ 0.93، وكان توزيع الدرجات تقريبًا طبيعيًا.
أشارت التحليلات الإضافية إلى أن درجات ATTARI-12 كانت مرتبطة إيجابيًا بالمواقف تجاه المساعدين الشخصيين الإلكترونيين (r = 0.60، p < 0.001) والروبوتات (r = 0.68، p < 0.001)، بينما لم تظهر أي علاقة ذات دلالة مع انحياز الرغبة الاجتماعية (r = 0.04، p = 0.411). كما أظهرت النسخة الألمانية من المقياس خصائص نفسية جيدة، مع قيم ألفا كرونباخ قدرها 0.91 و0.89 عبر نقطتين زمنيتين. كشفت تحليلات الانحدار الخطي الهرمي أنه بينما لم تتنبأ العمر والجنس بشكل كبير بدرجات ATTARI، فإن تضمين سمات الشخصية الخمسة الكبار وعقلية المؤامرة في الخطوات اللاحقة حسّن من قوة تفسير النموذج، حيث أوضح النموذج النهائي 14% من التباين في المواقف تجاه الذكاء الاصطناعي.
المناقشة
في هذا القسم، يناقش المؤلفون الهدف البحثي الثاني، الذي يركز على الروابط بين المواقف المتعلقة بالذكاء الاصطناعي وسمات الشخصية. درست الدراسات السابقة بشكل أساسي العوامل الديموغرافية والاجتماعية الثقافية التي تؤثر على قبول الجمهور للذكاء الاصطناعي، كاشفة أن النساء وكبار السن والأقليات العرقية والأفراد ذوي المستويات التعليمية المنخفضة يميلون إلى التعبير عن مزيد من المخاوف بشأن الذكاء الاصطناعي. بالإضافة إلى ذلك، تم تحديد عوامل مثل القيم المتساوية، وعدم الثقة في العلم، والتعرض للسرديات الديستوبية كمتنبئات بالمواقف السلبية تجاه الذكاء الاصطناعي. ومع ذلك، لا يزال تأثير سمات الشخصية على المواقف المتعلقة بالذكاء الاصطناعي غير مستكشف بشكل كافٍ، مما دفع المؤلفين إلى التحقيق في هذا المجال باستخدام أطر عمل معروفة مثل الخمسة الكبار والثالوث المظلم للشخصية، بالإضافة إلى البناء الناشئ لعقلية المؤامرة.
يحدد المؤلفون منهجيتهم من خلال ثلاث دراسات. تركز الدراسات 1 و2 على تطوير والتحقق من صحة مقياس ATTARI-12 لقياس المواقف تجاه الذكاء الاصطناعي، وتقييم موثوقيته وصلاحيته، واستكشاف ارتباطه بالمواقف تجاه تطبيقات الذكاء الاصطناعي المحددة. تهدف الدراسة 3 إلى ربط مواقف المشاركين تجاه الذكاء الاصطناعي بسماتهم الشخصية والعوامل الديموغرافية، مفترضة أن سمات مثل الانفتاح على التجربة، والضمير، والانبساط، والقابلية للتوافق، والعصابية ستؤثر بشكل كبير على المواقف تجاه الذكاء الاصطناعي. يقترح المؤلفون أيضًا فرضيات بشأن سمات الثالوث المظلم وعقلية المؤامرة، متوقعين أن المستويات الأعلى من الميكافيلية وعقلية المؤامرة ستتوافق مع مواقف أكثر سلبية تجاه الذكاء الاصطناعي، بينما قد ترتبط السيكوباتية والنرجسية بشكل إيجابي. علاوة على ذلك، يتوقعون تكرار النتائج السابقة التي تشير إلى أن النساء وكبار السن عمومًا يحملون مواقف أكثر سلبية تجاه الذكاء الاصطناعي.
القيود
تسلط القيود في هذا البحث الضوء على عدة عوامل حاسمة يجب أخذها في الاعتبار عند تفسير النتائج، خاصة من الدراسة الثالثة التي تتناول الفرضيات الرئيسية. على الرغم من أن العينة كانت متنوعة نسبيًا من حيث العمر والجنس والوضع الاجتماعي الاقتصادي، إلا أنها استُخلصت من مجموعة واحدة من المشاركين، مما يثير مخاوف بشأن قابلية تعميم النتائج. يزيد احتمال تداخل المشاركين بين الدراسات، على الرغم من المجموعة الكبيرة التي تم تجنيدهم منها، من تعقيد صلاحية النتائج. بالإضافة إلى ذلك، تشير الغالبية الملحوظة من الذكور في العينة إلى الحاجة إلى تمثيل أكثر توازنًا للجنس في الأبحاث المستقبلية، فضلاً عن استكشاف أوسع للتأثيرات الاجتماعية الاقتصادية والتعليمية على المواقف تجاه الذكاء الاصطناعي.
يجب أن تهدف الدراسات المستقبلية إلى تكرار هذه النتائج عبر سياقات ثقافية ومجموعات ديموغرافية مختلفة لتعزيز قوة الاستنتاجات. يوصي المؤلفون باستخدام مقياس ATTARI-12 في مجموعات سكانية متنوعة للتحقيق في كيفية تأثير سمات الشخصية، مثل تلك التي تلتقطها نموذج HEXACO، على المواقف تجاه الذكاء الاصطناعي. علاوة على ذلك، قد يوفر دمج العوامل السياقية والدافعية – مثل التعرض للخيال العلمي والتفاعلات السابقة مع الذكاء الاصطناعي – رؤى أعمق حول الديناميات التي تشكل تصورات الأفراد للتكنولوجيا. بشكل عام، بينما يظهر مقياس ATTARI-12 وعدًا كأداة قياس موثوقة، فإن مزيدًا من التحقق والاستكشاف لأبعاد الشخصية الإضافية والتأثيرات السياقية أمر ضروري لفهم شامل للمواقف تجاه الذكاء الاصطناعي.
DOI: https://doi.org/10.1038/s41598-024-53335-2
PMID: https://pubmed.ncbi.nlm.nih.gov/38316898
Publication Date: 2024-02-05
Author(s): Jan‐Philipp Stein et al.
Primary Topic: Misinformation and Its Impacts
Overview
The research paper presents a comprehensive examination of public attitudes towards artificial intelligence (AI), addressing significant gaps in the existing literature. The authors introduce the ATTARI-12, a novel questionnaire designed to measure attitudes towards AI as a unified construct, independent of specific applications. This measure demonstrates good reliability and validity across two studies involving 640 participants. In a subsequent study with 298 participants, the authors explore the relationship between AI attitudes and various personality traits, including the Big Five, the Dark Triad, and conspiracy mentality. Findings indicate that higher agreeableness and younger age correlate with more positive attitudes towards AI, while a tendency towards conspiracy beliefs is associated with negative perceptions.
The paper emphasizes the importance of understanding these attitudes, particularly as AI technologies become increasingly integrated into daily life. The authors argue that attitudes towards AI are shaped not only by specific features of the technology but also by broader psychological factors. They highlight the need for transparency in AI development to mitigate negative perceptions, especially among individuals prone to conspiracy thinking. The authors advocate for further research to refine the understanding of AI attitudes, suggesting a potential distinction between perceptions of AI capabilities related to agency versus experience, which could yield deeper insights into user interactions with AI technologies.
Results
The results of the study focused on the factorial validity of the ATTARI-12 scale, which measures attitudes towards AI. A confirmatory factor analysis was conducted, beginning with a single-factor model that assumed a general attitude construct. This model was found to have an inferior fit compared to a bifactor S-1 model, which included two orthogonal specific factors for cognitive and affective items. Notably, the bifactor model that accounted for negatively worded items demonstrated a strong fit with a comparative fit index (CFI) of 0.98, root mean squared error (RMSEA) of 0.03, and standardized root mean residual (SRMR) of 0.03. The analysis revealed that while 79% of the variance was explained by the general factor, the specific wording factor accounted for 21%. The internal consistency of the scale was excellent, with a Cronbach’s α of 0.93, and the distribution of scores was approximately normal.
Further analysis indicated that the ATTARI-12 scores were positively correlated with attitudes towards electronic personal assistants (r = 0.60, p < 0.001) and robots (r = 0.68, p < 0.001), while showing no significant relationship with social desirability bias (r = 0.04, p = 0.411). The German-language version of the scale also exhibited good psychometric properties, with Cronbach's α values of 0.91 and 0.89 across two time points. Hierarchical linear regression analyses revealed that while age and gender did not significantly predict ATTARI scores, the inclusion of the Big Five personality traits and conspiracy mentality in subsequent steps improved the model's explanatory power, with the final model explaining 14% of the variance in attitudes towards AI.
Discussion
In this section, the authors discuss the second research objective, which focuses on the associations between AI-related attitudes and personality traits. Previous studies have primarily examined demographic and sociocultural factors influencing public acceptance of AI, revealing that women, the elderly, ethnic minorities, and individuals with lower education levels tend to express more concerns about AI. Additionally, factors such as egalitarian values, distrust in science, and exposure to dystopian narratives have been identified as predictors of negative attitudes toward AI. However, the impact of personality traits on AI-related attitudes remains underexplored, prompting the authors to investigate this area using established frameworks like the Big Five and Dark Triad of personality, as well as the emerging construct of conspiracy mentality.
The authors outline their methodology through three studies. Studies 1 and 2 focus on developing and validating the ATTARI-12 scale for measuring attitudes toward AI, assessing its reliability and validity, and exploring its correlation with attitudes toward specific AI applications. Study 3 aims to connect participants’ attitudes toward AI with their personality traits and demographic factors, hypothesizing that traits such as openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism will significantly influence attitudes toward AI. The authors also propose hypotheses regarding the Dark Triad traits and conspiracy mentality, predicting that higher levels of Machiavellianism and conspiracy mentality will correlate with more negative attitudes toward AI, while psychopathy and narcissism may relate positively. Furthermore, they anticipate replicating previous findings that women and older individuals generally hold more negative attitudes toward AI.
Limitations
The limitations of this research highlight several critical factors that should be considered when interpreting the findings, particularly from the third study addressing the main hypotheses. Although the sample was relatively diverse in age, gender, and socioeconomic status, it was drawn from a single participant group, which raises concerns about the generalizability of the results. The potential for participant overlap between studies, despite the large pool from which they were recruited, further complicates the validity of the findings. Additionally, the notable male majority in the sample suggests a need for more balanced gender representation in future research, as well as a broader exploration of socio-economic and educational influences on attitudes toward AI.
Future studies should aim to replicate these findings across different cultural contexts and demographic groups to enhance the robustness of the conclusions. The authors recommend utilizing the ATTARI-12 scale in diverse populations to investigate how personality traits, such as those captured by the HEXACO model, influence attitudes toward AI. Furthermore, incorporating situational and motivational factors—such as exposure to science fiction and prior interactions with AI—could provide deeper insights into the dynamics shaping individuals’ perceptions of technology. Overall, while the ATTARI-12 scale shows promise as a reliable measurement tool, further validation and exploration of additional personality dimensions and contextual influences are essential for a comprehensive understanding of attitudes toward AI.
