presentation_adoption_de_IA_automobile.pptx

adoption de l'IA dans l'industrie automobile

1. University Ibn Tofail, ENSA Kenitra Adoption of IA in the automotive industry in Morocco Prepared by : BENKHADA Aya, ELMANSER Aicha Supervised by : Mr. Anis MOUMEN January 2, 2024

2. 1 Content Introduction Systematic Literature Review Objective Methodology Result Qualitative approach objective Methodology Result Quantitative approach objective Methodology Result Conclusion

3. 2 Introduction The adoption of artificial intelligence (AI) in the automotive industry represents a major global revolution. Our study, focused on this transformation, began with a Systematic literature review to identify global trends. We then sharpened our focus on Morocco, conducting qualitative and quantitative research with current and future industry professionals. This comprehensive approach offers an in-depth understanding of the impact of AI in the automotive sector, combining a global perspective with insights specific to the Moroccan context.

4. 3 Systematic Literature Review objective The objective of this Systematic Literature Review (SLR) is to take an in-depth examination of existing research on the adoption of artificial intelligence (AI) in the automotive industry, in order to synthesize knowledge and highlight key trends in this context.

5. 4 Systematic Literature Review Methodology We used a rigorous process aligned with PRISMA principles to develop our methodology. We searched ScienceDirect, Scopus, and IEEE databases for relevant articles. Initially, we found 65 promising articles, but after eliminating duplicates, we evaluated 60 articles in-depth. We ultimately selected 39 outstanding contributions that were relevant to the subject and minimized irrelevance. Figure: inclusion and exclusion criteria

6. 5 Systematic Literature Review Result VOSviewer Analysis of Authors Citations Three authors, Bajit, A., Barodi, A., Benbrahim, M., and Tamtaoui, A., stand out with significant contributions, collaborating on at least two papers each. These authors play a central role in the network of co-authors, demonstrating a strong influence in exploring topics related to the adoption of artificial intelligence (AI) in the automotive industry. Figure: Co-authorship

7. 6 Systematic Literature Review Result VOSviewer Analysis of Keywords Occurrence Many keywords present remarkable strengths in terms of "total link strength", underlining their importance and their robust links with other keywords. Some keywords include "Artificial Intelligence", "Automotive Industry", "Decision Making", "Deep Learning", "Industry 4.0", "Intelligent Systems", "Machine Learning", "Neural Networks", "Planning", "Supply Chains", and "Sustainable Development".

8. 7 Systematic Literature Review Result Descriptive Analysis by Year The descriptive analysis by year reveals consistent evolution between 2000 and 2017, followed by an increasing evolution starting from 2019 in publications related to the adoption of AI by the automotive industry. This evolution highlights a growing interest over the years Figure: Descriptive Analysis by Year

9. 8 Systematic Literature Review Result analysis grid Highlight the increasing adoption of artificial intelligence (AI) in the automotive industry. Key trends include the use of AI to enhance production efficiency, optimize vehicle design, and improve road safety. AI is also employed to balance the workload between robots and human operators, monitor vehicle performance in real- time, anticipate breakdowns, and develop advanced driver assistance systems. However, the adoption of AI requires adaptation to new technologies and addressing challenges such as regulation and data management.

10. 9 Systematic Literature Review Result Essential Aspects Emerging from Our Study Production Process Transformation: AI integration transforming the planning and execution of production workflows, marking intelligent management through Industry 4.0. Predictive Maintenance for Reliability: Machine learning powers predictive maintenance, ensuring vehicle safety while optimizing costs. Automated Driving Support: AI has a crucial role in automated driving safety, exploring standards and the balance between regulation and innovation.

11. 10 Systematic Literature Review Result Essential Aspects Emerging from Our Study Electric Vehicles and Artificial Intelligence: Explore recent advancements in electric vehicles and their infrastructure, highlighting the impact of artificial intelligence, while also underscoring existing challenges. Safety and Technological Challenges: Major challenges, including communications security, underscore the imperative of robust cybersecurity in this complex technological landscape.

12. 11 Qualitative approach objective The principal objective of this qualitatif study is to explore in depth the level of adoption of artificial intelligence (AI) by the automotive industry in Morocco. The study aims to understand the perspectives of companies in the sector and the challenges faced in this process.

13. 12 Qualitative approach Methodology Sample description: The study was conducted within five companies(n=5) in the automotive sector, selected to represent various aspects of the industry. Data collection: Remote semi- structured interviews were conducted with qualified interviewees, such as engineers or technicians, chosen for their relevant expertise.

14. 13 Qualitative approach Methodology Interview guide: An interview guide was drawn up, organized around six themes, including: the definition of artificial intelligence: the presence of AI- related Technologies: the forms of technological practice: the challenges facing the automotive sector in Morocco: the tools for adapting to technological advances: the existence of collaborations to adapt with technological revolution:

15. 14 Qualitative approach Methodology Data collection and Analysis Interviews were recorded in video format and transcribed verbatim, with reformulation, for a qualitative data analysis using NVIVO12 software. The principal steps of data analysis are as follows : Figure: Steps in Qualitative Data Analysis

16. 15 Qualitative approach Result Saturation level: A similarity analysis between the interviews was carried out using Pearson’s correlation coefficient, measuring the linear correlation between them. The results showed high coefficients, notably 0.638598 between "Interview with M5" and "Interview with M1", indicating strong similarity between these two particular interviews. Figure: elements grouped by word similarity

17. 16 Qualitative approach Result The qualitative study conducted among five companies in the automotive sector in Morocco revealed the following results: Definition of AI: Diverse perspectives, from the use of advanced technologies (M1) to a broader definition encompassing machine learning, natural language processing and complex problem solving (M4). Existence of AI- related technologies: Integration in specific sectors such as design, manufacturing and data management (M1 and M2). Specific examples of AI-based data analysis, including computer vision (M4).

18. 17 Qualitative approach Result Forms of Practice for AI Integration: Varied applications of AI, from autonomous driving to data analysis, ADAS(advanced driver assistance system) and predictive maintenance (M1, M2, M5).Opportunities for task automation and production management thanks to AI (M3). ► Challenges of AI Integration in the Moroccan Automotive Sector: Such as robust infrastructure, scarcity of skilled human resources, cybersecurity and social resistance (M1, M5). Emphasis on human challenges, such as resistance to change and the need for awareness of new technologies (M4).

19. 18 Qualitative approach Result Tools for Adapting to AI- related Technological Change: Use of tools such as JIRA based on AI algorithms for resource management (M2). Strategies such as constant technological monitoring and continuous training programs to stay competitive (M5). Collaborations to Adapt to Technological Advances in AI: Importance stressed of collaborations to innovate and adapt to technological advances, with concrete examples of strategic partnerships (M2).

20. 19 Qualitative approach Result Introducing the Technology Acceptance Model (TAM) provides insights into AI adoption dynamics, where perceived ease of use and usefulness influence the intention to use AI. Variables like improved security, training, user experience, and social influence further shape this adoption context. Figure: Contextualized Conceptual Model

21. 20 Quantitative approach objective The objective of our quantitative study is to analyze the factors influencing the acceptance of artificial intelligence (AI) in the automotive industry in Morocco, focusing on the overall perception (usefulness and ease) of AI and its link with usage intention among future industrialists.

22. 21 Quantitative approach Methodology Population studied: The sample for this quantitative study was made up of students at ENSA Kenitra, mainly in the Industrial Engineering and Intelligent Industry and Digital Technologies. Sampling techniques The sample was randomly selected from among students in these branch, ensuring a diverse representation of opinions and levels of experience. Population size: 69 Confidence level:80 Margin of error:5 Sample size:49

23. 22 Quantitative approach Methodology Data Collection Instrument: A questionnaire was developed to collect quantitative data. The questionnaire included structured questions based on the TAM model aimed at measuring students’ perceptions of the adoption of artificial intelligence in the automotive industry in Morocco. Figure: Contextualized Conceptual Model

24. 23 Quantitative approach Methodology Variables studied: included perceived usefulness of AI, perceived ease of use of AI, intention to use AI Figure: items from PEU Figure: items from PU

25. 24 Quantitative approach Result Research hypothesis: positive overall perception (usefulness and ease) of AI is associated with higher usage intention among future industrialists.

26. 25 Quantitative approach Result Results of the dimensions studied: PEU1: Completely Agree (1), Agree (2), Neutral (3), Disagree (4), Completely Disagree (5) Figure: distribution of responses for PEU

27. 26 Quantitative approach Result Results of the dimensions studied: PU1: Completely Agree (1), Agree (2), Neutral (3), Disagree (4), Completely Disagree (5) Figure: distribution of responses for PU

28. 27 Quantitative approach Result Results of the dimensions studied: IU1: Completely Agree (1), Agree (2), Neutral (3), Disagree (4), Completely Disagree (5) Figure: distribution of responses for IU

29. 28 Qualitative approach Result Statistical models: Linear regression of PEU as a function of PU : PEU = 0.8771 * PU + 5.7174 . The relationship is significant, shows that each unit increase in PU is associated with a 0.8771 increase in PEU. Figure: Linear regression of PEU as a function of PU

30. 29 Quantitative approach Result Statistical models: Linear regression of IU as a function of PEU : IU = 0.12041 * PEU + 1.71297 The relationship is significant, shows that each unit increase in PEU is associated with a 0.12041 increase in IU. Figure: Linear regression of IU as a function of PU

31. 30 Quantitative approach Result Statistical models: Linear regression of IU as a function of PU : IU = 0.21274 * PU + 1.66517 The relationship is significant, shows that each unit increase in PU is associated with a 0.21274 increase in IU. Figure: Linear regression of IU as a function of PEU

32. 31 Quantitative approach Result Normality test for models of PEU and PU The residuals follow a normal distribution, as the p- value (0.1015) is above the 0.05 significance level. Thus, there is insufficient evidence to reject the normality hypothesis for the residuals. Normality test for models of IU and PU The residuals do not follow a normal distribution, as the p-value (0.003302) is below the 0.05 significance level. There is significant evidence of non-normality in the residuals, which may indicate limitations in the predictive ability of this model.

33. 32 Quantitative approach Result Normality test for models of IU and PEU Residuals do not follow a normal distribution, as the p-value (0.02524) is below the 0.05 significance level. There is significant evidence of non-normality in the residuals, suggesting limitations in the predictive ability of this model.

34. 33 Conclusion This project examines the integration of artificial intelligence (AI) in the automotive industry in Morocco. The study consists of a systematic literature review and a qualitative approach involving industry experts. The research identifies key trends, such as autonomous driving and predictive maintenance, as well as highlighting various AI applications. It also sheds light on challenges such as cybersecurity and resistance to change. The findings from the quantitative approach emphasize that the overall perception of AI, in terms of its usefulness and ease of use, influences the intention to use it. Ultimately, this study emphasizes the significant impact of AI in the Moroccan automotive sector and highlights the need for a strategic approach to address the emerging challenges.

35. Thank you for your attention