Artificial Intelligence lecture slides.pptx

A short insight into AI

1. Artificial Intelligence (AI): Concepts, Applications, and Ethical Considerations COMPILED BY Mr B Maimela

2. Introduction to AI AI refers to machines or applications performing tasks requiring human-like intelligence (Dumont et.al, 2016; Artificial Intelligence Class IX, n.d.).

3. Key Domains: Data Science:   Extracts insights from structured/unstructured data (Artificial Intelligence Class IX, n.d.). Computer Vision:   Trains machines to identify objects like humans (Artificial Intelligence Class IX, n.d.). Natural Language Processing (NLP):   Enables human-computer language interactions (Artificial Intelligence Class IX, n.d.).

4. Advantages and Disadvantages of AI Advantages Reduced errors (Artificial Intelligence Class IX, n.d.). Improved decision-making (Artificial Intelligence Class IX, n.d.). 24/7 operation (Artificial Intelligence Class IX, n.d.).

5. DISADVANTAGES High implementation costs (Artificial Intelligence Class IX, n.d.). Dependence on hardware/software (Artificial Intelligence Class IX, n.d.). Limited to predefined tasks (Artificial Intelligence Class IX, n.d.).

6. AI Project Cycle 1. Problem Scoping  – Define the problem 4. Modelling  – Develop AI algorithms 3. Data Exploration  – Identify patterns 5. Evaluation  – Test model accuracy 6. Deployment  – Implement in real-world scenarios 2. Data Acquisition  – Collect reliable data 

7. AI in Action – Real-World Applications Healthcare:  AI aids in diagnosis and treatment (Smith, 2017). Finance:  Fraud detection and risk management ( Wesi , n.d.). Retail:  Chatbots for customer service (Biswas, n.d.). Autonomous Vehicles:  Self-driving cars (Smith, 2017).

8.   Ethical Considerations in AI Accountability:  Who is responsible for AI decisions? (Artificial Intelligence Class IX, n.d.). Bias:  AI can inherit biases from training data & therefore could make biased decisions (Artificial Intelligence Class IX, n.d.). Privacy:  Data security risks (Biswas, n.d.). Transparency:  Need for explainable AI (Biswas, n.d.).

9. The Future of AI Agentic AI:   Agentic AI represents the next evolution of artificial intelligence, where systems move beyond simple reactive responses to become proactive, autonomous entities capable of: Goal-directed behavior:  Setting and pursuing objectives without constant human oversight e.g., an AI concierge like Hilton's "Connie" that learns guest preferences over time (Smith, 2017, slide 2) Dynamic task decomposition:  Breaking complex goals into sub-tasks e.g., planning a trip by autonomously booking flights, hotels, and activities (Biswas, n.d., slide 17) Self-improvement:  Continuously refining strategies through reinforcement learning ( Wesi , n.d., slide 8) Key Distinction from Traditional AI: While conventional AI follows predefined rules (like IBM's chess-playing Deep Blue), agentic AI exhibits emergent behaviors - as demonstrated by Microsoft's AI "objects" that independently perform tasks (Biswas, n.d., slide 4). 1. Agentic AI: Autonomous, Goal-Driven Systems

10. 2. Human-AI Collaboration: Enhancing Productivity Operational Models: Augmented Intelligence:  AI assists human decision-making e.g., IBM's automated radiologist highlights potential issues for doctor review (Smith, 2017, slide 31) Symbiotic Workflows: Humans provide oversight and ethical judgment AI handles repetitive tasks (Smith, 2017, slide 32) Sector-Specific Examples: Healthcare:  AI analyzes treatment options while physicians make final decisions (Smith, 2017, slide 22) Manufacturing:  Predictive maintenance systems alert human technicians only when needed ( Wesi , n.d., slide 5) Productivity Impact: Contrary to replacement fears, only 5% of European AI startups focus on job displacement - most enhance human capabilities (Dumont et.al, 2016, slide 16).

11. 3. Regulations: Ensuring Ethical AI Development Current Frameworks: EU GDPR (2018): Right to contest automated decisions (Smith, 2017, slide 45) Requires transparency in AI decision-making processes IBM's Ethical AI Principles (Smith, 2017, slides 40-42): Purpose:  AI should aid, not replace humans (symbolic relationship) Transparency:  Disclose training data and control mechanisms Skills:  Train human workers to use AI tools effectively

12. Emerging Challenges: Algorithmic Bias:  "We often have no way of knowing when and why people are biased" (Wachter in Smith, 2017, slide 44) Privacy Risks:  LLMs may leak training data properties (Biswas, n.d., slide 48) Accountability:  Need for explainability in multi-agent systems (Biswas, n.d., slide 49) Implementation Case: Salesforce's " Agentforce " incorporates ethical guardrails for autonomous AI agents in customer service (Biswas, n.d., slide 4).

13.   Conclusion AI is transforming industries but requires ethical oversight. Improper use of AI can have severe consequences. Continuous learning and adaptation are key. In a rapidly evolving world, the question we have to ask ourselves is: How can we ensure AI benefits society responsibly?

14. References: Biswas, D. (n.d.). A comprehensive guide to agentic AI systems [PDF slides]. SlideShare. Retrieved August 18, 2025, from https://www.slideshare.net/slideshow/a-comprehensive-guide-to-agentic-ai-systems-c742/274678426 Dumont, J.-B.; Verdillion , L. & Gossart , E. (2016). The AI rush [PDF slides]. SlideShare. https://www.slideshare.net/slideshow/the-ai-rush/81139551 Smith, C. (2017, October 13). AI and machine learning demystified by Carol Smith at Midwest UX 2017 [PowerPoint slides]. SlideShare. https://www.slideshare.net/slideshow/ai-and-machine-learning-demystified-by-carol-smith-at-midwest-ux-2017/80840514 SweetandSourCandy , S. (n.d.). Artificial intelligence PPT-Class IX [PDF slides]. SlideShare. Retrieved August 18, 2025, from https://www.slideshare.net/slideshow/artificial-intelligence-ppt-class-ix-pdf/271627720 Wesi , M. (n.d.). Agentic AI: The next wave of intelligence [PowerPoint slides]. SlideShare. Retrieved August 18, 2025, from https://www.slideshare.net/slideshow/agentic-ai-the-next-wave-of-intelligence-pptx/275308782

15. IMAGES: Emory University. (n.d.). Ethics-themed visual from Emory Responsible AI promotes the ethical use of Artificial Intelligence [AI-Ethics1/JPEG]. Emory University, Emory Continuing Education: Responsible AI programme article. Retrieved August 18, 2025, from https://ece.emory.edu/articles-news/responsible-ai-program.php Goddard, W. (2020, November 8). Where is AI used today? [where-is-ai-used-1024x683 (1)/JPEG]. IT Chronicles . Retrieved August 18, 2025, from https://itchronicles.com/artificial-intelligence/where-is-ai-used-today/ Shoolini University. (n.d.). [Applications-of-AI/PNG]. In AI applications take the world by storm (blog post). Shoolini University. Retrieved August 18, 2025, from https://shooliniuniversity.com/blog/ai-applications-take-the-world-by-storm/