Module 4
TRENDS IN AI
AI and Ethical Concerns
It addresses the moral, social, and practical challenges arising from the widespread adoption of AI.
Ethical Use of AI – Ensuring AI is designed and used in ways that are fair, transparent, and aligned with
human values.
AI and Bias – How AI models can inherit and amplify biases present in training data (e.g., gender,
racial, or socioeconomic bias).
AI Safety and Risks – Concerns about autonomous systems, job displacement, privacy invasion, and
misuse of AI in surveillance or warfare.
Trustworthy AI – Principles for building AI that is explainable, accountable, and auditable.
Impact on Jobs – How automation and AI might change the workforce, including both displacement
and the creation of new roles.
Ethical Frameworks – Guidelines and regulations (e.g., from the EU, IEEE) for responsible AI
development.
AI as a Service (AIaaS)
the cloud-based delivery of AI tools and services, enabling businesses to leverage AI without building
their own infrastructure.
Definition and Model – AIaaS provides pre-built AI capabilities (like NLP, computer vision, predictive
analytics) via APIs or platforms (e.g., Google AI, AWS SageMaker, Azure AI).
Growth Drivers – Factors such as lower costs, scalability, ease of integration, and faster deployment.
Use Cases – Chatbots, recommendation engines, fraud detection, automated data analysis.
Challenges – Data security, vendor lock-in, limited customization, and latency in real-time applications.
Vendors – Major players include Google Cloud AI, Microsoft Azure AI, IBM Watson, and Amazon AI
services.
AIaaS democratizes AI, allowing small and medium businesses to adopt AI without deep technical expertise
or large investments.
Recent Trends in AI
Highlights cutting-edge developments and emerging directions in AI technology and application.
Collaborative Systems (Human-AI Collaboration) – AI systems designed to work alongside humans
(e.g., cobots in manufacturing, AI-assisted diagnostics in healthcare).
Algorithmic Game Theory & Computational Social Choice – Using AI to model strategic
interactions in economics, politics, and multi-agent environments.
Multi-Agent Reinforcement Learning (MARL) – Multiple AI agents learning to cooperate or compete
in shared environments (e.g., autonomous vehicles, robotic teams).
Neuromorphic Computing – Hardware inspired by the human brain, designed to process information
more efficiently and with lower power consumption (e.g., Intel’s Loihi chip).
AI in Edge Computing – Running AI algorithms on local devices (like smartphones, IoT sensors) rather
than in the cloud for faster, more private processing.
Generative AI – Models like GPT (ChatGPT), DALL-E, and other creative AI tools for text, image, and
music generation (covered in later chapters and examples).
These trends show AI evolving toward more interactive, efficient, and integrated systems, moving
beyond standalone algorithms to embedded, collaborative intelligence.
Expert System
An expert system is an AI system that mimics the decision-making ability of a human expert in a specific
domain.
Components of an Expert System:
1. Knowledge Base – Stores facts and rules about the domain.
2. Inference Engine – Applies rules to facts to deduce new information.
3. User Interface – Allows non-experts to interact with the system.
4. Explanation Facility – Explains how the system reached a conclusion.
5. Knowledge Acquisition Facility – Helps experts input knowledge.
Characteristics:
o Domain-specific
o Rule-based or logic-based
o Capable of explaining decisions
o Can handle uncertainty (in some systems)
Examples Given:
o Medical diagnosis systems
o Credit approval systems
o Fault diagnosis in machinery
Advantages:
Consistent decision-making
Preserves expert knowledge
Available 24/7
Can train new experts
Limitations:
Limited to narrow domains
Cannot learn from new data (unless combined with ML)
Expensive to build and maintain
Cannot handle "common sense" reasoning
Development Process:
Knowledge acquisition
Knowledge representation
System development
Testing and validation
Internet of Things (IoT)
The Internet of Things refers to interconnected devices that collect and exchange data over the internet
without human intervention.
Basic Components:
1. Sensors – Collect data from the environment
2. Connectivity – Wi-Fi, Bluetooth, 5G, LoRaWAN
3. Data Processing – Edge or cloud processing
4. User Interface – Apps, dashboards for control
Applications/Examples:
o Smart homes (thermostats, lights, security)
o Wearable health devices
o Smart agriculture (soil sensors, automated irrigation)
o Industrial IoT (predictive maintenance)
o Smart cities (traffic management, waste management)
Challenges Discussed:
1. Security – Vulnerable to hacking
2. Privacy – Massive data collection
3. Interoperability – Different standards and protocols
4. Scalability – Managing millions of devices
5. Power Consumption – Battery limitations
IoT Products Mentioned:
o Smart thermostats (Nest)
o Fitness trackers (Fitbit)
o Connected cars
o Industrial sensors
How IoT Works:
o Sensors collect data
o Data transmitted to gateway
o Data processed in cloud/edge
o Actions taken based on analysis
Artificial Intelligence of Things (AIoT)
Artificial Intelligence of Things is the integration of AI technologies with IoT infrastructure to create
smarter, autonomous systems.
How AIoT Works:
1. Data Collection – IoT devices gather real-time data
2. AI Processing – Machine learning models analyze the data
3. Intelligent Action – Systems make decisions or predictions
4. Feedback Loop – Systems learn and improve over time
Where AI Unlocks IoT Value:
o Predictive Analytics – Predicting failures before they happen
o Automated Decision Making – Real-time responses without human intervention
o Pattern Recognition – Identifying trends in massive data streams
o Natural Language Processing – Voice-controlled IoT devices
Applications/Examples:
1. Smart Cities – AIoT for traffic optimization, energy management
2. Healthcare – Wearables that predict health issues
3. Manufacturing – Predictive maintenance, quality control
4. Agriculture – AI-driven irrigation and pest control
5. Retail – Smart inventory management
Benefits of AIoT:
o Improved efficiency
o Real-time decision making
o Reduced human intervention
o Enhanced predictive capabilities
o Better resource utilization
Challenges:
1. Complexity – Integrating AI with IoT systems
2. Computational Requirements – Need for edge AI processing
3. Data Quality – Garbage in, garbage out problem
4. Security Risks – More sophisticated attacks possible
5. Cost – Higher implementation costs
Future of AIoT:
o Edge AI – Processing data locally on devices
o 5G Integration – Faster, more reliable connectivity
o Autonomous Systems – Self-managing IoT networks
o Explainable AI – Making AI decisions transparent in critical applications
Expert Systems → IoT/AIoT:
Early AI (expert systems) shows the value of domain-specific intelligence
Modern IoT/AIoT systems often incorporate rule-based reasoning alongside ML
IoT → AIoT:
IoT provides the data infrastructure
AI provides the intelligence
Together they create autonomous, intelligent systems
Practical Progression:
1990s: Expert Systems (rule-based intelligence)
2000s: IoT (connected devices, data collection)
2010s-Now: AIoT (intelligent, autonomous systems)