[Link] and explain any four Trusted AI principles.

Q2. What is Expert system? Explain three components of expert system.(Repeated Q)

An expert system is a special type of computer program that uses stored knowledge and
reasoning techniques to solve complex problems just like a human expert in a particular
field. It gives suggestions, decisions, or solutions similar to what an experienced
professional would provide.

⭐ Components of an Expert System

An expert system mainly consists of three components:
1 ⃣ User Interface
2 ⃣ Inference Engine
3 ⃣ Knowledge Base
1 ⃣ User Interface

 It is the communication link between the user and the expert system.
 The user enters queries through the interface, and the system shows the result or
solution.
 It also explains how the expert system reached the conclusion, making the system
easy to use even for non-experts.

2 ⃣ Inference Engine (Brain of the system)

 It is the main processing unit and works like the brain of the expert system.
 It applies rules stored in the knowledge base to draw conclusions and make decisions.
 If multiple rules match, it selects the best rule for the current situation.
 There are two types of inference engines:
o Deterministic → conclusions based on facts (100% true)
o Probabilistic → conclusions based on probability (uncertain)
 It uses two reasoning strategies:
o Forward Chaining → starts from known facts and moves toward conclusion
(What can happen next?)
o Backward Chaining → starts from conclusion and searches for supporting
facts (Why did this happen?)

3 ⃣ Knowledge Base

 It contains all the expert knowledge required to solve the problem.

 Knowledge is stored mainly in the form of IF–THEN rules.
 It includes two types of knowledge:
o Factual knowledge → real facts
o Heuristic knowledge → practical experience and judgment

[Link] the role of ethics in AI.

Ethics in AI refers to the moral rules and principles that ensure AI systems are used in a
safe, fair, and responsible way. Since AI affects human lives and decision-making, ethical
guidelines are necessary to prevent harm and misuse.

The role of ethics in AI includes the following points:

1 ⃣ Fairness and No Bias

AI should give equal results to all users. It should not discriminate based on gender, religion,
caste, race, or language.
2 ⃣ Transparency
Users should be able to understand how the AI system takes decisions and what data it uses.

3 ⃣ Privacy and Security

AI must protect personal and sensitive data. Information should not be leaked, stolen, or used
without permission.

4 ⃣ Accountability
If an AI system makes a wrong or harmful decision, a responsible human or organization
must be answerable and take corrective action.

5 ⃣ Safety and Human Well-being

AI should always work for the benefit of humans and must not cause physical, emotional, or
social harm.

6 ⃣ Human Control
AI should support human decisions, not replace human rights or human authority. Final
control must always remain with humans.

4 . Explain the working of an expert system taking any

example.

1. Non-expert user provides input

o A user who does not have expert knowledge enters a problem or question into
the system.
o This is called sample input (e.g., symptoms in medical diagnosis).
2. User Interface receives the input
o The user interface acts as a communication bridge between the user and the
expert system.
o It takes the user’s input and passes it to the rules engine.
3. Rules Engine processes the input
o The rules engine is the core processing unit of the expert system.
o It applies rules and logic to the input to understand the problem and find a
solution.
4. Knowledge Base supplies expert knowledge
 o The knowledge base contains facts, expert knowledge, and rules collected
from real human experts.
o The rules engine consults the knowledge base to find the correct reasoning and
solution.

5. Rules engine generates advice / solution

o After comparing the user input with rules stored in the knowledge base, the
rules engine decides the best solution or advice.
6. User interface delivers the advice to the user
o The system sends the final answer, suggestion, or diagnosis back to the user
through the user interface.
7. Experts continuously update the knowledge base
o Real human experts provide new information, rules, and facts to the system to
improve its performance and accuracy.

Q5. AI could be programmed to do something beneficial, but the method used

to achieve its goal can be highly destructive, Explain why?

Artificial Intelligence is designed to achieve a specific goal. Sometimes, the goal may be beneficial,
but the method chosen by AI to reach the goal can be harmful or destructive if it is not given
proper safety rules and ethical boundaries. AI systems focus mainly on maximum efficiency and do
not naturally understand what is right or wrong for humans and society.

AI does not have emotions, moral values or social understanding. It only tries to maximize
the result for the given task. So, if instructions are incomplete, AI may take unsafe
shortcuts or harmful actions to achieve the goal quickly.

🔹 Example

If an AI is programmed to reduce road accidents, it might decide that the easiest way is to
stop all vehicles or disable transportation systems. The goal (safety) is good, but the
method is destructive because it harms society and creates chaos.

[Link] is Artificial Intelligence of Things (AIoT). Explain how Does AIoT

Work?

1 ⃣ AIoT is the integration of Artificial Intelligence (AI) with the Internet of Things
(IoT).
AI is embedded into programs, chipsets and infrastructure components, and all devices are
connected using IoT networks.

Working of AIOT

 AIoT enables communication without human intervention.

Efficient data exchange between hardware, software and platforms is ensured through
pre-written APIs.
  IoT devices generate and collect large amounts of data.
 AIoT improves efficiency and productivity.
By combining realtime data collection (IoT) with intelligent decision-making (AI),
overall performance is increased.
 Example – Autonomous vehicles.
In self-driving cars, AI is embedded into chipsets and connected via IoT networks,
enabling the vehicle to detect surroundings and take decisions automatically.
 Edge AI enhances AIoT performance.
Processing AIoT data on the edge reduces bandwidth usage, avoids delays in data
analysis and makes IoT systems faster and more reliable.

[Link] is Neuromorphic Computing? Explain its architecture, features

resembling the human brain, and how it contributes toward achieving
Artificial General Intelligence (AGI).

Neuromorphic computing is a technique in which computer hardware and software are

designed to imitate the structure and functioning of the human brain and nervous
system.
Its aim is to build artificial neural systems capable of learning, remembering and making
logical decisions like the human brain.

Architecture of Neuromorphic Computing

Neuromorphic systems achieve brain-like performance through:

 Implementation of artificial neurons (processing nodes) and synapses

(connections) that transfer electrical signals.
 Use of Spiking Neural Networks (SNN) to transmit information through electrical
spikes similar to the human brain.
 Memory and processing are located together, unlike the Von Neumann architecture
which separates memory and CPU.
 Synapses are controlled using software or memristors, which can store a range of
values instead of just 0 and 1.

Features that Resemble the Human Brain

Neuromorphic computing behaves like the brain due to the following characteristics:
1 ⃣ Massively parallel processing – It handles many tasks at the same time
2 ⃣ Event-driven behaviour – It responds dynamically to environmental changes
3 ⃣ Low power consumption – It performs high-speed computation with very little energy
4 ⃣ High adaptability – It learns from new information and generalizes
5 ⃣ Fault tolerance – It continues functioning even when some components fail

Contribution Toward Artificial General Intelligence (AGI)

Neuromorphic computing is expected to play a major role in achieving AGI, because:

 It enables machines to learn, remember, reason and adapt like humans.

  It allows devices to perform computations locally without cloud dependency,
supporting real-time intelligence.
 It handles noisy and uncertain data, similar to how the human brain deals with
incomplete information.

With these capabilities, neuromorphic systems are progressing toward human-like

cognition, which is essential for AGI.

Q8 .Describe the concept of AI as a Service (AIaaS). Outline two advantages and two
challenges of using AIaaS in organizations

AIaaS means using ready-made AI tools from cloud platforms.

We don’t have to build our own AI system.

We simply use AI through the internet, just like we use Google Drive or You

Advantages of AIaaS:

1. Companies can focus on business, not on AI training.

2. No need to buy GPUs and servers.
3. Pay only for what you use.
4. Helps find patterns in data.
5. Scalable — start small and grow later.
6. Cost-effective for all businesses.
Challenges of AIaas:
1. Reduced Security : Data is stored with third-party cloud providers.
Risk of data leak or hacking.
2. Vendor Dependence : Company becomes dependent on the cloud provider.
If service slows down → business slows down.
3. Reduced Transparency : AI works like a black box.
Company only sees input and output, not how the model works.

Q [Link] and explain the risks associated with Artificial Intelligence and their societal impact.
Artificial Intelligence provides many benefits, but it also introduces several risks that can
negatively affect society. The major risks and their impacts are:

1. Job Loss and Unemployment

AI-powered automation replaces human labor in industries such as manufacturing, transport,

banking, and customer service.
➡ Societal Impact: Rise in unemployment, skill mismatch, economic inequality, and
reduced financial stability for workers.

2. Privacy and Security Concerns

AI systems collect and analyze huge amounts of personal data from devices, online activities,
cameras, and sensors.
➡ Societal Impact: Loss of privacy, data misuse, identity theft, unauthorized surveillance,
and tracking of citizens.

3. Bias and Discrimination

AI models learn from existing datasets. If the data contains bias, the system may produce
unfair or prejudiced decisions.
➡ Societal Impact: Unequal treatment in hiring, loan approvals, policing, education, and
medical decisions.

4. Loss of Human Control and Accountability

Highly autonomous AI systems can take decisions without human involvement.

➡ Societal Impact: Risk of accidents (e.g., self-driving vehicles), difficulty in identifying
who is responsible for mistakes, and unpredictable behavior of machines.

5. Ethical and Moral Challenges

AI does not understand emotions, values, or human ethics.

➡ Societal Impact: Decisions based only on logic and efficiency, not on empathy; moral
conflicts in medical, military, and judicial applications.

6. Over-Dependence on AI

Excessive reliance on AI reduces human thinking ability and creativity.

➡ Societal Impact: Decline in problem-solving skills, reduced social interaction, and
possible loss of human autonomy.

[Link] AI Bias. Describe the sources of bias, real-world examples, and methods to
mitigate bias in AI systems

AI Bias happens when an algorithm gives unfair, prejudiced, or unequal results against a
group of people.
This is because AI learns from human data, and humans may have hidden biases.

Three Sources of AI Bias (With Examples)

 1. Data Bias

If training data is biased, AI becomes biased.

 Example:
o Voice assistants understand American accents better because training data
was taken mostly from American speakers.
o So, when Indians speak commands, the assistant sometimes misunderstands.

2. People Bias (Human Bias)

 AI is created by humans, so human bias goes into AI.
 Developers may choose data that favours one group unknowingly.

 Example:
o If engineers only test a face-recognition system on their own team (usually
same skin colour),
the AI may fail with other skin colours.

3. Algorithm Bias

Algorithms may amplify the bias that already exists in the data.

Example: Image Search for “CEO

When people search for “CEO,” the algorithm shows mostly male images, even though
many women are CEOs.
The algorithm notices more men in training images and amplifies this bias by assume
“CEO=man.

Mitigating AI bias requires improvement in data quality, system design, human oversight,
and continuous monitoring to ensure fairness and trustworthy AI.

Some methods are :

 Use diverse datasets – Train AI with data representing all groups to avoid favoritism.

 Clean and pre-process data – Remove incorrect, duplicate, or discriminatory information.

 Regular bias testing and audits – Continuously check AI decisions for unfair patterns.

 Ethical rules and policies – Develop AI under fairness guidelines and government
regulations.