Page 85 - Ai Book - 8

Types of AI Biases

There are three main types of AI a biases.
1. Perceptive Biases: Perceptive biases happen when the way we see things is influenced by our feelings or
opinions. It’s like wearing sunglasses that make everything look a bit different. For example, if we really
like someone, we might think all their actions are amazing, even if they’re not. Imagine your best friend
draws a picture. Because you like your friend so much, you might think it’s the best picture ever, even if
there are parts that could be improved.
2. Incomplete Data Bias: Incomplete data bias occurs when we make decisions based on only part of the
information. It’s like trying to solve a puzzle with missing pieces. If you don’t have all the pieces, your
picture (or decision) might not be complete or accurate. Think about doing a school project with only
some of the information. If you don’t have all the details, your project might not be as good as it could be.

3. People Bias: People bias happens when we treat others differently based on how they look, where they’re
from, or other reasons. It’s like judging a book by its cover without knowing the interesting stories inside.
Imagine not playing with someone at school just because they wear different clothes. You might miss out
on a great friend because you didn’t give them a chance.
Understanding these types of biases helps us work towards making AI that’s fair and friendly. As AI becomes a
bigger part of our lives, making sure it treats everyone equally is super important.

Knowledge Botwledge Bot
Kno

Researchers and Scientists believe that AI will take over 16% of current jobs within the next
10 years.

Pop Quiz Quiz
Pop

State ‘T’ for True or ‘F’ for False statements.

1. We cannot differentiate between good and bad things.
2. The term ‘AI ethics’ refers to a set of moral standards or principles that governs the use
of AI systems.
3. The field of AI raises various complex ethical issues all over the world.

4. Chatbots are not grasping human jobs.

BIAS IN AI: UNDERSTANDING GIGO PRINCIPLE

AI systems are intelligent creations that make decisions
based on input data. If the input data is biased, AI
systems can produce biased outcomes. Example:
Amazon developed an AI system for hiring software
engineers in 2014 based on the data from resumes
submitted over the past 10 years. The algorithm
favoured men over women since historical data
showed a higher number of male candidates being
hired. As a result there was gender discrimination
due to biased input data.

85
85