Page 75 - CT_AI_Class-7
P. 75
This is another example of bias. The teacher is making a decision based on who is loud or confident,
rather than who has the best ideas or skills for the project. This is unfair because the quieter
students may have important contributions but aren’t given the chance to share them.
In both cases, decisions are being made based on unfair factors, not taking into account the
abilities or potential of everyone equally. Bias in decision-making can cause missed opportunities
for many students.
Types of Bias
Bias in AI happens when the system makes unfair decisions based on wrong or incomplete
information. Let's look at some types of bias that can affect AI systems:
Data bias: Data bias happens when the information used to train an AI system is incomplete
or unfair. If the data doesn't represent everyone equally, the AI might not make fair decisions.
For example, imagine a school’s AI system that picks students for a competition, but it only
looks at the achievements of boys. The AI might unfairly choose more boys, even though girls
are just as talented. This is data bias because the system was trained on data that didn’t
consider all students equally.
Historical bias: Historical bias happens when AI learns from past decisions that were unfair.
If the data includes decisions that favoured one group, the AI might keep making the same
unfair choices. For example, if a company has hired more men than women in the past, an
AI system trained on that data might keep favouring men. This is historical bias because the
system is learning from the past, which may not have been fair.
Measurement bias: Measurement bias happens when the wrong things are used to judge or
measure something. If the wrong factors are considered, the AI might give the wrong results.
For example, imagine a school that judges students based only on how many hours they
study, instead of how well they perform in tests. This could be unfair because it doesn't show
how well the students understand the material. This is measurement bias because the wrong
criteria are being used to make decisions.
Ethics and AI Bias Awareness 73
