Page 67 - CT_AI_Class-8

Discuss the results and identify whether the AI makes any incorrect predictions due to limited or
biased training data. This activity helps students understand how supervised learning works and
how AI systems learn to identify objects using labelled image data.

Checkpoint Fill in the blanks.
1. Unbalanced data can make AI's predictions or incorrect.
2. means the data is unbalanced or unfair.

3. Many online platforms use AI to videos or articles that users might enjoy.
4. It’s crucial to understand how an AI system works and how it makes its .

21 st #Media Literacy
Century #Technology Literacy
ai in action Skills

Moral Machine is an online platform created by MIT to explore human moral preferences
regarding autonomous vehicles. It presents users with ethical dilemmas, where they must
decide how self-driving cars should react in life-threatening situations, such as choosing
whom to harm in an unavoidable crash. The platform collects data on these decisions to
better understand global ethical views on machine decision-making. It helps inform how AI systems, like
autonomous vehicles, should be programmed to make ethical choices in real-world scenarios.
Visit the given link or scan the QR code to play the game: https://www.moralmachine.net/

21 st
Century #Information Literacy
Skills
ADDRESSING GENDER BIAS IN AI HIRING SYSTEMS
TechSolutions Inc., a tech company, implemented an AI-powered hiring tool to streamline
recruitment by analysing past hiring data. However, after several months, the company noticed
a bias in the system, where male candidates, particularly for technical roles, were disproportionately
favoured. Upon investigation, the company found that the dataset used to train the AI contained historical
biases, including male-dominated hiring patterns and biased interview ratings that favoured men. This
led to the AI system perpetuating the gender imbalance, undermining the company’s diversity goals.
To address this, TechSolutions Inc. rebalanced the dataset to ensure gender parity and applied bias
mitigation techniques, such as fairness constraints during model training. The company also introduced
regular audits to monitor the AI system's performance and implemented human oversight in the final
hiring decisions. These changes helped achieve a more balanced gender representation in new hires,
aligning the company’s recruitment process with its diversity and inclusion objectives.
Based on the above case, answer the following questions:

1. What was the main issue TechSolutions Inc. faced with its AI-powered hiring tool?

2. What steps did TechSolutions Inc. take to address the gender bias in the AI system?

Data and Fairness in AI 65