How to keep human bias out of AI (Kriti Sharma)

https://www.ted.com/talks/kriti_sharma_how_to_keep_human_bias_out_of_ai/transcript#t-535359

Transcript of "How to keep human bias out of AI"

 

1. What information do I learn about the speaker and their background?

 

She is a developer who makes algorithms of AI, joining forums and sharing ideas about AI. In that process, she found people’s bias we had not noticed, such as gender problem. She sometimes got others’ doubt that she is not qualified to say something about AI, but she struggled herself to prove what she is saying, by making her own profiles to do some experiences. She noticed women’s ideas are less likely to be accepted than men’s, so she should try to hide herself.

 

2. Summarize the main idea of the talk in one sentence.

 

AI strengthens human bias but improve the quality of people’s lives.

 

3. What supporting details or examples does the speaker use to support their ideas?

 

Though AI makes our life convenient, our bias is getting reinforced in the process of interacting with AI. For example, some of us might not have noticed, well-known AI secretaries are all female. This can make people have prejudice that secretaries should be women, especially strong to young children who are accepting everything without sense to discriminate something true or false.

And she mentioned women developers are less evaluated, which showed that if women hide her gender, people trust her developments four percent higher. That survey improved the reliability of the bias.

 

4. How does the speaker engage the audience? For example, using images, charts, humor, actions.

 

She used signposting phrases properly, which helped audience’s understanding (That is what I need I think). When she was giving a speech about what we must remember in this high-tech society at the end of her speech, she suggested two things that we should remember. Due to those signposting phrases, listeners can organize the contents of her speech in a structure.

 

5. What is your opinion of the talk? 

 

We must think how much decisions in our daily life are made by AI, not myself. That gave me a huge shock and got to realize there are a lot of AI-based information around us. I really did not imagine we were getting influence of AI, though we are the ones who control and make the algorithms of AI. And there were many prejudices we did not know, and I felt strongly that these would affect children, making a society in a bad way in the future. I kept in mind we all should try from the beginning to get rid of these biases. And I think it would be important for me to use signposting phrases when making a presentation, which helped me to understand easily.