Ethics & AI: Equal Access and Algorithmic Bias
About this video
AI could help society in education, healthcare and science, but it is already influencing important decisions about people's lives. Because machine learning depends entirely on the data it is fed, biased real-world data can make AI repeat racial and other biases, for example in home loans or criminal charges. The speakers argue for including vulnerable and marginalised people, spreading AI's benefits to everyone, and bringing more diverse perspectives into AI development.
You'll be able to
- explain how biased data can lead to unfair AI decisions.
- explain why the perspectives of vulnerable people matter when building AI.
- describe how AI could be democratised so it benefits everyone.
Key words
- Algorithmic bias
- When an AI system's decisions are unfair to some groups because of patterns in its data or design.
- Marginalised
- Pushed to the edge of society and given less power or attention.
- Underprivileged
- Having fewer advantages, such as money or opportunities, than others.
- Democratisation
- Making something available to everyone, not just a few.
- Machine learning
- The type of AI dominating today's applications, which learns from the data it is fed.
Transcript
No matter what field you end up going into, it's quite likely that AI is going to have some impact on what you're doing. The potential for AI to help society is enormous! It's something that is influencing a lot of very important decisions about real humans and their lives.
It could be used in education, to be more of an equalizer between people. It could be used in healthcare to develop new drugs. It could be used in science to develop new technologies. Like any technology its application will depend on how it is utilized.
At the same time, we need to think about the risks that are associated with doing that. The consequences are huge! The kind of artificial intelligence technology that really dominates the applications we're seeing today is machine learning. Machine learning depends entirely on the information that you feed it.
The problem is that with real world data, there's often information in there that you didn't intend to be in there, but is captured because of the bias in the data collection process. So if you're building an AI to determine who gets a home loan, or who should be charged with a crime, it could definitely bubble up the racial biases that humans and our current society already does.
A lot of what it means to build less harmful AI is really systems that are including the perspectives of those that are most vulnerable or most marginalized, most likely to be hurt by the deployment of that system. In many ways I've worried that the people who are particularly vulnerable to AI are the people who are already underprivileged in many respects.
Most people in the world just have AI applied to them, rather than playing an active role in guiding what AI gets applied to. Everybody you know has a computer in their pocket. That's young people, old people, rich people, poor people. To me, that's actually quite exciting from a democratization of technology perspective.
It means that AI, powerful as it is, could theoretically be in everybody's pockets, benefiting everybody. We should strive to make sure that things that provide value for society can be reached to anybody. How do we give a greater voice to the people who are being impacted by AI, to in turn be able to turn around and impact how AI is used for them?
Every time when you're looking at a new problem, you have an opportunity to change the world. Sometimes we succeed, sometimes we don't, but we always try. It's really critically important that we have as many diverse perspectives as possible, influencing the development of AI.
We need the participation of more women, more people of color, to provide a different perspective and a different lens on which problems matter and how we should approach these problems.
Video © Khan Academy, used under CC BY-NC-SA. Watch on Khan Academy. Puzzicle is not affiliated with Khan Academy.