0:0023:03

Does Artificial Intelligence (AI) Support Diversity, Equity and Inclusion?

Patrice Caire 23:03 2022

A job ad hidden from anyone over 54 shows how AI quietly discriminates

A job ad sat open on one woman's screen while her 60-year-old friend could not see it at all: it was shown only to profiles aged 18 to 54. Computer scientist Patrice Caire, a consulting scientist at Stanford University's Robotics Lab, traces cases like this, and a husband granted 20 times his wife's credit limit, back to the data AI learns from. What can users, companies and governments actually do about it?

Learn how biased training data skews AI and what users, companies and governments can do

Watch on YouTube

In short

Computer scientist Patrice Caire argues that AI systems often reinforce society's inequalities because the data they learn from, and the teams that build them, lack diversity. Using cases such as age-targeted job ads, mislabelled images, CEO image searches and Apple Card credit limits, she explains why training data matters. She calls for vigilant users, transparent companies, diverse teams and enforceable AI standards.

“No wonder then that the point of view of women is not included in the design and development of the artificial intelligence tools and systems that we use every day.”
“That was a bit of a quick fix, but probably it was cheaper than to address systematically the issue.”

What you will learn

  • Explain how the size and diversity of a training data set shape what an AI system can recognize and decide.
  • Recognize everyday cases where AI discriminates by age, gender or race, from job ads to image search and credit limits.
  • Distinguish the statistical and computational biases that company toolkits address from the human and systemic biases that NIST says remain below the surface.
  • Apply practical steps as a user: report ethical issues, demand data set transparency and ask companies to disclose automated decision making.
  • Follow the regulatory and standards work of bodies such as NIST and IEEE to help close the legal void around AI.

Key takeaways from the speaker

  • Understand that today, AI impacts almost everyone, but that the negative impacts are particularly affecting women and minorities.
  • Learn why scientific experts are urging governments and companies to take action and develop AI standards and regulations.
  • See how critical it is for all citizens be aware of AI biases and learn how to identify them.
  • Grasp the significance of the input data used to train AI models.
  • Realize that to prepare the future, women and minorities should become full participants in the design and development of the AI technologies that shape our world.

Details

Level
Beginner
Format
Talk
Technical depth
Introductory technical
For
All career stages
Length
23:03
Published
July 2022

Questions this talk answers

Why do AI systems produce biased results?

They learn from training data, so when a group is missing or mislabelled in that data, the system cannot handle it correctly. Caire calls this the zebra problem.

Are quick fixes enough to remove AI bias?

No. Google removed the gorilla label instead of fixing the data, and the patched CEO image search fell back on outdated results with no women or people of color after a slight change in the search term.

What can ordinary users do about biased AI?

Stay on the lookout, report ethical issues, talk to each other and demand that companies be transparent about their data sets and their use of automated decision systems.

How can AI be regulated when there is no clear legal framework?

Caire advises following what governments and bodies such as the IEEE publish, joining petitions and discussions, and staying informed about advances in AI standards.

Why did digital assistants have female voices?

According to the social science research Caire cites, female voices seemed more passive and amiable, which people preferred in an assistant. She notes Apple only added a gender-neutral Siri voice that year.

About this talk

Today, Artificial Intelligence (AI) is deployed everywhere—from smartphones to search engines—raising ethical questions particularly concerning women and minorities. However, governments and companies have been slow to address AI biases and discriminations. In this talk, I focus on the question whether AI supports diversity, equity, and inclusion. Using daily life use cases and examples, I illustrate the significance of the input data while training models. Then I stress the importance to recognize, understand and remove biases in this input. All along, I present recommendations like establishing AI standards and regulations in order to fill the legal void in which companies operate and including diverse demographics that reflect the composition of our society, to design, develop and test current and future AI technologies.

Topics

Related talks

Browse the library