Skip to the content.

Computer Bias Homework

Homework and Popcorn hacks from computer bias lesson

Popcorn Hack 1

Example:

In the facial recognition software used by some law enforcement agencies, studies have shown that it is less accurate in identifying people with darker skin tones.

Who is affected:

This bias affects people of color, particularly Black and Brown individuals, who are more likely to be misidentified. This can lead to wrongful arrests or increased surveillance.

Potential Cause:

A potential cause of this bias is the lack of diverse training data. If the software is trained mostly on images of lighter-skinned individuals, it struggles to accurately recognize faces with darker skin tones.

Popcorn Hack 2

There was a time when I asked ChatGPT for help with a coding problem, but the response didn’t fully solve the issue. I was trying to debug an error in my code, but the explanation was too vague, and the solution didn’t address the specific bug I was facing. It was frustrating because I wanted a clear, working fix but had to spend extra time searching elsewhere. One way ChatGPT could be improved is by offering more detailed, step-by-step explanations and asking clarifying questions to better understand complex coding issues.

Popcorn Hack 3

Bias could sneak into a fitness tracking app by providing one-size-fits-all recommendations that overlook differences in physical abilities, age, or health conditions. For example, setting a universal step goal may exclude users with mobility challenges, while calorie estimates may be inaccurate for people with varying metabolic rates. To make the app more inclusive, I would offer customizable goals, alternative activity tracking for different abilities, and personalized progress markers based on each user’s unique needs.

Homework Hack 1

Digital Tool: TikTok

Propose a Solution: TikTok could update the algorithm to highlight a wider range of creators. They could also add tools for users to explore diverse content and regularly check for bias in the recommendations.