Artificial intelligence (AI) is changing the way we live and work by doing tasks automatically and making important discoveries in healthcare. It has a lot of potential to make us more productive and innovative. But, it’s clear that AI is not helping everyone equally. It might even make the differences between people worse, especially when it comes to things like race.
Leaders in businesses and government need to make sure that everyone can benefit from AI. Right now, it seems like AI is causing problems, and people are only trying to fix them after they happen, or sometimes not fixing them at all. To really solve the issues caused by AI, we need to take a proactive approach from the start.
If policymakers and business leaders want AI to be fair for everyone, they should first understand three ways AI can make things unequal. We suggest a simple framework to help with this. It covers these three ways but also looks at how AI changes what people want and need, which is another way it can cause inequality.
Our framework has three parts:
- Technological forces: Algorithmic biasThis happens when AI makes choices that unfairly hurt certain groups of people. For example, a healthcare AI might not give Black patients enough care because it wasn’t trained on enough data from Black patients. This is a problem because it’s not just unfair; it can harm people’s health. Bias in AI happens because the data it learns from might not include enough information about certain groups or because there are unfair ideas in the data.But just fixing bias isn’t enough. There are more complicated reasons why AI can cause inequality. We need to understand how AI changes the supply and demand for things, which also makes some people worse off.
- Supply-side forces: Automation and augmentationAI can make it cheaper to do some jobs by using robots or helping people work better. Some jobs can be done by robots, and these jobs are often the ones done by Black and Hispanic workers. This isn’t because AI is being unfair; it’s because it’s easier and cheaper to use AI for these jobs. But since many people of color have these jobs, using AI can make inequality worse.
- Demand-side forces: Audience evaluationsWhen people know that a professional is using AI, they might not want to use their services. For example, if you knew your doctor was using AI to help diagnose your illness, you might not trust them as much. This can make people want these services less, and it’s not just in healthcare; it happens in many other jobs too.Our research shows that people have different opinions about AI. Some are worried about it, some love it, and some are in between. This affects what people want and what they value. These differences can make inequality worse.
Understanding how people value AI-augmented work is important but often overlooked. It’s a big part of why AI can make some people better off and others worse off, especially when it mixes with bias against certain groups.
To make AI more fair, we need to think about all three of these things together. They are connected, and changes in one can affect the others.
For example, if a doctor decides not to use AI because patients don’t like it, it can hurt the doctor’s business and the patients. Also, it can mean that AI doesn’t learn from certain groups of patients, making it worse for them in the long run. So, it’s like a tripod; if one leg is weak, the whole thing falls.
To stop this from happening, we should create ways for people to understand AI better. If companies that use AI explain how it helps people instead of replacing them, it might help. And it’s not just about fixing bias and making AI work better; we need everyone—companies, governments, and experts—to work together to make sure AI benefits everyone. This way, we can have a future where AI helps everyone equally.
Image Source: https://cointelegraph.com/news/bias-in-ai-what-can-blockchains-do-to-ensure-fairness