AgricultureAutomotiveBusiness + ManagementData Science for GoodEducationFeatured PostGovernmentMediaOpinionAIData Science for Goodinequalitysolving inequality with aiposted by Ava Burcham, ODSC September 6, 2019 Ava Burcham, ODSC
When people talk about artificial intelligence, they often praise it as an unemotional, efficient task manager, and, to an extent, that’s what it is. But when we think about it in bigger terms than Siri or Alexa, we can see real ways people are solving inequality with AI. In this article, we’ll look at a few ways AI is already fixing real sociological problems of inequality, as well as consider some ways it could be implemented in the future.
3 Ways We’re Already Solving Inequality with AI
Automating Court and Hiring Processes
One AI application that’s gotten a lot of media coverage is automated court and hiring processes. The concept is that companies or the government can use machine learning to predict good candidates to hire or the likelihood of repeat offenses to decide on sentencing, which, in turn, would reduce individual biases. Unfortunately, most of the press has been negative, due to faulty programming and data bias—where the machine is learning from biased data that predicts men to be better candidates or white people to be less risky for shorter sentences.
Despite the rocky start to this application, companies are now re-training their machines to account for that bias and there’s real potential to this programming.
In industries that are and have always been dominated by one gender—due to a variety of reasons, usually not actually related to one gender’s higher aptitude to that job—hiring managers may be biased against the opposite gender and turn down qualified candidates. By incorporating AI into this process, we can remove that gender bias and undo that inequality, creating a company where candidates are truly hired based on merit over assumptions.
In the government there are many ways this technology could be implemented. This includes basics like saving money by automating menial tasks, but can be most impactful when applied to the prison and court system. There have been many cases where judges seem to give lower sentences to one race or another. This isn’t necessarily purposeful racism, but can be rooted in stereotypes and biases that can be hard for us to detect within ourselves. One company is exploring the idea of applying machine learning to the judges themselves, pointing out when they’ve shown a history of bias against one group so they may self-correct. Another option is having the machine create suggested sentencing based on the specifics of the case, to be approved or appealed by a judge. While there are risks to both of these options, if we’re careful about the data sets we train our programs with, these programs could be an incredible but simple step to reducing a huge problem of inequality.
Saving the Environment
There are actually quite a few ways AI can impact and better the environment, and while some wouldn’t consider the environment as an entity which can experience inequality, the deterioration of our planet does create inequality. Places which experience higher levels of pollution also experience higher levels of poverty. Monoculture—the farming of just one plant for miles and miles—creates huge detriments to our ecosystem, and is often carried out by huge corporations that employ and exploit underprivileged workers, and use heavy pesticide which also impacts our health. Water pollution is often highest in low-income areas, like in the case of Flint Michigan or developing countries, and impacts the health of those residents. Luckily, with AI, we have tangible solutions to all of these problems, and only need to focus on implementing them at scale.
With AI, air pollution can be monitored better and give cities better data to make decisions with. For example, London has “low emission zones,” and other cities increase their encouragement of carpooling, public transport, or bicycle use. Beyond monitoring, intelligent technology also lowers the amount of pollution we put out in the first place—as the use of electric cars spreads we’ll emit less exhaust, and hybrid cars are made to switch between gas and electric depending on which is more efficient in the moment.
Likewise, we now have better monitoring capabilities of water, meaning fewer people are harmed if or when water becomes unsafe, but we also have ways to clean up the pollution we’ve created. To tackle the trash already at sea, we have AI-based trash collectors which float around and suck in trash, without harming or “inhaling” the animals that live in the waters it explores. And to minimize the water we waste, a few different companies have implemented or created systems to track water-use. In one case, it was done to notify homes that were potentially overusing water during the California drought, in another case, it was used by farmers to make sure they weren’t over- or under-watering their crops.
Like the above example, AI can also help us better understand and solve land pollution or misuse. Farmers have begun using AI to track where their plants are being killed by insects, which means they can use pesticide only where they need to, which, in turn, means there’s fewer chemicals in our environment and food. AI systems can even help our endangered species by tracking their movement with footprint analysis and hidden cameras, which benefits our ecosystem on a whole. And, AI robots have been created that can accurately sort trash and recyclables.
Minimizing Food Shortage
Just as AI can help farmers capitalize on their profits and protect the environment, it can also help us make better use of food and reduce waste. This technology can sort crops and decide if the food is “good” or “bad” just like a person could, but it can also decide what a less-than-perfect crop would be best for, rather than just ending up in the trash. Maybe it’s just an odd shape, and it can go to Imperfect Produce, maybe it’s a bit squished, but would be perfect for making a canned product, the options go on. It would reduce food waste on the whole. AI and machine learning can also be used to predict food-related crises in cities—food shortages, crop failure, or droughts—and help prepare for them, rather than being caught off guard and suffering.
Both of these applications can be incredibly helpful to help with the food shortage we face. While we may still need to expand this reach to help developing countries, some of the highest level of waste and hunger happen in wealthy countries with high levels of inequality.
Potential Uses of AI Against Inequality
There are quite a few AI applications we’ve already seen introduced into the industry, but the options and benefits of those are still semi-unexplored. By considering our technology for the specific purpose of solving inequality, we can make strides that might have otherwise been overlooked.
Spotting Fake News
While AI that spots fake news isn’t necessarily new, the way we use it could create a huge impact on underprivileged people. It’s no secret that the media shapes public opinion, and the more fake news there is, the higher the risk we’re influenced into thoughts and actions that put others at a disadvantage.
There’s been fake news created to inflate stereotypes about and encourage hatred towards entire groups of people, with deadly repercussions. So too, fake news seems to have changed opinion about political candidates and sway entire elections. It’s a few steps removed from the direct inequality, but the use of AI to find and manage fake news could have a great impact on individuals’ lives, if we only use it as such.
This isn’t necessarily a piece of AI, but it is important: data science is becoming an incredibly accessible field to get into. There are more education opportunities popping up every day, many of which are free or low-cost, and, with practice, provide equal training for this constantly developing industry. These low-cost options remove the barriers to entry that so many private (or even public) universities have created, which allows low-income or underprivileged peoples access to this incredibly profitable and world-changing industry.
If hiring managers recognize that a degree doesn’t equate experience, data science could become one of the most diverse industries in the world. With more diversity comes more ideas and backgrounds to influence the programs created, and pushes the boundaries of what those programs can be used for.
The applications we currently have are doing huge good for the world, but the possibilities become endless when we consider the breadth of talent that could be joining the industry in the coming years. Soon enough, solving inequality with AI could be the obvious goal, or maybe even a goal of the past.