So you’ve probably heard a bit about artificial intelligence in the media. You may have seen snatches of AI breakthroughs on the news, or watched that Simpsons episode where the house takes over and tries to kill Homer. There is a lot of sensationalism out there, and we know that sensationalism pays. However, some of these fears are becoming close(ish) to reality. These days it is not entirely off the wall to imagine our jobs being taken over by robots, self-driving cars running over pedestrians, and our personal data being stolen.
If that isn’t bad enough, AI has been shown to amplify bias against certain races and genders due to algorithms being built on data reflecting societal biases. In one example, an algorithm used in courtrooms across the United States of America was shown to be prejudiced against black defendants. The programme, Correctional Offender Management Profiling for Alternative Sanctions (COMPAS), was created in order to identify whether criminals were more or less likely to reoffend. A report, compiled by ProPublica, asserts that the programme wrongly flagged black defendants as twice as likely to reoffend in comparison to white people. 45% of black defendants were falsely labelled as at higher risk of reoffending, whereas only 24% of white defendants were incorrectly put in the same category. Further evidence showed that only 28% of black defendants were labelled as being at lower risk of reoffending, compared to 48% of white people.
Clearly, we have good reason to be anxious about biased AI programmes. But, the robots are coming, whether or not we want them to. So, instead of fear mongering or burying our heads in the sand, we have to work on a way to build trust in AI as a society. Demonstrating how AI can be used to produce helpful outcomes is key to building trust. It is also up to governments and international organisations to form policies to ease people’s fears.
To build more trust in AI there are a number of areas that need to be addressed:
Enabling more women and other underrepresented groups to take leading roles in the creation of AI.
Understanding and addressing human prejudices, which can often be written into data and programmes, skewing the actions and outcomes of AI tools.
Focusing on addressing gaps (such as equal representation across genders, races, and age groups) in big data.
Finding solutions for people who have lost their jobs, or are at risk of losing them, to automation.
With all that said, it’s not as if our governments and international organisations are doing nothing. In fact, a lot of time and money is going into solving these issues, and a number of organisations are making important steps in the right direction.
The United Nations Global Pulse is a network of innovation labs that focuses on ‘harnessing big data for development and humanitarian action’ and helping public sector institutions create well-informed policy that uses big data and AI for public good. UN Global Pulse works on a number of projects to help close data gaps while also contributing research to help achieve the Sustainable Development Goals (SDGs) through the use of real-time data and AI.
Pulse Lab Kampala is one such UN Global Pulse initiative. It has developed a radio application that monitors what local radio discussions are focusing on. The aim of the project was to demonstrate that UN agencies can gain insight into local concerns through the use of AI and big data analytics. Findings from the initiative were then used to help design local programmes to enact the SDGs.
In 2016, the US Government’s National Science and Technology Council formed the Subcommittee on Machine Learning and Artificial Intelligence. It was created with the aim of deriving the greatest possible benefit from AI, as well as addressing the challenges this new technology could pose both now and in the future. The Subcommittee went on to publish the National Artificial Intelligence Research and Development Strategic Plan, with the specific aim of developing policy relating to the use of AI. National policy efforts around AI such as these helped contribute to the US’s high ranking of second place in – Dr. Rami Shaheen’ 2017 Government AI Readiness Index. It remains to be seen whether the work of the new administration will help or hinder the country’s score in the 2019 version of the Index, due to be published later this month.
Many countries have started to develop AI strategies which explicitly tackle the four areas crucial for building trust in AI mentioned above, which is helping citizens not only become more aware of the technology, but also to trust it more. France, Italy and Canada are great examples of countries that have invested time and resources into developing strategies and solutions to the problems faced when it comes to the use of AI and data.
So, should we be scared of AI at all? Our fears are not unfounded, but as we have seen, there are numerous measures being put in place to protect us and our future. So long as governments and international organisations continue to work towards enacting policy that both protects us (whether that’s from job loss and/or out of control self-driving cars) and encourages more diversity in the field, then we may be on track for AI to be a constructive addition to society. But for now, I’ll still remain a little cautious, just in case (as in the Simpsons) my house is taken over by an AI Pierce Brosnan.
Do you want to learn more about AI? Check out our ‘AI in a week’ free online course, which introduces the key issues and debates around AI.