In AI We Trust

Imam Hoque asks what needs to be done to ensure AI is trustworthy and how we can introduce some transparency into its ‘black box’ decision making

- Imam Hoque

Artificial intelligence is predicted to contribute up to $15.7tn to the global economy by 2030 and is seen as the key to tackling many of the biggest challenges we face worldwide, from global health and wellbeing to fighting organized crime, as well as reliable legal and democratic systems. As a transformative technology, it presents a huge opportunity to raise growth and prosperity and change the world as we know it.

AI is already becoming part of the fabric of our society, from chatbots to recommendations of which articles to read online – but as its use continues to grow, so do concerns about the ethics around black-box AI solutions. How do we trust something that is now becoming so sophisticated it is often beyond our understanding? How can we know if the results are biased? And how can we regulate its use to ensure its further uptake and development to better our lives and our planet?

THE BLACK BOX

Trust is crucial in all forms of software. Trust mechanisms are based on the deterministic nature of the software application, where behavior is determined in such a manner as to make it innately predictable. Consequently, any outcome or decision the software suggests or makes will be entirely explainable and it is this transparency that protects against any decision-making that might be unfair.