What is artificial intelligence?

In 1948, the computer science pioneer Alan Turing said, “A computer would deserve to be called intelligent if it could deceive a human into believing that it was human.” Although the processing speed and analytical power of a modern AI-driven computer would have seemed unbelievable to Turing, he nonetheless would likely have understood the ethical quandaries presented by that level of power. As AI gets better at understanding us and mimicking us, it increasingly seems human. And as we generate increasing amounts of personal data via digital channels, we need – more and more – to be able to trust the AI applications that underpin so many of our daily activities. Below are a few examples of ethical challenges that business leaders need to be aware of and monitor.

• Ethical use of customer data In the 2020s, the bulk of information we share and gather as businesses or individuals is done via digitally connected channels. At the start of 2020, there were over 3.5 billion smartphones in the world, all sharing enormous amounts of data – from their GPS location to their users’ personal details and preferences, via social media and search behaviors. As businesses gain wider access to their customers’ personal information, it becomes extremely important to set up benchmarks and ever-developing protocols to protect privacy and minimize risk.
• AI bias Bias can creep into an AI system due to human bias in the programming of its algorithms or through systemic prejudices that can be propagated via faulty assumptions in the machine learning process. In the first instance, it is more obvious to see how this might occur. But in the second, it can be harder to spot and avoid. A well-known example of AI bias occurred within the U.S. healthcare system, where AI applications were being used to allocate standards of care. The algorithm learned that certain demographic groups were less able to pay for care. It then extrapolated this information to mistakenly correlate that group as being less entitled to care. After discovering this unfortunate mistake, computer scientists at UC Berkeley worked with the developers to modify the algorithmic variables, thus reducing bias by 84%.
• AI transparency and explainable AI Transparency in AI is the ability to determine how and why an algorithm arrived at a particular conclusion or decision. AI and machine learning algorithms that inform outcomes – and the outcomes themselves – are often so complex as to be beyond human understanding. Algorithms such as this are known as “black box” models. For businesses, it is important to ensure that data models are fair, unbiased, and can be explained and externally scrutinized. Especially in areas such as aviation or medicine, human lives are at stake. It is therefore vital that the humans who use this data are taking data governance initiatives extremely seriously.
• Deepfakes and fake news Deepfake is a portmanteau of “deep learning” and “fake.” It is a technique that uses AI and machine learning to – typically – superimpose one person’s face on to another’s body in a video, with such accuracy as to be indistinguishable from the original. In its innocent form, it can result in amazing visual effects such as the 30-year de-aging of Robert De Niro and Joe Pesci in the film The Irishman. Unfortunately, its more common use is to create believable fake news stories or to put celebrities into graphic or compromising videos in which they never originally appeared.