In spite of high-profile arguments between tech giants about the impact of artificial intelligence (AI) on the safety and future of human civilisation, the conversation around AI in business and the associated risks is a much more nuanced debate than one might think, writes Dr. Rob Walker, the vice president of decision management and analytics at Pegasystems.
To appreciate these nuances, it is important to be aware that AI comes in two particular flavours: Transparent AI and Opaque AI. Transparent AI is a system whose insights can be understood and audited, allowing one to reverse engineer each of its outcomes to observe how it arrived at any given decision. Conversely, Opaque AI is an AI system that cannot easily reveal how it works. Not unlike our own minds, it can be challenging for it to try to explain exactly how it has arrived at a certain insight or decision.
So how does a business decide between the different types of AI and which is most appropriate?
For starters, there is no ‘good’ or ‘bad’ AI – only appropriate or inappropriate use of each system depending on one’s own needs. Although it may initially come across as risky, Opaque AI has a number of positive aspects which can prove useful in the right situations. In some instances, being transparent can be a constraint on AI and will limit its power and effectiveness – so in more complex use cases an Opaque system might be preferable.
But the issue of bias poses a potential problem in an Opaque system. Without complete human control, an Opaque AI system may begin to favour policies that contradict an organisation’s brand promise, all without the organisation’s knowledge. It’s quite easy for an AI system to use neutral data to work out customer details, which it can then use to make non-neutral decisions. An Opaque AI may, for instance, interpret customer data and use it to start offering better deals to people based on race, gender or other demographics – leading, of course, to potentially disastrous outcomes.
Fortunately, organisations can work out whether or not they are using AI correctly, and which kind of AI is appropriate, by determining the extent to which they are willing to trust it. To fully trust an AI system, either the AI must be transparent so that business management can understand how it works – or, if the AI is Opaque, it has to be tested before it is taken into production. Tests must be extensive, going beyond merely searching for viability in delivering business outcomes and looking for the kind of unintended biases suggested above.
There are also other factors at play, particularly for those organisations using AI as part of a customer engagement system. With GDPR coming into effect in May 2018, companies will have to be able to explain exactly how they reach certain algorithmic-based decisions about their customers. Organisations could use some sort of a switch, a ‘T-Switch’, to increase transparency by forcing the methods used by AI to make decisions from Opaque to Transparent. As they’ll be more easily able to comply, they gain a distinct advantage.
Public debates between the likes of Elon Musk and Mark Zuckerberg often ignore the fact that AI is something that’s already in widespread use by many in a business context today, and the current risks associated with it aren’t about whether it will leave us all in destruction. Instead of worrying about such doomsday scenarios, we should focus on the very real dangers posed by AI technology being used incorrectly today. Risks can include diminished business value, significant brand damage and violations of regulation. Though these don’t spell the end of human civilisation, they can still have a tremendous impact on the success or failure of an organisation
