“This mission is too important for me to allow you to jeopardize it.”
These are exactly the sort of words that would make you launch your phone into the nearest river, if they had been uttered by Siri. Fortunately, they are the fictional words of HAL 9000, the sentient artificial intelligence system in the 1968 Stanley Kubrick film, A Space Odyssey: 2001. It told the story of a mission to Jupiter and the gradual realisation of the human crew that the perfect piece of AI designed to help them, was in fact fallible, and plotting against them to preserve its existence.
We don’t appear to have come much further in our collective sentiment towards trusting AI. The term “killer robots” has been splashed across the press headlines quite a bit recently, with some heavyweight names behind them, highlighting the potential dangers of using AI technology in warfare. Some of these warnings around the ethical usage of AI are undoubtedly justified. How do you prevent AI from learning bad characteristics as well as good? It doesn’t necessarily need to be as dramatic as the use of AI in war. It could be as simple as AI learning some of the sadly still intrinsic bias in society, such as that boys wear blue and girls wear pink. The stock archive this technology is likely to learn from has been written by humans. And humans have prejudices, fears, and ideas that they want to promote. AI may not be able to help learning some of these, and apportioning blame to the technology would be a mistake, but they could still have an impact on the service provided to us.
Ethical issues aside, nervousness around AI in the workplace is much closer to home, and again in many ways, there is justification for some jitters from employees. Technology has a history of replacing humans in the workforce—and the initial stages of this can be painful. Printing presses, weaving machines, mechanised farming, automated production lines, to name a few that have disrupted the workforce across industries. AI in banking could undoubtedly do the same if deployed without a long-term, sustainable plan from banks.
In our upcoming series of reports on AI in financial services, Accenture looks at the potential advantages and pitfalls of embracing AI in banking, capital markets and insurance.
“People x Process x Data = AI” is our view on the success of AI in the workplace. The process and data side we will come back to another time—but an equally important part of AI are the “people” that this technology will work alongside. Many have years of experience, most of which will not be written down for an AI colleague to pick up and assimilate into its own bank of knowledge. The importance of people is particularly significant in banking, where interaction between the bank and customer is still of vital importance to most, and must become a priority. Fifty-three percent of customers still go into their branch once a month or more. Customers like and want the reassurance of being able to speak to a person.
That is not to say that many would not be happy with a “phygital” blend of interaction with their bank. But if this is to be a success, then the workforce needs to be ready and able to use this technology. And with 30 percent of banking executives unsure that their current workforce has the necessary skills and experience to use AI technology to its optimum, there is cause for concern that the rollout of this technology could pose a problem for banks.
For it to be a success, a fully detailed proof of concept should be in place, with an inventory of the workforce skillset being of primary importance, before any decisions are made on how and where to use AI. Easier said than done? It needn’t be. Some simple “best practice” steps should help this along. To name a few:
- Involving the workforce in decision-making and investigations into how and where AI could help them in their roles would go a long way towards easing any transition of jobs
- Providing training to understand what the technology involves, and showing its limitations as well as its advantages
- Showing how AI could take away some of the more repetitive and frustrating parts of a function, leaving the employee to do the more interesting parts of their role, and take part in more creative and stimulating work
- Introducing roles that will make use of AI to create value within the business, and which need some human imagination to create. The lack of differentiation in products has long been lamented by customers. Using AI to simulate how a new product might work for a bank, in a fail-fast, low-risk environment, has its obvious advantages
Maybe the ethics of using AI is less around whether there is a risk it will learn our worst traits, and more around what our intentions are from using it. If it is just to slash the costs of the workforce, then employers are missing a trick, and could find themselves on the receiving end of public and regulatory disapproval. Their employees have something AI cannot learn: empathy and understanding of human nature. Both of which are vital in a customer-facing service, and which in its current format AI cannot provide on its own, meaning a combined AI/human workforce is necessary to get the best from this technology. The future is bright; the future is still human.