That AI is set to grow and become part of our decision-making world is not in dispute. Indeed the flurry of statistics and surveys out there point to a world where AI makes many of the decisions that humans still make.
According to Juniper Research, the amount of loans agreed by AI ‘advisors’ will reach $315 billion by 2025. The total value of assets under management by ‘roboadvisors’ by that time will be $3 trillion. This is slightly more than the GDP of India.
This is great if you sell AI solutions. And there are predictions out there for many other industry sectors where black and white, yes/no decisions can be made using ever more sophisticated lookup tables and rules. The legal profession is one ripe for a real AI makeover.
Whilst this growth is good for the technology industry, AI taking over so many jobs and job functions must come with a dose of common sense.
One area where AI is set to grow is in customer service, with the use of chatbots. In principle this is fine but the use and implementation of a chatbot service must be first-rate. If it is not, or AI is implemented in areas where there is a level of discussion to be had, then these areas should be avoided.
How often can you say that you used a chatbot on a site and were absolutely happy with the service? More often than not, our experience has been a circular, frustrating interaction (that sometimes feels like an argument) where the same list of suggestions or options pops up, even if you have said none of them is relevant.
Looking at the banking statistics must also raise some questions. Of the $315 billion worth of loans approved by AI in 2025, what proportion have circumstances that an AI would not be able to make a judgement about. But still does. And if an AI does make a judgement against approving a loan and a human would have approved it, what effect will that have on the customer concerned. Will his credit rating take a battering, will his insurance premiums increase, will he be blacklisted somewhere?
As AI is set to grow exponentially, so we must make sure that we only implement it in areas where it is least likely to cause customers to reject it, indeed some companies are refusing to implement AI in areas such as customer service until it is more like Artificial General Intelligence than simply Artificial Intelligence.