Depending on the skill and experience of the programmers and system designers, you may get serviced by a soulless robot with an algorithm designed to see if you are committing fraud, a robot customer service AI which will ask you endless questions to see if you meet their byzantine requirements for a low risk customer, or possibly a well-designed robot system that provides immediate access to a human being in case of trouble.
Which of these three AI systems would you prefer to deal with in your business transactions?
If you choose AI system number 3, you can summon live help when a problem occurs. If you have the misfortune to deal with either of the other two choices, you are dealing with machine logic designed by an inexperienced programmer and when things go wrong you will put your business at risk because the company has decided to save money on salaries and you will not get human help without threats of litigation or public exposure via the internet.
There is a growing awareness of the need for human intervention in AI systems, but you may still encounter a system powered by AIs in the first two categories. If that happens, you should add your voice to the growing number of people writing about customer support problems with AI.
This is what the best AI designers are working toward:
Provide paths forward from failure. The trick isn’t to avoid failure, but to find it and make it just as user-centered as the rest of your product. No matter how hard you work to ensure a well-functioning system, AI is probabilistic by nature, and like all systems, will fail at some point. When this happens, the product needs to provide ways for the user to continue their task and to help the AI improve.