18 Jan 2017

The Challenges of Machine-to-Human Communication

Humanity stands at an odd junction in its relationship with machines. We’re sophisticated enough to have machines that talk to us as they check out our groceries, but jaded enough that these machines annoy us.

Empty lifts inform non-existent passengers that they are ascending; we’re naggingly told that the train doors are opening even though we can see that perfectly well for ourselves; self-checkout tills announce notes are dispensed below the scanner even when only coins are due in change.

Machines are slower than us, grammatically incorrect and annoy us with unnecessary information.

Linguistics Professor Geoff Pullum pinpointed some of these frustrations in an article for the Chronicle of Higher Education.

At the supermarket self-checkout machine, I press “Pay with cash.” The electronic voice says: “Insert cash or select payment type!” (I already selected payment type.) I feed in a £20 note, and again it says: “Insert cash or select payment type!” (No condition saying “unless cash already inserted.”) The change due is less than the smallest banknote (£5, roughly $7.70), hence must be all coins; yet after “Please take your change!” the voice adds: “Notes are dispensed below the scanner!” (No condition meaning “if change due ≥ £5.”)

The problem seems to be that user experience just isn’t taken into account when these machines are designed. Simple testing would quickly reveal these issues before release yet not only are they not corrected but they are cloned across multiple versions of the same machine, spreading the frustration to more users.

Professor Pullum blames thoughtless programming for these redundancies. His argument is that senior managers don’t care enough about how the machines communicate with their users to catch these irritating errors.

It may take very little effort for your answer machine to announce you have ‘one new message’ rather than ‘one new messages’ but someone has to catch this and care whether it is fixed.

Why machine communication matters

Although the frustrations of machine-to-human communication aren’t confined to self-checkout machines, it’s worthwhile focusing on this particular element of human-machine interaction. One of the reasons Tesco’s Fresh & Easy grocery store venture failed to take off in the US was that US customers failed to warm to the concept of self-checkout.

Whilst it’s possible that even the best user experience at machine checkout may still have been rejected by US customers, a better user experience may have helped shoppers accept the novel concept.

From a business perspective, there are huge benefits associated with persuading users towards machine self-service.

McDonald’s fast food chain found that customer who used the self-service kiosks were more likely to ‘supersize’ their meals (ie opt for the bigger serving size) and spent an average of 30% more on their order. It’s thought this is because they didn’t feel as judged when making this decision as they did when ordering from a human server.

Getting it right matters, and can make a big difference to the bottom line. If each self-service kiosk increases order value, then higher user acceptance (ie more customers willing to use the terminals) has positive effects on profitability. It’s a strong argument for ensuring this user experience is as good as it can be within the confines of the technology.

One answer is that machines are only as good as the people that program them. But a more in-depth version of this answer is that the people that manage the people that program them, and the customers that buy the machines, need to have higher requirements for the standard of communication on offer.

The importance of user testing

The next challenge for the machine designers is to take user experience far more into account when building the next generation of communicative machine technology. This includes building in better logic so that self-checkout machines don’t tell you where the notes come out when you’re only getting coins in your change. It’s a way to make the process less confusing and more human.

As the potential uses of artificial intelligence unfold and expand, it’s likely that we’re going to be spending more time in the company of these communicative machines.

This means more interactions and potentially far more frustrations. That’s why it’s so important to get the user experience right.

With user experience now being increasingly valued as a discipline in its own right when it comes to online tools such as apps and websites, it’s time UX became part of the testing process when machine-to-human communication devices are being built.

Those working in the field could learn much from the process of building for the web: with user testing now such an important part of any quality web build the discipline could contribute a lot to the field of machine communication.


Sign up to our newsletter

Get our blog articles straight to your inbox.