One of the basic principles of UX design is to assign functionality appropriately between the user and the computerized system. A classic example is mathematical calculations: computers are better at this, so we try to let computers do calculations instead of forcing the user to do them. At the other extreme, humans tend to be better than computers at certain kinds of judgment and interpretation, so we generally let humans do that.
In the grey zone between human strengths and machine strengths, designers are increasingly trying to differentiate their products by automating borderline functionalities, and the results are not always the best. For example, take my new high-definition home theatre system. The HDMI standard seems to require that when I turn on my Blu-Ray player, my receiver and TV should intelligently and automatically turn on. Oh but were this intelligent.
The reality is, the Blu-Ray turns on, and sometimes the other components follow automatically and sometimes they don’t. But there’s a lag of 5 to 15 seconds before I know what they’re in the mood to do that day. So half the time I go to turn my TV on, and discover that I’m actually turning it off. Sometimes that also seems to have the effect of turning my Blu-Ray off – I guess it figures if there’s no screen to display itself on, life isn’t worth living.
So I can rotate infinitely through an amusing Marx Brothers scenario of human-computer mis-communication – a situation that the engineers who designed this standard didn’t foresee when they patted themselves on the back for providing so much labour-saving intelligence in the user interface.
This example illustrates the difference between real intelligence and pseudo-intelligence in a user interface. In real intelligence, the interface adapts appropriately to the complexity of different contexts. In pseudo-intelligence, adaptation is rigid and incomplete.
We are increasingly able to provide pseudo-intelligence in interfaces, but real intelligence remains a difficult challenge. I think the general principle should remain the Hippocratic oath: do no harm. A pseudo-intelligent interface is worse than no intelligence at all.