I have been thinking about software robots, yes I know I should get out more. It’s been on the horizon for quite a while but after a workshop at IBM it’s now feeling like it’s about to pass the tipping point and become a hot topic.
As a concept the software robot is a fudge, a bodge, a cut-and-shunt, a sticking plaster, steam car or whatever derogatory term you want to use for a non-ideal solution that you know will be unstable, unreliable and need constant tinkering to keep it working.
They are a bridge to transfer data between systems that have been designed to work using a human to re-key data would be much better transferred directly machine to machine using an agreed digital protocol. Emulating a human with their own login ID and deliberate delays to stop the interface falling over the software robot is an ugly and ‘stupid’ use of the machines.
However, its potentially less ugly and stupid fix than asking your human workers to do this task and often its workers with relatively high skills levels doing this critical but frustrating low skill work which makes it even worse. Teachers, social workers, medics, police and many more skilled people can end up re-keying data because the systems don’t interface.
However, in the real world being an ugly stupid process is not the best basis for a change business case.
In the real world that machine to machine link may not happen any time soon. It may not happen because you don’t own all the systems involved, it may not happen because it’s too expensive or just because it’s not high enough up the priority list. Or you may have to wait for obsolescence and replacement in a few more years. In the real world you may need that bodge.
As bodges go the software robot is not a bad one as long as you accept its limits along with its strengths. It will reliably do predictable and repetitive work, its relatively cheap, it’s relatively easy to configure and adapt and you can clone as many as you want when you get it right.
On the down side it doesn’t cope with change or variation, like the socially inept savant, it will need constant supervision by someone who understands it to deliver its potential.
In time they will get smarter and more tolerant of the mess us humans are happy to live with, but right now I think they are smart enough to deliver some big benefits.
That’s going to make for some interesting relationships in the work place as we integrate these robots into our team structures. We will have to learn when and how to deploy them and how to apply fixes to them in real time when things go wrong. In return they will take a lot of the drudgery out of our jobs and allow us to do more of the interesting value-add stuff. Used well they will improve team efficiency.
But, it won’t take these things a century to go from black Model T to Tesla Model S it’s probably going to be less than a decade before they are intelligent enough to move up the office pecking order starting to start making decisions with us and for us.
So we better make some space for them….