Our recent Global CemPocess Conference on process optimisation in the cement industry was a bit of an eye-opener. The event started with a presentation by Jose Favilla of the IBM Industry Academy, ostensibly about ‘Industry 4.0’ in manufacturing. This concept has been heralded as the next industrial revolution, although it could also be called ‘the last industrial revolution,’ since after this one, there will be nothing left to improve.
Industry 4.0 will see the rise of cyber-physical systems, which is really just a fancier way of saying ‘very sophisticated factories.’ Data will be collected from all points in the factory (new factories might have 10,000 I/Os) and this data will be fed into increasingly sophisticated control systems that can operate the factories better than any human. This point is critical: the machines will operate the factory more efficiently than any human can, even better than the best operator in the world. When a machine can do this, why would we not hand over the controls and consign our human operators to the scrapheap of history? When a series of fuzzy-logic controls, algorithms, model predictive process control, setpoint-based controls, rule-based logic controls and, yes, artificial intelligence, can be directed at the problem of running a wallboard line at its optimum efficiency level, (with no loss in concentration, no tummy-bugs or hangovers, no holidays and no mulling-over of one’s mid-life crisis) then it’s time for us humans to turn off the lights, pack the fishing gear and go off on a permanent vacation.
Various companies are working on AI, each with their own areas of strength and weakness. At the moment, it is debatable whether there is true machine-based artificial intelligence. Machines have beaten the best chess Grandmasters in the world, have beaten the greatest player of Go (an Asian board game that is even more complicated than chess) and have even beaten humans at Jeopardy - the US TV game show where contestants are presented with answers and must guess the question. Computers can now even look at a book, recognise the words through optical character recognition, and then make sense of the words and sentences themselves, in a process of mining so-called ‘dark-data.’ Machines now have access to all the knowledge in the world, and perhaps we are lurching towards the moment of ‘singularity.’1 These are all examples of fantastic programming, fast processing, some very smart algorithms and even machine ‘learning,’ - but are they truly intelligent?
In some ways it doesn’t matter. As a reader of Global Gypsum Magazine, you are displaying your intelligence (and, may I say, your good taste). You can do wonderful things that machines struggle to do, like drive a car (oh, they can do that now), land a spaceship (er... they do that too, on a barge, in the middle of the ocean), draw, paint and compose music (ahem, computers do all of these things). Machines now pass the Turing Test - whereby we can’t tell whether we’re talking to a machine or not.
“Ah,” you might say, “but what about the flights of genius that we’re capable of, those flashes of inspiration, of intuition, those leaps of logic that humans are so good at. What about that then, eh?” Hmmm. As often as not, these logical leaps are just wrong, based on false rules-of-thumb, incorrect prejudices or ‘heuristics’ that just don’t work (as tediously laid out in the bafflingly best-selling book ‘Thinking Fast and Slow’). Some of our human decisions are not so good. For example, look at the state of the world today. It’s a mess: humans did this. Our decisions are not always so good.
Jose suggested that increasing automation will allow humans to rise up the ‘cognitive value chain’ to be able to focus on more challenging tasks. I disagree. It turns out that some of the professionals that are currently among the best paid worldwide, such as doctors, lawyers and accountants, are going to be among the first to be replaced by algorithms. The jobs that are least likely to be replaced are those ‘down’ the cognitive value chain that require empathy (such as nurses and care-workers) and those that need manual dexterity. It turns out that workers in the giant warehouses that now store all the goods that we order online are more dextrous at picking out of the storage bins the exact goods that are required than the current generation of robots. However, to get the most out of these human ‘operatives’ a computer-based algorithm tells them exactly what to pick and in exactly what order to pick it in.
In the same way, the robot brain that will soon be operating your wallboard plant will require human operatives to do its bidding when it comes to finickity maintenance tasks, like tightening bolts or changing a fuse. Make no mistake though, it will be the computer that will decide if and when that needs to be done, and it will send out instructions for humans to go and do its bidding. Yes, it’s best to get ready to obey our robot overlords.
1 https://en.wikipedia.org/wiki/Technological_singularity
