First session this morning is a Maverick session, not recognised Gartner research, but a personal view. "Judgement day, letting machines run the world." How far can we use machines to make decisions?
We're moving from automating the simple, to automating the complex. In the next 40 years there will be big changes to what machines can do.
Why do machines make better business decisions than humans? When we make business decisions we think we're acting rationally. But, it's anything but rational. Lot of behavioural factors affecting decision making, we're very influenced by behavioural and cognitive biases. Good example, why is there always a €250 bottle of wine on wine list, so you spend more on medium wine so you don't look like a cheapskate.
Also, so many pressures on decision making, we find it difficult to cope with them. Very volatile economic climate, information volumes exploding,. We're now creating as much information in 2 days as we did up to 2003. Impact of social media, pervasive mobile computing, loss of control. Traditional decision making methods can't cope.
Motivation systems can force bad decisions. Eg in a recent fatal airline crash, the pilot ignored 15 automated warnings to go round again. Turned out pilots were being given bonus if they saved fuel and this was a contributory factor. Pay for performance incentives work best for mechanical tasks. For cognitive tasks, monetary bonuses can result in worse performance.
Business decisions are all about predicting the future, choosing between different future outcomes. Machines win in all areas. Machine-based algorithms beat humans on clinical diagnosis, hiring staff, predicting the wine harvest....
Predictive modelling is big business now, and growing. Machine based models can ask the questions that no-one dare ask and challenge perceived wisdom. Talent management algorithms can be used to hire staff, it lowers turnover, improves retention and has better employee effectiveness.
Machine learning is well established and growing in sophistication. Combination of statistics and brain models. Cognitive chips launched by IBM in August 2011, a combination of programmable and learning synapses.
Most big business decisions are made by a small team of people with little transparency. Quality of decision making today is not good. Research shows that the more people are involved, the better the decision. Collaborative decision making is a technology that allows people to be involved in decisions, brings greater transparency, and improves the decision making process.
IBMs Watson computer, massively parallel probabilistic evidence based architecture, developed to win a game show, Jeopardy in the US. They give you the answer, you have to give the question. It beat two of the champions. and you can see videos on YouTube. It worked by looking for patterns in masses of data, not thinking like we do. Often humans got the answer, but Watson got there quicker.
Will see a new generation of modelling tools over next few years to analyse masses of complex data.
Machines now can do more than we ever envisaged. We are just at the start of this new wave of technologies. Machines won't replicate what we do, but they imitate what we do.
Today, we are in a phase of augmented decision making. Humans plus computer agents are better predictors than either on their own. Already in use in supply chain planning and insurance underwriting. Algorithm and model design will become the must have skills during the next 10 years.
Within 10 years we will be able to hand over most of running a business to machines and they'll do it better than us. Computers will control operational decision making. Humans will focus more on strategic decisions, innovation and risk management.
In the far future, systems that run corporate entities will be linked together, Skynet based economic system! They'll share data and make rational decisions about how resources should be allocated, potentially optimising the global economy.
Some impacts? Humans may lose key skills, pilots already forgetting how to fly planes because they rely on computers so much. It may stifle real innovation, and too much automation could cause a system that's too big to fail, or cause massive shocks to the global economy if the models are wrong