I think, therefore I aim: The path to artificial intelligence, and its true value for business

October 11, 2019

The world around us is growing smarter – if we are to believe all the hype about smart kettles, smart houses, smart cities, applications and AI-driven cloud services, says Renewtrak.

And yet there is still a lot of very un-smart things happening – including business systems wrapped in so many layers of security that they no longer justify the label “solution”. Are we too stupid to tell the difference between real and artificial intelligence?

For decades developers have been challenging AI to learn simple actions and play complex games. In 1997 for the first time a reigning world chess champion, Garry Kasparov, was defeated by a chess computer developed by IBM. “Deep Blue” had been trained to recognise winning situations and to look ahead at all possible moves leading towards a winning position. Twenty years later, in 2017, Google demonstrated a program called AlphaGo Zero that instead taught itself to play Go – a game that is strategically far more complicated than chess – rather than needing to be programmed to recognise winning positions.

Real life business decisions might be limited by certain rules, but they are seldom played in a rigid environment – market conditions, customer expectations, economic and regulatory factors all lie outside control and can change unexpectedly. The number of players involved might be limited, but those players can be affected by family problems, illness, and swings of mood and loyalty. If AI is to deliver broader business benefits, it will need a lot more of both types of learning: being taught the rules of the game, as well as being given a whole load of material to study and teach itself strategy.
 
Automation provides the answer. The greatest business value of automation is not so much in saving labour as in saving the sort of boring, repetitive and yet complicated labour that humans tend to do worst – either growing bored and rebellious or else performing the actions in a trance state where human errors multiply and often lead to long term problems

So automated monitoring, as in thousands of connected devices on the Internet of Things (IoT), or traffic monitoring embedded in networking components, can harvest more real world, real time data than could ever be managed by human agency alone. Add an element of machine learning, and you are on the path to developing that higher-level intelligence that goes beyond taught skills to actually having something to teach us.

Do companies offer real AI? Strictly the answer today is that they do not. But with the automation and learning ability of their solutions, they are paving the way to ever-growing intelligence. Companies that adopt the best of today’s automated, smart solutions are indeed taking a significant step along the path towards real AI.

Leave a Reply