How Is Artificial Intelligence Developing In The Enterprise?
Artificial Intelligence (AI) has moved from science fiction into the enterprise in recent years. Many companies are using AI systems today to intelligently analyse large volumes of changing data and to notice or predict patterns. Typical business uses today include examples such as:
- Rail operators predicting train delays before they happen because the AI system can extrapolate from small delays to predict the impact on the entire network.
- Customer service agents being advised on how to help customers by systems that know the answer to every question a customer has asked in the past.
- Alexa knows how to answer your question because it immediately processes your voice and determines what you are asking before creating an answer.
- Netflix knows which movie you might want to watch because they know your past behaviour and how similar customers have also behaved.
AI really is all around us today, in the enterprise and as consumers of services. In the present environment it would now be unusual for any company to not be exploring how AI can improve their business.
But AI does have one a fundamental flaw, it is always limited to working on a very specific problem. This means that you can have a very complex system that knows everything that your customer may ask when they call for help or the system may understand how to play chess or Go, but these individual tasks are all that it can do. There is no inherent awareness of the environment around the system – although we use the term intelligence, it’s not really aware or sentient. An AI system that can play Go cannot plan the best route on a map.
This means that the system can only solve the problems it was designed for. Some might argue that this is a benefit, because it means that however good our AI systems get, they never move into the realm of awareness and all the problems that a conscious system might create.
A recent experiment by IBM has demonstrated that AI is developing rapidly though. They demonstrated how an AI system could be asked a random question and it would then have the ability to debate that subject. For example, in the video clip that I watched the system was asked if pre-school facilities should be subsidised by the government. It gave a response, arguing why subsidies are useful for 4 minutes.
This system has been pre-loaded with information on millions of subjects and objects. It’s stuffed full of encyclopedia content and research. But even with all this data it is quite an achievement to turn that into information and then a coherent argument.
Essentially this system is starting to show that perhaps an Artificial General Intelligence might be possible. It would need to be pre-loaded with an enormous amount of general data, and then would need a Machine Learning system to continue learning, but it is looking more feasible than even a year or two ago when Elon Musk started warning that we are heading for an ‘AI apocalypse’ because the machines will eventually have more intelligence than the humans.
I don’t think we will be seeing many business case studies featuring general intelligence just yet, but AI in the form we already know it will certainly be more important. AI is offering companies a chance to identify patterns and trends they could never see manually and this will be a strong source of competitive advantage in the next few years.