We are in the midst of the Fourth Industrial Revolution where one of the main driving factors is artificial intelligence (AI). AI will unleash another step-change increase in productivity and reshape our workforce, similar to the industrial revolution.
In one estimate, Accenture forecasts that AI could double the economic growth rates of 12 major countries by 2035. These gains will come from computers doing what they have always been good at — freeing us to make better use of our time.
For instance, AI makes it possible to automate routine tasks such as responding to simple customer questions. It can also be used to spot changes in customer preferences, identify visual problems in manufacturing, further automate agriculture, identify fraud, and better inform business decision-making. But what is AI and how is it going to transform your business?
While those are complex questions with no simple answer, you can get a long way toward answering them in the context of your organization by focusing on three concepts: recent advances in computer programming, the importance of quality data, and how AI will change organizational workflows.
Understanding algorithms and deep learning
Computer programmers create algorithms to tell computers how to complete tasks and solve problems. The difference between AI systems and earlier generations of computing is that they can learn as they go, then devise their own solutions to problems. This is made possible by deep-learning algorithms, which review large amounts of information and can better handle messy or “unstructured” data, including human language.
These capabilities make AI systems useful for completing tasks that would have been impossible even a few years ago. For example, computers couldn’t distinguish between a cat and a dog until 2012, when they had been taught enough about the differences between the two species and programmers came up with better algorithms.
Another example is using AI to analyze sentiment, such as what your customers think based on their social media posts. Sentiment analysis searches language to pinpoint the opinions of your customers on, say, Twitter and estimates the emotional reactions shown toward a product. This could allow you to see if a marketing campaign was working in real time.
However, this analysis is hard to perfect due to language changing all the time. If we all used clean, plain English with no accents, we could get AI working relatively well, and soon. But teenagers say things like “YOLO,” which AI algorithms must try to understand. In practice, people will keep making stuff up and AI will have to continuously adapt.
Another problem to keep in mind with algorithms is bias. The way algorithms solve problems can reflect the biases of the programmers or organizations that wrote them. In the era of AI, the effects of these biases can compound as the algorithms themselves continue to learn and develop their own approaches. This has already been seen in examples such as social media bots that turn racist over time.
The message here is that if you’re going to build your company’s future on AI, you will be building it on algorithms with certain ways of solving problems and learning. In turn, you need a good understanding of how the machine is operating — and who made it.
The importance of data and new workflows
Another key point to understand about AI is that it is only as good as the data — specifically, the “training data” — you give it. Like any computing system, if you put garbage in, you will get garbage out. And just because a system is artificially intelligent, it doesn’t mean it will be self-healing if it is given the wrong information to work with.
One of AI’s most powerful capabilities is being able to identify items or patterns and then apply that understanding to future activities. To enable an AI solution to do this, you need to feed it a large amount of data describing past events and teach it the attributes of those events. This requires the labor-intensive process of establishing a large body of high-quality data, then typically labeling items as positive, negative, or neutral. Teaching computers to recognize cats, as above, involved labeling lots of images of cats as positive.
To get AI to take over a business process, you will typically need to feed it about two years’ worth of training data. This can be made easier by using services such as Amazon Mechanical Turk, paying it to get people to label data to teach your AI system.
An example might be having a machine take over the task of replying to customer queries. To achieve this, you would collate years’ worth of common queries and the replies your organization sent. The AI will become able to select — or even construct — replies to queries based on past patterns. And it will be able to do this more flexibly than a traditional computing system because it can better cope with natural language and novel situations.
The better you train your AI, the more reliable and responsible it will be. To maintain customers’ trust in your new AI-powered processes, you should also make sure their data is being held very securely on your premises or in the cloud.
Your next move
AI is a complex field and I am the first to say that we computer scientists have not progressed as far as many people believe. For instance, we currently have no credible research path to any kind of conscious AI algorithm and there are no robots that are truly autonomous or able to make their own decisions — so don’t worry about walking terminators.
However, it’s also clear that AI is a very significant evolution in computing and that it will underpin many of the future advances — and leaps — in business productivity. For this reason, I would strongly urge all business and government leaders to actively explore AI. Learn what it is, how it works, what it can do, what it can’t do, and where it’s headed. Then ask what all of that means for your organization, your use of technology, and your business model.
Like early PCs and the Internet, it might not look like much now, but there is no doubt that AI has the potential to create enormous change not only in business but also society.