AI centred on deep learning
Deep learning, which is an extension of machine learning (ML), is a major driver of artificial intelligence (AI).
So said Alexander Linden, research VP at Gartner, commenting on the key trends in AI and ML during a roundtable discussion at the Gartner Symposium/ITxpo in Cape Town.
Deep learning delivers superior data fusion capabilities over other ML approaches, and Gartner predicts that, by 2019, it will be a critical driver for best-in-class performance for demand, fraud and failure predictions.
Linden labelled AI an "awesome innovation", pointing out that although it has become important for the industry, referring to it as artificial intelligence is a bit of a stretch.
AI boils down to certain types of technologies that have been discussed for at least 20 years and when digging deeper into more research, you will see that in the early to mid-90s the theme of data mining came up.
Data mining was quite popular and if looking at all the technologies that were being discussed around data mining, you will see it is the same as predictive analytics and advanced analytics.
Five years ago, big data was the big buzz, and was split into infrastructure and data science. Increasingly, people started to talk about machine learning and then about four years ago a new breed called deep learning developed, he said.
These days, Linden argues that AI is deep learning. "By a far margin, most of the innovations that are really awesome right now are centred around deep learning."
He explained: "AI is a much more different level concept. AI is made up of certain capabilities that need to be awesome because if they are not then we wouldn't call them artificial intelligence. They are awesome because they can solve complex issues that were only able to be solved by humans recently.
"Deep learning can, for example, give promising results when interpreting medical images in order to diagnose cancer early. It can also help improve the sight of visually impaired people, control self-driving vehicles, or recognise and understand a specific person's speech."
Dispelling AI myths
Similar to sentiments shared by his counterparts this week, Linden highlighted some of the misconceptions and anxieties posed by intelligent machines.
If people believe technology can think like humans, then they will soon think the technology is much better than humans, will take people's jobs, win Nobel and Pulitzer prizes, and kill everyone.
This couldn't be further from the truth, he said. "I'm 100% sure and have no single doubt in my mind that this won't happen.
"We don't want technology to think like a human, we want technology to be even better than humans, not complain, not get tired, consistent, more adaptable and fast."
Linden also pointed out that using ML and AI to add value to a business is complicated. "Don't deliberately meet all ML prerequisites exactly; instead find the right problem to solve.
"It is a good idea to start ML by using the same data you use in your popular reports, such as orders by a region. Then you can apply ML to make forward-looking predictions; for example, a forecast for the same orders by a region for the next month. This way it extends on the after-the-fact reports to show business stakeholders the art of the possible with ML.
"What's hard for people is easy for ML, and what's hard for ML is easy for people," he concluded.