Treasure Data's VP of marketing Kiyoto Tamura envisions AI moving from very specific, mundane operations to much broader—and more exciting—applications.
“In the past, it was more like, ‘Find the optimal route for package delivery … or the most relevant websites for a search query.’ Now, we are starting to see, ‘Play a game of Go really well; drive a car safely,’ etc. All of this is cool, but humans still need to feed objective functions to the computer, and at least for now, this is going to be the case.”
Data scientists, machine learning researchers, and computational linguists are increasingly sought out, says MindMeld CEO Tim Tuttle. He cites a VentureScanner study that counted 910 AI companies emerging from March to October 2016, more than half of which focus on deep learning/machine learning and natural language processing.
“Not only do these categories win in numbers, but they’ve also received the most funding, to the tune of $4.5 billion,” Tuttle says. “With the recent explosion of interest in conversational applications, there has been a mismatch between supply and demand. As a result, subject-matter experts will remain a valuable commodity until academia and industry can rebalance the equation.”
A form of artificial intelligence, machine learning can take massive amounts of data to very quickly find patterns—like facial recognition—and solve problems, like recommending a movie to stream, without being explicitly programmed to do so.
“Cognitive technologies, aided by bots and machine learning, will start to add value as organizations strive to find the ‘signals in the noise,’” says Patrick Spedding, senior director of BI R&D for Rocket Software. “Machine learning is, after all, based on mature analytics capabilities—formerly known as ‘data mining’—which really have been waiting for a suitable platform to become more ‘consumable.’”
How should developers who want to expand into machine learning develop skills in this area?
Abrams, of Seven Peaks Ventures, points to a highly regarded online class: “Andrew Ng's seminal course on machine learning on Coursera is a great example. Students who took his course via Coursera actually did better in Kaggle competitions than some longtime practitioners.”
Not every developer working in machine learning comes from a computer science background, though it’s helpful, says Solvvy CTO and co-founder Mehdi Samadi, who sees some Ph.D.s without CS degrees being recruited and trained to become machine learning engineers.
“Core contributions in the field of machine learning require running a lot of experiments using the real data, observing from the result of the model, and improving the model,” he says. “Having a CS degree or core engineering background usually would benefit the engineers to be more successful in their job in order to be able to continuously run experiments and improve machine learning models.”
Sign up for CIO Asia eNewsletters.