This article first appeared in The Irish Times, on November 23 2017.
Knowing how to apply machine learning confounds many business owners
Artificial Intelligence, or AI, refers to any kind of machine-based learning developed to perform various tasks and functions that, heretofore, was the responsibility of humans.
While AI has been around for decades in various guises, the technology has come on leaps and bounds in the past 10 years, providing countless ways to make any business operate more efficiently with less manpower.
The technology can be – and, in many cases, is already being – exploited by many different firms to achieve greater efficiencies and higher output. “Many large companies are using it in their businesses now,” explains Tom Davenport, Professor of Information Technology and Management at Babson College and Co-Author of Only Humans Need Apply: Winners and Losers in the Age of Smart Machines.
And yet it remains an anomaly in terms of its integration into everyday business practices. Specialist expertise is required in house to build solutions powered by machine-based learning. An organisation needs a development team consisting of statisticians, computer engineers and business savvy experts combined. But most SMEs don’t necessarily have that mix of expertise readily available to them - placing AI beyond their capabilities.
The Top 3 AI applications for business:
According to Davenport, there are three major types of applications:
1) The first application he refers to as “robotic process automation’, which automates structured, back-office processes involving interaction with multiple computer systems. “It’s not the smartest technology but it replicates a human doing low-level information processing,” he says. Tasks such as accessing data, transferring it into other systems and reading and responding to emails all come under this application. “It is easy to implement and typically has high ROI,” he adds.
2) The second is known as “cognitive insights applications” where machine learning is used to generate more models with higher granularity that learn from data over time. They are typically used for marketing and sales purposes (e.g. propensity models that predict which customer will be most likely to buy a certain product or service), detecting fraudulent transactions, and combining data on similar customer records across different databases.
3) Finally, “cognitive engagement” involves the use of smart machines to interact with customers or employees using natural language. “Sometimes these are used for customer service applications, but they are also used in HR or IT departments to answer employee questions.”
Of course, some industries will be more likely to exploit AI tech than others. “The major industries currently impacted by AI are healthcare, manufacturing, transportation, customer service and finance,” explains Shourjya Sanyal, a Digital Skills data science lecturer.
Sanyal sees two possible ways in which SMEs and mid-sized businesses can address the need for AI. “One is to hire engineers who will work closely with the customer discovery team and use a SAAS application like IBM Watson, AWS Deep Learning AMI or Google. ai to implement a working solution,” he says. “The other possibility is to bring a great AI engineer onto your team. The AI Engineer will design, develop and test hypotheses tuned to your needs in an iterative way (using programming languages like Python or R), to deliver the maximum value.”
The most valuable digital currency on the market: data
AI is hugely important to data analytics. “Online businesses like Google and Facebook were the earliest adopters,” states Davenport. “The financial services sector has also been a big user. Telecom is growing rapidly. All of these industries have a lot of data on consumers. It’s somewhat more challenging to use the technologies when you have businesses as customers.”
But data consumer giants aren’t the only users. There are already a growing number of Irish SMEs ahead of the pack. Data science expert Shourjya Sanyal is also CEO of smart medical-grade wearable technology manufacturer, Think Biosolution, which is building AI-powered wearable fitness trackers that can tell users how to optimise their exercise to build endurance. “Other Irish SMEs are using AI in everything from voice-enabled apps and services for children to searching for brand logos,” he says. “Nuritas is using AI to redefine biology as we know it. Artomatix is using AI to give users a great visual experience. Soapbox Labs is using AI to power voice-enabled apps and services for children. LogoGrab is using AI to search for brand logos on the internet. The list goes on.”
A ‘race to the bottom’?
Of course, too much automation comes at a price. “I refer to pure automation as a ‘race to the bottom’ from a strategy standpoint,” says Davenport. “It lowers your costs but also eventually those of your competitors, and eventually lowers margins. It can also be more difficult to innovate if many processes are automated. Automated customer services have rarely been of high quality.”
Government authorities and regulators are already on a steep learning curve, but a balance will need to be found between exploiting the potential of this new technology whilst also remaining cognisant of the potential negative effects it might have. “There is always the risk of data security that companies, big and small, must be prepared to address,” warns Sanyal. “There is also the challenge of monitoring unexpected system behaviours and how that affects end users. Of course, automation will impact the creation of new jobs. So companies and government must work together to train the next generation towards optimising human potential in an automated ecosystem.”
