AI: Diving Into The Deep End?
Aeroplane engineers who work on landing gears manufactured by GE Aviation Systems have a glimpse into the future. The company’s Predix operating system is one of the best-known examples of artificial intelligence (AI) in industry, using machine learning to process historical and real-time data and deduce the best time to bring a part in for maintenance. Predix is also used widely in oil and gas infrastructure, cutting costs by optimising schedules and downtime for routine tasks.
AI is slowly being integrated into workflows of firms all over the world. Around 20% of attendees at a recent MIT conference, EmTechDigital, claimed that their firm already has an enterprise AI strategy showing success. Depending on who you talk to, South African companies are doing OK, or have quite a way to go, when it comes to integrating AI into their business processes today. There are pockets of excellence amid slightly patchy take-up of business intelligence, automation, and other ‘AI-lite’ type capabilities.
“Our market has remained relatively protected, [so] the urgency often is less because competition is not like in the US where companies are completely wiped out because they didn’t transform,” says Rapelang Rabana, chief digital officer at ICT services business, BCX.
“I don’t think South Africa is doing particularly badly, especially compared to other emerging markets, and there are pockets where we are ahead of the developed world. But to be more globally competitive and to outpace our global competitors, we need to do a lot more to catch up.”
The risk is that the gap between the AI ‘haves and have-nots’ will widen. As early adopters start seeing the business benefits of deploying AI in their organisations, their advances could be exponential and the laggards are never going to catch up. Worse, at the same MIT conference, some 80% of attendees said that a lack of skills was one of the biggest challenges to adopting an enterprise AI programme. Those who don’t move quickly will miss out on the best talent.
The catch-22, however, is that for established organisations with existing business processes and customers, there is no such thing as a fast track. They need to do their time getting the fundamentals right before attempting the deep end.
Take automation, which, although it brings immediate benefits, does not yet optimise underlying processes.
“We are still doing things the way humans do it, but using software,” says Bruce Watson, head of the Information Science department at the University of Stellenbosch. “For instance, opening a bank account online still follows the old procedures.”
“You essentially go through the same steps as you would if you walked into a bank branch. What we really should be thinking of is completely new ways of running a business. Can we come up with completely new processes for signing up new customers? There is a very strong value proposition here.”
Graeme Welton, director of Flow Software, a company that helps the manufacturing industry create data-rich dashboards for decision-making, agrees that there is no point using AI if you are still collecting company data in Excel.
“You have to walk the steps,” he says. “You have to put the basic building blocks in place and fix the foundation before you put another layer on top.”
As a bridging strategy, technology company Clevva suggests something it calls intelligence augmentation (IA), which allows companies to perfect back-end logic and intelligence before tackling the front-end experience. For instance, instead of replacing call centre staff with AI-powered chatbots from the get-go, rather support them with digital intelligence.
“This gives you room to learn and make mistakes because your staff can step in when your digital logic is found wanting,” says Ryan Falkenberg, Clevva co-CEO. “It means approaching AI from a both/and position, not an either/or one. In other words, including, not excluding, staff in your digital mix. Only once you have perfected your digital logic do you then look to adopt purer forms of digital autonomy.”
Meanwhile, humans get to polish their emotional quotient, which is something we still beat the machines at, and learn to work hand-in-microchip with the robots. This idea of humans and robots working side-by-side is reinforced by retailer TFG’s approach to its entire digitalisation strategy. At board level, the digital transformation champions are group CIO Brent Curry and group retail director Shani Naidoo, who has a master’s degree in industrial psychology. This combination of right- and left-brain thinking “keeps everyone in check” says Curry, and in particular has affected how new technologies are communicated and rolled out to the rest of the organisation, which ultimately drives uptake with customers, and, therefore, success or failure.
Curry describes his approach as conservative, for instance when piloting AI-powered facial recognition software in TFG stores. For now, the software is being used to measure sentiment – whether someone in a queue looks disgruntled, for instance – and gather anonymised demographic data. But in future it could, with permission, capture biometric information and contribute to TFG’s wider customer experience play, which includes hyper-personalisation across channels. This all ultimately feeds into TFG’s vision of what an increasingly disrupted retail industry will look like in 2025.