Artificial intelligence (AI) – teaching machines to think and learn – is either the magic solution that fixes all our problems, or it destroys us. That’s the binary discussion taking place in tech circles about the learning algorithms and computer software that are changing our world.
Engineering wizard Elon Musk sees a future of thinking and killing robots if we don’t constrain humanity’s AI ambitions. At best, we become “house cats” to machines; at worst, the human race is expendable. Either way, AI is something to fear and watch very, very closely. Facebook’s Mark Zuckerberg, on the other hand, says Musk is a doomsayer, and AI in the form of self-driving cars and the like will save lives and make the world a better place.
These impassioned discussions seem far removed from daily South African life and business, but a new report produced by Accenture and the Gordon Institute of Business Science (GIBS) says that AI is already here. The report urges businesses to jump on the AI bandwagon so that they do not get left behind in this technology boom.
Rory Moore, innovation lead at Accenture and a co-author of the report, says: “AI is not something you have a polite, civilised conversation about: people start shouting at each other.”
But Moore, a strong proponent of AI, says: “Every few years, you have the opportunity to change the trajectory of a country or economy. Whether we like it or not, it’s coming. We can either put our heads in the sand and hope it goes away, or embrace it.”
As is usually the case with any polarised technology conversation, both sides ignore the grey area. AI is a vast umbrella encompassing a broad range of technology and applications, something which is canvassed in the GIBS-Accenture report, not simply an omnipotent robot with vengeance and blood-lust melting its circuits.
What is AI?
AI encompasses computer software that tries to perform tasks usually reserved for humans. These include sensing, understanding, optimised decision-making, pattern and language recognition, learning, and many other cognitive tasks that we take for granted.
In their report, AI: Is South Africa ready?, the authors detail five tangible forms this technology can take: virtual agents, speech analytics, identity analytics, recommendation systems, and cognitive robots.
Virtual agents or online chatbots are able to engage with customers in the place of call centre and sales support staff. Speech analytics includes software that recognises speech patterns, emotions in voices, allows computers to “understand” what people are saying. Identity analytics, on the other hand, are more about security: they allow people to protect their data and control access to systems. Recommendation systems – which are already at play when Google optimises your search results, Amazon suggests books you’d like, or Facebook tailors advertising to you – focus on content targeting. Then there’s also cognitive robots, in which robots learn from their experiences, the environment and on their own. This is the application that gets most people hot under the collar: for Musk, these learning robots become killers, for Zuckerberg, they are driverless cars.
For example, Cape Town-based start-up Clevva, which was established in 2011, aims to help its clients improve decision-making.
“[It] is looking to be the global leader in compliant decision and action navigation,” says co-CEO Ryan Falkenberg. What this means is that they “effectively [offer] staff a decision-making GPS that will ensure, irrespective of context, they consistently ask the right questions, consider the right information, make the right decisions and take the right actions, in line with prescribed rules”.
But AI will influence most industries. “Every industry where human decision-making forms a large part of the value chain, AI will disrupt,” says Falkenberg. “It’s not just where repetitive administrative decisions are being made; it includes autonomous vehicles (impacting professional drivers), medical practitioners, lawyers etc. – everyone who currently extracts value from making complex decisions based on pre-defined rules.”
These AI technologies are already seeping into businesses and the world’s social fabric.
Andrew Ng, a computer scientist and a co-founder of online learning platform Coursera, infamously said: “Just as electricity transformed almost everything 100 years ago, today I actually have a hard time thinking of an industry that I don’t think AI will transform in the next several years.”
Jeff Chen, a senior lecturer at GIBS and co-author on the report, is in neither the Musk nor the Zuckerberg camp, but says that we need to be realistic about AI’s capabilities and consequences. “Any technology that makes things more effective and efficient eventually takes off, whether we want it to or not,” he says.
It is estimated that the AI market will be worth $35-billion and double annual economic growth rates, the AI report notes.
But AI itself is nothing new: “AI is not a new term, it’s been around for a long time,” says Chen, “but it’s just that now the technology has become more sophisticated. I think AI can help us in many areas, but with casualties.”
Jobs will be lost...
“We need to understand the other side of AI, regardless of how we romanticise it: somewhere down the line, someone is going to lose jobs and we need to understand the social economic repercussions of that,” Chen says.
Technological leaps often come at the expense of current jobs and livelihoods: whether it was the combine harvester, washing machines, or mechanised drilling, technology and machines replaced people. While those technologies ultimately resulted in more productivity and greater employment overall, in the decades following their introduction there was a spike in unemployment.
Clevva's Falkenberg says that while many routine jobs will disappear as soon as a specific AI technology is into that industry, an option for other jobs is to “migrate people to new forms of work by offering them augmented AI”. “It’s like first offering drivers a GPS to help them drive better, and then ultimately replacing them from driving with autonomous vehicles.
“Our preference is, in as many jobs as possible, to start giving people augmented AI before simply replacing them altogether. That said, many routine-based jobs will simply disappear once AI gets more mainstream. And as a result, many people throughout Africa will find themselves even more removed from the formal job market.”
Currently, South Africa’s official employment is at 27.7%, which excludes those who have given up looking for a job. Unofficial estimates reckon that unemployment is significantly higher at more than 40%.
...and jobs will be gained
“There is no doubt that AI will generate a swathe of jobs that we do not yet have names for and that new skills in areas like robotics and pattern recognition will flourish,” says the report. “For example, just a few years ago there was no job like an app developer.
“But it is also true that AI will eliminate other jobs, potentially worsen inequality and erode incomes for some parts of the population. The imperative – starting now – is for policymakers to proactively address and pre-empt the downsides of AI.
“For example, they must identify the groups that are at risk of being affected disproportionately by job displacement and create strategies that focus on reintegrating them into the AI-driven economy,” says the report.
Accenture’s Moore expects that in two years’ time, all service-based sites – like those of banks or other service providers – will have chatbots, rather than people dealing with customers’ queries. “Just think: I’m a bank, and my call centre is flooded with people who have forgotten their pin. A bot reduces traffic, and I can redeploy those people.
“Our stance is not to retrench, but redeploy. Business can become more competitive, can [ultimately] employ more people, but not in the same job,” he says.
This idea needs to be pushed within companies, Moore says. Executives need to choose to adopt this stance for the better of the country, as well as at a national policy level, with government offering incentives to companies that choose to reskill workers rather than retrench.
But this boils down to policy and regulation. South Africa's authorities have not covered themselves in glory when it comes to legislating to include new, disruptive technology, with Uber being the case in point. The e-hailing service, through which a passenger can order a taxi via an app and pay automatically with their credit card, has exploded into the South African transport space. The American technology company has been in the country since 2013, but four years later there is still confusion about the regulations and licensing process of these e-taxis. Also, metered taxi drivers have responded to the competition from the cheaper service with violence, with metered taxi drivers saying that Uber is taking away their livelihoods.
“We need a more enlightened approach from regulation,” says Moore, noting that over regulation can also stifle budding technologies.
He maintains that the only way that government can regulate AI, which is such a fast-moving space, is through a strong lead from business.
“There’s no way government can cope – it moves too quickly for regulation to work. Business needs to guide government,” he says.
“Stuff is happening, and South Africa needs to get ready to compete. It is being held back by a cumbersome regulatory framework.”
However, the report recognises a number of obstacles ahead for AI in the country. “South Africa is facing myriad structural and cultural hurdles as well as social dilemmas, which may hamper or delay businesses and governments from fully integrating AI technologies into the economy, ultimately impacting the potential upside to growth and competitiveness that AI presents,” the authors write.
No magic wand
But while technological disruptors such as AI can only flourish in an enabling environment, simply sticking AI into a company does not mean that the company will flourish.
“In the end, it’s never, ever, ever about the technology. It’s about how cleverly you use the technology,” says Chen. “AI can only be powerful if it is fed the right kind of data, it gives you an answer based on the information it gathers.”
This helps to level the playing field between larger and smaller companies: “For the company with fewer resources, it doesn’t mean they can’t compete. They just have to compete smarter. AI isn’t your determinant. It’s your strategy.”
Asked whether he thought AI would doom or save humanity, Chen says: “I’m not in the Elon Musk or the Zuckerberg camp, but I think AI is something important. If you don’t use AI, another company will and it will give them a competitive edge.”
But that competitive edge hinges on having a strategy: “A company must have a plan, and if they don’t have a plan for how to utilise this technology – not only immediately, but also long term – they’ll be left behind,” he says.
“At best, we become ‘house cats’ to machines; at worst, the human race is expendable”
“...AI can help us in many areas, but with casualties”
“We need a more enlightened approach from regulation”