Godwin Emefiele’s term as governor of the Central Bank of Nigeria (CBN) officially ends on 2 June. His dual exchange rate structure, which the IMF has long urged Nigeria to scrap, may not last much longer.
Artificial Intelligence hits African companies
It follows that Africa, with its current economic worries, could do with a productivity bump. But the adoption of new technology anywhere in the world is not a foregone conclusion. The majority of smartphone owners, for instance, only use about 10% of the features on their phones.
We should not forget the importance of ‘ergonomics’ — user-friendliness — in the adoption of new technology in the emerging world, especially in Africa. For example, the mobile boom in Africa was due as much to the convenience of the technology as it was to its price.
So how can we make adopting new technologies more convenient? Artificial intelligence (AI) can empower average or even mediocre workers to fit better into high-performance environments. The result is raised productivity and the unleashing of innovative business models, which are so often hobbled by the lack of talent.
Often when people think of automation, their mental images are of robots on a production line or behind a fast-food counter. But the most pervasive form of automation in the workplace today is actually software, often in the form of enterprise resource planning (ERP) platforms and customer relationship management (CRM) systems.
Here is a dirty little secret: Companies have been spending millions in business automation for years at every level of the enterprise. These efforts frequently fail, and the impact on productivity has been less than impressive.
In Africa, as I came to learn during my enterprise software advisory days, there is an epidemic of failure related to current workplace automation tools. A few years ago, I was a consultant to one of the world’s biggest custom software groups. Our strategy in Africa was to target traditional ERP vendors by pointing to the abysmal success rates of deploying ERPs and CRMs on the continent.
When one digs into the source of the failures of most of these programs, it becomes obvious that the problem is poor user interfaces. Crucially, these programs have to work by understanding how people communicate with each other. Clearly, for such tools to fit into the African enterprise and help spur productivity, much more intelligence is needed to adapt them to their users. In this context a fascinating link between AI and jobs emerges.
In the early-medium term — say the next 10 to 25 years – AI is likely to save rather than destroy jobs in many sectors around the world, including in Africa. For AI fear-mongers, robots will simply take over end-to-end. Yet the truth is that we are so far away from being able to build such robots.
There are more plausible alternatives to the possibility of robots quickly replacing all workers. For example, we are more likely to see AI-enhanced training tools to bridge the skills gap. Imagine web-training videos that can interact with a learner, adjusting settings to suit they’re evolving understanding.
The critical insight here is that ‘specialized AI’ that complements average human performers is accelerating at a much faster pace than ‘general AI’ that can effectively and fully replace an average worker. Hence the usual suggestion that AI will replace mid-level workers and leave only the super-talented at the helm refers to a much more distant horizon than current socio-economic forecasting tools can handle. And frankly, diehard futurists aside, such timescales are not particularly useful for human planning.
Africa’s biggest economic challenge today is filling up large sections of its economy with average workers primed to perform average tasks far better than most workers are currently managing to do. In South Africa, as many as 31% of employers cannot fill their vacancies-a major component of overall unemployment.
The tools that exist today to train and empower such workers are dumb. This frustrates enterprises and entrepreneurs who are forced to scale down the absorption of labour to meet expansion and innovation requirements. With cleverer personnel augmentation and empowerment tools, more workers will fit better into more roles in a multi-tasking, fast-paced and rapidly mutating work environment — thereby boosting productivity and lifting overall employment levels on the continent.
We are already seeing the impact of near-AI or pseudo-AI capabilities in several segments of the continent’s labour landscape. African software developers, many of whom do not have the experience and exposure of their peers elsewhere, are using tools like ‘Jade’ to smooth the learning curves for building complex enterprise software architectures, and Chef to better control and harmonize the infrastructure on which the code runs.
Without many of these ever-smarter tools, it is highly unlikely that the thousands of developers filling programming roles on the continent would have been up to scratch. The software being built by tech startups all over the continent would not have been built here, and the jobs they have created would not have existed.
Similar signs are now visible across the design and engineering domains, which are precisely the ones where the skills shortage on the continent is most intense and where the unfilled vacancies and unharnessed job opportunities are most concentrated.
But is AI not too advanced for Africa? Should we be mastering simpler technologies first? Well, the truth is that the days when a society could ‘strategically choose’ to be ‘technologically backward’ are fast disappearing. Technology systems are much too deeply interconnected across global ecosystems for such choices to be viable.
An even more critical point anyway is that AI will actually make technology easier to adopt and harness in Africa. AI, looked at this way, ties strongly with Africa’s current priorities, and African technologists are more primed for it than many might imagine.
Bright Simons invented mobile-phone counterfeit product detectors, and works for mPedigree, which is building intelligent systems to predict the flow of the counterfeits before they show up.
From the March 2017 Print edition