Artificial intelligence (AI) is working its way deeper into more and more aspects of our daily lives. The newest generation of smartphones are coming out with specialised processors dedicated to performing AI tasks.
Internet companies like Google and Facebook are using complex AI algorithms to tailor the content they deliver to suit each individual’s interests and preferences.
The more advanced AI becomes, the less likely we are to even notice it’s there, and the more the lines become blurred between ‘real’ and ‘artificial’ intelligence. The goal, as it should be with any technology, is ostensibly to make our lives better, but to many of us there is something fundamentally discomforting about the prospect of machines rivalling – or even surpassing – the power of human thought. There are also purely practical concerns about AI and the ways it can be used.
The first and most often talked about danger of AI is the risk that it will usurp people’s jobs. The more advanced the technology becomes, the better it will become at being able to automate many of the tasks performed by humans. Those who are working at the leading edge of advancing AI are quick to assure the public that it is meant as a way of augmenting, rather than replacing, the roles of people – but then they would, wouldn’t they? It wouldn’t do to have the populace becoming biased against the very thing they’re in business to develop.
Another danger comes in the form of actual physical harm. Military drones are at the extreme end of this spectrum, seeing as how inflicting damage is at the very core of their purpose. Perhaps more worryingly, though, is the prospect that even AI deployed with good intentions can either be manipulated or simply fail, with potentially lethal consequences. Autonomous vehicles – the poster child of the AI revolution – came under negative scrutiny for just such an occurrence recently, when an Uber self-driving car hit and killed a pedestrian in Arizona, USA.
Silicon Valley tech giants do not have a monopoly on AI development, however. CNN’s Marketplace Africa programme recently explored the potential impact AI could have on South African businesses and industry. In the televised segment, the work of the University of Johannesburg (UJ) was highlighted, with vice chancellor Tshilidzi Marwala boasting that the institution has the largest concentration of people with PHDs in AI, and is the leading centre for AI, on the African continent.
Cape Town, a city that is widely considered as a leader in South Africa’s adoption of ‘smart city’ technologies – a key area for AI – recently hosted an event welcoming SA into the global City.AI community. City.AI is a non-profit organisation that gathers AI practitioners on a quarterly basis across more than forty cities, to share challenges and lessons in applied AI.
Deep learning (while not quite the same thing as AI) has also been the subject of a major development at Johannesburg-based ASIC Design Services. The company has developed a framework for implementing deep learning on FPGAs, and exhibited the technology at this year’s FPGA Design Symposium in the USA as well as at Embedded World in Germany. Robert Green, who presented a paper on the framework at the FPGA Design Symposium, writes about its development in his article on ‘Embedded deep learning framework for FPGAs’.
AI will inevitably come to disrupt many industries in South Africa, just as it will across the rest of the world. We might as well embrace it.
Brett van den Bosch