In 2011, the global scientific community was abuzz with the biggest news in years, an achievement that had the potential to bring science fiction to life in a way not seen up till then.

It was about an intelligent machine that understood the complexities of human language and learnt from mistakes, getting progressively better at giving solutions with practice.

In a world where the words “smart” and “intelligent” have become so overused that they have practically lost meaning — simply connecting to the Internet bestows the prefix “smart” upon a mobile phone — this was a real revolution, a tantalising prospect for the advancement of science and of humanity.

The machine was Watson, a room-sized supercomputer invented by IBM, which had managed to beat two of the world’s best players in Jeopardy! a televised trivia competition in the US that requires players to not only have a repertoire of information at their fingertips, but also figure out the question.

Questions are asked indirectly, for example, instead of asking “Who wrote Hamlet?” a typical Jeopardy! question would be framed thus: “This Globe Theatre playwright penned a tragedy about an indecisive Dane.” The answer, of course, is Shakespeare.

“Up till Watson, computers were not capable of understanding the nuances, complexities and subtleties of human language,” says Rob High, IBM Fellow, Vice-President and Chief Technology Officer of Watson.

“Humans are naturals at indirect speech, but machines are very literal — without the right keywords, for example, it’s nearly impossible to find the information you need when using a search engine. Watson is different; the machine is able to understand what you are really asking, even if it is not framed directly.”

In essence, Watson is able to interpret veiled questions, assess the sources needed to answer a question, be knowledgeable across a wide range of subjects, and deliver the answer fast, in less than five seconds.

In the three years since that first well-publicised win, IBM has been working to transform Watson from a gimmicky game-show winning machine to a commercially viable system that can be used to solve any number of complex problems that not only require massive troves of information, but the ability to find relevant connections in implicit and roundabout communication, as it often as it occurs in real life and as just naturally as humans would — but faster and more reliably than conventional computer systems.

This is the era of cognitive computing, in which systems and software are not programmed, but actually improve by learning so they can discover answers to questions and uncover insights by analysing colossal amounts of Big Data: Call logs, mobile banking transactions, online user-generated content such as blog posts and tweets, online searches, satellite images, and many other large socioeconomic datasets.

Last week, IBM showcased Watson’s capabilities for the African context at a colloquium, whose theme was “Africa in the New Era of Computing,” at Nairobi’s Catholic University.

READ: IBM starts Watson supercomputer roll out in Africa

The company is positioning itself to capture Africa’s virgin Big Data market. Christened Project Lucy, IBM’s $100 million, 10-year project looks to build a cognitive hub in Kenya for Africa and roll out Big Data analytics large-scale to governments, industries, financial markets, schools, hospitals, and any institution that relies on data for decision-making.

In the 21st century, that is practically everywhere.

“With the ability to learn from emerging patterns and discover new correlations, Watson’s cognitive capabilities hold enormous potential in Africa, helping it to achieve in the next two decades what today’s developed markets have achieved over two centuries,” said Kamal Bhattacharya, the Africa Director of IBM Research.

It is part of the bigger trend that has been unfolding over the past decade — that of Africa leapfrogging the Industrial Age right into the Information Age. So far, mobile phones and the Internet have driven the digital revolution, but IBM hopes that cognitive computing could represent the biggest jump yet.

“In medicine, for example, there are more than 21 million articles published in research database PubMed in the biomedical and life sciences; more than one million papers are published every year,” said Robert Morris, Vice-President of Global Labs, IBM Research at the colloquium.

“A doctor can’t keep up with that kind of information. Cognitive computing helps doctors access the particular research that is relevant for a particular patient, link it with other studies that caution or provide contrary advice. It helps the doctor make a better-informed decision.”

In education, Watson could identify the link between a contaminated water borehole, an epidemic of cholera and the subsequent low levels of school attendance in the region. Watson could also help teachers tailor personalised homework for each student, for example, by finding that a student’s poor performance in mathematics is because of a poor grasp of geometry. The system then customises remedial exercises in geometry.

But critics say that the capabilities of Big Data are being oversold, and that Africa’s problems do not actually require such massive amounts of data to solve.

“Big Data gets people excited, but in my opinion, it’s more of hype than reality,” says Kwame Owino, CEO of the Institute for Economic Affairs in Nairobi. “In the African context today, I just don’t see it being worth the trouble, when you consider the cost of collecting, archiving, storing and analysing Big Data”
Mr Owino argues that applying Big Data to personalise students’ homework or reduce congestion on the roads is an “overkill”.

“The additional information one gets from sampling 5,000 individuals and one million is not much, as long as the small sample is sufficiently random. It’s not cost-effective to deploy such technology in most cases,” says Mr Owino.

Further, with massive quantities of data there is a risk of focusing exclusively on finding patterns or correlations and subsequently rushing to judgements without a solid understanding of the deeper dynamics at play — for example, the inherent personality differences between students that makes them react to challenges at school and at home differently.

Such a wealth of data “tempts some researchers to believe that they can see everything at a 30,000-foot view. It is the kind of data that encourages the practice of apophenia: Seeing patterns where none actually exist, simply because massive quantities of data can offer connections that radiate in all directions,” says a 2012 UN report on Big Data for Development.

The report states that such data also “intensifies the search for interesting correlations, which might be (mis)interpreted as causal relationships.” 

Charity Wayua, research scientist at IBM, admits that the system does have some blind spots.

“In education, for example, the system just makes recommendations, it’s up to the teacher to implement it. There’s also the question of where to get all the data to make it work. We’re talking of digitising all students’ records — not just overall scores but what they answered in each and every question — demographic data, data on the home environment and much more. It could take ages to gather all this,” she says.

Dorothy Gordon, director-general of the Kofi Annan Centre for Excellence in Ghana, agrees that perhaps the biggest problem will not actually be overkill, but getting the critical mass of data in the first place.

“At the moment, we don’t have enough data for Watson to work well in Africa,” she said at the colloquium. “We need more research done, more Africa-specific data gathered.”

The vast majority — 70 per cent of adults and 88 per cent of children — infected with HIV worldwide live in sub-Saharan Africa, but almost all of the treatment developed to date has been designed using the research into the North American and European HIV strains.

Big Data and real-time analytics are obviously no modern panacea for age-old development challenges. But Wolfgang Fengler, sector leader of the World Bank and former lead economist for Kenya, argues that Africa is the continent best positioned to harness Big Data for development.

“Africa is in a sweet spot for innovative knowledge generation. There’s high cell phone penetration, low communication costs and an education dividend. But humans and Watson need each other; dumping lots of data into a machine is not cognitive computing. Even the smartest machine cannot make the judgements demanded by humans. But smart machines can help humans make better judgements.”