Let the children play – with their minds

Parents, whatever you do, do not — I repeat do not — curb your children’s enthusiasm for online gaming.

I present to you Prof Benjamin Rosman, whose early interest in computer games of strategy and tactics led him from that small screen in his pre-teen bedroom straight to the cutting edge of artificial intelligence (AI).

He now runs the robotics, autonomous intelligence and learning (Rail) laboratory at the school of computer science and applied mathematics at the University of the Witwatersrand.

He is the director of the national e-science postgraduate teaching and training platform (NEPTTP); the chief science officer of Lelapa AI, building AI for Africans by Africans; and founder of the Deep Learning Indaba summer school, with a focus on strengthening African machine learning, which now has satellite events in 47 African countries. 

He has won numerous global awards and to top this litany of smarts — and the reason for our lunch — he is the founding director of the Machine Intelligence and Neural Discovery (Mind) Institute at Wits, launched last November.

I say let the children play.

We meet at the Wits Club for the consistently wonderful delights supplied by the Olives and Plates outpost where he is clearly a regular. Over a wrap and an omelette, Benjamin  tells me he was way ahead of the curve with the AI tsunami.

He tells me an amusing story about presenting at a conference where a baffled gentleman eventually piped up to ask why they were talking about artificial insemination. No chance of such mistakes any more. AI is very much with us. His area of specialisation is reinforcement learning, which he focused on in his postgrad studies at the University of Edinburgh.

But he always knew he would return home because one of his perennial dreams has been to make South Africa, and Johannesburg in particular, an AI hub. Not just as passive recipients of global tech but as active creators of AI that is African and directed at African issues.

“Most people around here doing AI are just taking off the shelf stuff and applying it to their problems. There's hardly anybody in Africa who's actually building the next generation and that, to me, is a problem. The actual development of AI is something we've got to be doing as well. Otherwise, we're just consumers of technologies.”

His AI research lab at Wits is now the largest in Africa and possibly the biggest in the southern hemisphere.

If we look at the history of AI, part of that came from trying to understand our own intelligence.

—  Benjamin Rosman, chief science officer of Lelapa AI

“In 2017, with seven other people, we started a movement called Deep Learning Indaba, which was a group of us who were either expats or had studied abroad and come back, who wanted to give something back. Already at that point there were all these nonsense business conferences around AI and we wanted something really deep and technical to help the next generation get to the cutting edge.”

More than 16,000 people attended the most recent one last year and they have more than addressed the paucity of academic papers published out of African institutions. And it has led to numerous startups and projects.

“A couple of years back, kind of growing out of that community, a group of six of us started a startup that's doing AI for African languages. There are 2,000 plus languages spoken in Africa and basically anyone who doesn't speak English or French is excluded from accessing any technology. And so we've got a company called Lelapa AI that's building the lapa.” As in home. 

He is delightfully down to earth, charming and given to good simple explanations of stuff that can seem opaque to non-techies like me. And unusually self deprecating for a seriously smart and driven individual. He is the antithesis of the tech bro trope. 

I will resist the temptation to give airtime to our discussion of Elon Musk, but rather turn to Mind, the fascinating interdisciplinary project he is now leading at Wits.

It’s as if the archetypal renaissance man and the hive mind got together and reproduced a little baby institute for our times. Its founding premise is the ideal you imagine for human learning — that our intelligence is sparked and that we learn best not in silos of niche activities but in the magically generative space where exploratory conversations and scientific research happen across a multiplicity of disciplines.

“If we look at the history of AI, part of that came from trying to understand our own intelligence. I love being collaborative. I love working with other people. And so we set this up as an institute that looks at intelligence rather than artificial intelligence. So it's artificial and natural intelligence. Probably a third of the institute is AI people but the rest are all these other disciplines. We've got a lot of amazing psychologists, behavioural psychologists, neuropsychologists, there's a huge neuroanatomy group. There are people doing brain evolution, amazing philosophers, a whole lot of people working in ethics and governance, linguists, all these interesting people.

“And so the idea of the institute was to bring all these people together and say this kind of collaboration is something that I don't think most universities could pull off, but we can.

“We are trying to think holistically and at the same time sitting on government advisory committees —  so really trying to make sure we're living up to our potential in this space.”