AI Trends and Challenges for the Future with Johan Steyn

So grab a coffee , sit back and let's get ready to blast off . I'm your host , frans van der Toij .

So , yeah , it's a really interesting thing how you've carved out that sort of association that you know . You've built that brand , that look . You love AI . You talk a lot about AI on all sides of it .

So just maybe , briefly , just tell us a little bit about yourself , just to give people a sense of who you are and what you do , and then we'll get into this amazing conversation today .

So I do quite a lot of training with universities and with companies to make sure they understand this technology . I consult and make sure that the strategy around this technology works for them . And then a lot of public speaking and writing . But I often say I speak as a father of a 10-year-old more than as an AI guy .

We have to think seriously about our children's future and how this technology will potentially very negatively impact the world we live in and how this technology will potentially very negatively impact the world we live in .

You're also now part of the working group on AI strategy in South Africa specifically . I know you do work all over as well , but what do you see as the most pressing of challenges in developing our national AI framework ? Because that's the interesting part , right , the EU has come out with their legislation .

I'm not sure about the US , actually , but you know , I mean , the EU is always acting on these kinds of things and we want to follow suit . So what are some of the challenges ?

But luckily there is a lot happening . So this working group there's quite a lot of people involved . I'm very happy that it's not just academics or people from the government . They also have included quite a lot of people working in the field . So you know we can learn a lot from the EU's AI Act , like you referred to .

In fact , there's a lot of countries in Africa already and , of course , across the world , busy working on these things . So I think we shouldn't reinvent the wheel . We should take what we can learn from , especially the EU Act , but then we also have to make sure that it's uniquely African and South African .

You know we have societal issues that's not necessarily as prevalent in the so-called global north Europe , northern America , china and the like . We need to look at our joblessness , which will possibly increase even more with this technology Things like chatbots .

You know we have more than 3,000 languages and dialects in Africa , and only a very , very small percentage of those languages are served by AI or conversational AI technology . The same is true for education and healthcare . I always say we need African solutions , created by Africans for Africa . So I think it's good that this working group is busy .

We need to make sure that it's not overly academic . Of course it's a process . We need government's involvement , we need various stakeholders , but at least it's happening . I just hope that it will happen at pace and won't take too long . Regulation is there to protect us .

You know , when we started driving cars and there were the first few car crashes and people died , that's when the regulation of seatbelts came in . But the question with AI is what needs to go wrong and to what extent will that be quite severe for us to really make sure that we increase our speed with regulatory frameworks . But things are happening .

We should support it and I can't stress how important it is that we get it right .

In fact , he wrote a whole book about this topic , because I don't think we talk about or think about this enough . I'm quite interested , just maybe , johan . I mean you mentioned the Global North . How does our approach to getting the regulation together for AI ? How does that compare to what the Global North is doing or how ?

That essentially excludes us from doing business with Europe and other areas . So it's like with GDPR and POPIA . You know our privacy regulation is fairly closely aligned to Europe's GDPR . We'll probably do the same , but again , we can't just copy and paste the AI Act in the EU . We need to look at very specific situations here in our country .

Now the balancing act here is between innovation and regulation , because you can also not over-regulate it so much that we can't compete globally with other businesses . We have to use AI and automation technologies more and more , especially financial services , space , the telecom space , insurance and other areas . So can you over-regulate it ?

Probably , but we should make sure we find that balance and we'll also , given the speed at which this technology is advancing , we'll need to update that framework annually probably .

Regulation and regulators are always behind innovation , but my worry is just that the split between how quickly we can regulate effectively and how quickly this technology is increasing in power . That split will just become more and more that we might get to a point where it's almost not worth looking at updating the regulation .

So there's a lot that we need to do . We all need to be involved , as parents , as teachers and working with government . So it's not a kind of a scaremongering thing .

But I think , franschbach , if we just think of deep fakes , you know the fact that people can use my image and voice and yours and present it in the court of law and it's almost indistinguishable from the real footage . Think of what they can do with our children .

There's already a big problem in America where they take the pictures of young girls on social media and they produce pornographic material with them . Then they email the parents for money to pay them off not to publish these pictures . So there's a bit of an Orwellian world awaiting us if we don't get this right .

Also , you know , given South Africa's I mean our socioeconomic landscape and the way that things are , the specific challenges and things that we have how can AI be leveraged to address some of those challenges here ? Do you think ?

Think of secondary and tertiary education as well . There's a lot we can do around what we call hyper-personalized education . What I mean with that is that we don't necessarily have to change the curricula , but we can change how it is delivered .

So if you have enough data on your students or your staff , for instance , you can see that one person learns better in a classroom , another person on their own , some people like audio books , some people like videos , some people like mind mapping to really get the best out of them .

So I think we already have across the world , but in our country , a lot of challenges around education . I don't think we are preparing our youngsters for the work and to be the workforce of tomorrow .

Because of the rate of technological change Again in healthcare , to really bring down the costs to scan for cancer and for tuberculosis , especially in rural areas , to use drones to deliver blood or other medicine .

There's a lot we can do to serve our population and I think that's one of the key things in the work that I'm doing is the fact that this technology because it can , but that it should serve humans we have to democratize it . You know , at this stage , sadly , the power of AI lies in a few hands . You know Metza , facebook , google , chatjet or OpenAI .

It's almost like an oligarchy of technological ownership that doesn't serve all of us and it's a big money-making game , you know . So , again , if we do it right , if we look at the societal problems we can solve , for we can definitely use AI . Another thing is predicting weather patterns .

We've had a lot of severe weather instances over the last few months flooding , etc . Can we predict it better through AI and make sure that people are moved from the areas that will probably be hit by a tornado ? People are moved from the areas that will probably be hit by a tornado .

The other one is uprisings , like we had in KZM some months back , can we look at without infringing on people's privacy ? But look at chatter on the Internet and WhatsApp , et cetera , to pick up where there might be a terrorist incident or an uprising . So again , it's there to serve the population and to serve people .

The police should be using this technology like crazy Security firms as well . And again , that fine balance is what is my right . You know , for instance , my face is also my private data . If I walk down the street and the security camera films me without my consent , they are breaking the law .

Now do I just not walk on the streets of Santon anymore , or should there be a bright yellow sign saying you're being filmed , so that we know ? So , again , this power in the hands of the wrong people will be dramatic and severe , but if we use it right , we can do so much good . With the French one . That's sadly human nature . You know . We grab power .

We want to be richer than others , we want to be more powerful . If you're in an elected party , if you are in government across the world , your primary goal is to win the next election , even more so than serving the people . Now imagine what governments can do spying on people changing their minds , which we see already .

Imagine what governments can do spying on people changing their minds , which we see already . America's election in November will be interesting when we think about fake news , social media feeds . So I'm leaning more toward the negative side that we need to be careful of it .

You know , it's also crypto is . You know , there were people giving themselves out as being specialists and experts and all kinds of thing , when it's only been around for two , three years . At that point in time , I'm like I don't even think , like I don't know how long . It's been about 11 , 12 years now .

I still don't think that we have experts , because I think there's still a lot of things that we don't understand about that , and AI even more so . But you also I mean , you're talking about , you know sort of the impact and the ways that we can use it in South Africa . But you also wrote in Business Day about cultural sensitivity .

That's also an important consideration . When it comes to AI , there's obviously a lot of South African businesses , and particularly when we work with financial planners and advisors we work with financial planners and advisors .

Some of us are starting to use some of this technology you know , maybe more in the back office than client-facing kind of situations , but one of the things that we need to be really cognizant of , you know , if we develop apps or if we incorporate AI into what it is that we do , and particularly , I guess , when it comes to to the client facing side of

things Frans

was so important .

But if you know Greek mythology you would have recognized that it's Bacchus and his cohort and that's what they were celebrating . But it made me think of the fact that we , as humans , we are symbol-driven creatures .

We think through symbolism , we tell stories through symbolism and metaphors , and we need to just keep in mind those very sensitive cultural nuances when we especially have applications and platforms that deal with our customer base . Think , for instance , of chatbots or conversational AI .

I mean we mostly use English chatbots because it's already developed , the data sets are there . There's a lot of work being done in South Africa by some firms and universities to make sure that we incorporate more of our native languages .

But one of the key things here is you have to test it with your target audience , first , a small subset , to make sure that this chatbot is not using language that's difficult to understand or that is even offensive . It could be gender-based reasons or others .

You know , for instance , in a lot of our cultures , when you speak to people who are older , there's a lot more protocol involved seeing them as elders and having respect for them .

So we need to build in those nuances , without , of course , over-engineering it , because we're also running businesses and you know we need to release these things quickly to make sure we increase our market share . That's why I always say make sure that the core team in your company that develop these systems are not just vendors or your technology team .

Make sure that it's a multidisciplinary team . You have to have HR , the legal department , the procurement department , the organizational design and change management people and others and financial people involved . But then , of course , you need diversity .

Make sure that that core team is multicultural , multi-ethnic and so forth , just to make sure that the bias that we have because we all have biases and that's not bad , it's human nature as long as we're aware of it . You know , if I build an AI algorithm , it will probably be a result of who I think somebody is and how the world works .

You know an example I often use Francois . If I have to build an algorithm for a bank to look at all their customers , or say 600,000 customers and to try and identify family units because we might want to sell educational , banking products or others . My worldview is that the family is a man and a woman , potentially with children , and that's correct .

But then a colleague needs to say what about same-sex couples ? What about people who can't or don't want to have children ? What about single parents ? And there's probably 12 or 13 iterations . So we will build things in our image . It's human nature .

But when we build something , especially for our customers , make sure that the customers have a voice , but also that the team internally that build it ensures that it is culturally relevant to the target database . So it's not just the tech , it's the human aspect on top of the tech that we really need to think about .

But , particularly as an educator , which skills do you think is really really crucial for South Africans specifically to develop and to enable them to thrive in this new AI era that we are already in ? I don't think it's not coming anymore .

He discovers stuff on our smart TV or smart speaker that I didn't even exist but that's a small portion of our population .

But to make sure that our youngsters understand technology , that they don't fear it , but also that they understand how to use it responsibly you know there's a lot happening at this stage regarding sexual predators online , because their power to automate and to customize the messages to youngsters is incredible , especially if you use large language models .

So our children should learn that when they chat to a chatbot , for instance , not to reveal private information . My son was speaking to his smart speaker a year ago and this app started asking him what is his daddy's name , where do we live and which school is he in . I mean , I had a backflip over my desk to stop this whole thing .

But they shouldn't fear it , but they should use it responsibly . So , to enter a workforce where they are familiar with computers , what computers can do , prompting large language models . So that's the one hand . On the other hand , you know there are skills that we will need .

That's things like critical thinking , problem solving , working in a corporate , for instance , or with a team , understanding the politics . You know how to encourage colleagues , how to deal with depression , how to work with money .

So it's those two lenses that I think we need across the world , but especially here , because a lot of children don't learn these things at home . Parents almost outsource that responsibility to teachers and they don't have the capacity . We , as parents should teach our children those things how to treat women with respect .

That's a big thing I'm trying to do with my son Make sure they are technologically fit , but also make sure that they are well-rounded humans who can work with others for the greater good .

So I mean , how do you think you know , if you think about the potential impacts that AI can have on the job market , specifically here in South Africa , because I think it's going to be different in different parts of the world .

It's not necessarily going to be exactly the same , but do you maybe have any views around what the possible impact on the overall job market is going to be in South Africa and also what would be the best ways to maybe prepare for that now ?

So but now balancing the experience that already sits in that administrative team with technology is is the trick , because you't , if somebody has been a receptionist or administrator or a payments clerk or a debtors or a creditors clerk , for instance , for 30 years , you can't just automate them away .

I think a lot of the lower level tasks again the extracting the deal from documents , automating certain business tasks in the background we can do that Now . That will inevitably mean that we might only now need one data clock and not five , because the technology can do the rest .

Any work , like even mining , which is one of the large sectors in our country , a lot of mining work is already being automated away . So now here's the balancing act , because I have been thinking about this Do we need to call it a BEE kind of regulation , but when it comes to automation ?

So in other words , where the government forces companies to not automate away more than 10% of their workforce . But then what do we do about being globally competitive ? So that is going to be a challenge . You know , in our country every wage earner , on average , looks after eight or nine other people , so the impact will be severe .

Now then , people will start speaking about the you know , global income . I'm trying to think of what they call it now the name will come back to me In other words , where we finance people who lose their jobs through AI . But again , how are we going to fund it ? Even in our country , we are struggling to fund our social grants .

So what if our jobless number doubles in the next five years ? The impact could be again , from a social uprising point of view hungry people desperate for food for their children . The crime levels will move up . Again , it comes back to education . It comes back to being globally competitive .

You know there's a lot happening in our country where organizations like BPSA are bringing in call center work into our country . They're doing great work with a number of organizations . You know , culturally we adapt to a lot of other countries with our sense of humor , with our diversity , but we should create 10 000 , 100 000 more jobs from call centers .

But now call centers are being automated by conversational ai as well . So I think the the long there's a long answer for the point is I don't know what government can do . We need to figure it out .

You know we're going to need to figure out , like , how do we do that ? And that's going to be be so , so important .

Be honest , as far as um , you know you think about um , everything that we've spoken about today , um , which is quite , quite interesting to me , and there's a lot of things that you mentioned that I have not really even spent one minute to think about um . So thank you very much for that .

I think a lot of people listening are exactly like , like , oh , I need to to consider this um .

But if you think about all the ai regulation and you know also innovation and things that's coming and how these two are going to try and balance each other out , um , you know , what do you think is going to , what that's going to look like in the next five years or so ?

I mean , if you take our privacy regulation for PR , we've had some serious breaches over the last few years and I've not seen one board of directors go to prison or pay the 10 million rand fine as per the regulation , because it's difficult to prosecute these large firms .

Now we're bringing even more regulation , so I would just my encouragement to people listening in now is familiarize yourself with this . Don't have to become an AI expert or a data scientist . Read a lot , attend events like these , speak to people . Make sure that you build technology for the future that will be ethically but also legally compliant .

Don't fear the technology , but please , please , please , don't just try and AI everything . Start with the business case . I , the French , I often say tongue in cheek if you're not using Excel properly yet , don't even look at AI . So don't have a mandate for people to automate an AI stuff . Fix your key business problems .

It might not even be a technological solution , for instance , a toxic leadership structure , people who hate their jobs , who don't understand their jobs , clients who don't understand your technology . That's not a technological discussion , that's a leadership and a human to human discussion .

So , yeah , be familiar with this tech , realize the limitations because there are many , and prepare yourself and your children for the future . That would be , I think , my main message amazing .

You have a gift and a talent for that . Thank you so much , really really appreciate it If people do want to maybe follow you I know you've got an incredible newsletter and things that you send out and you write incredible articles all the time , and where's best for people to follow you and to maybe connect with you ? Yeah , francho , thanks .

No , how does it go ? Stay safe , be blessed and prosper and remember to continue thank you very much .

If you enjoyed this episode of the podcast , hit that subscribe button or the follow button and share it with your friends , colleagues and anyone else you think would get something out of it . Thanks for tuning in and I'll catch you in the next one .