Interview with Mr Steve Nouri

Mr Tan Cheng Peng: Hello everyone, welcome to SMI Horizon series. Today, I am pleased to have Mr Steve Nouri from Australia. Steve is a data science leader who has evolved the way people look at AI and innovations. He is regularly recognised as one of the top 10 global experts in the fields of Artificial Intelligence and Data Science. He is also a member and contributor to the Forbes Technology Council, International Standard Organization, Harvard Business Review and several other organisations. Steve is the co-founder of Hackmakers and the founder of AI4Diversity, a global non-profit initiative with more than 10,000 volunteers that engages and teaches the diverse communities about AI. He is currently the Chief AI evangelist at Wand, and also is involved in developing international standards towards a more ethically responsible AI. I am pleased to have Steve with me today as SMI’s Distinguished Visitor, to share his insights on how AI is evolving and being applied in many different fields, which could allow us to draw inspiration to do things better with AI in the maritime industry. Hello Steve, welcome to Singapore. Thank you for being on our SMI Horizon series where we interview and feature thought leaders, movers and shakers, for their insights and experiences that could shine the way ahead for Maritime Singapore. Would you like to begin with a few words on what you do as Founder of AI4Diversity?

Mr Steve Nouri: Of course, thank you. AI4Diversity is very close to my heart. It is a non-for-profit initiative I started less than a year ago. It’s a bottom-up approach, community led project that we would like to bring together people from all over the world to help enable under-represented groups and people who are probably going to miss out when the AI is implemented because they don’t have a digital footprint. In general sense, this is a community’s initiative to help each other, find ways to collaborate and also learn AI together. In future, we would like to leverage these huge, inclusive opportunities to help other enterprises to leverage it, and ensure they also come up with projects that are beneficial for everyone, and nobody is left behind.

Mr Tan: Thank you. AI’s definition has changed tremendously from the early 80s and 90s from a quantification of human thinking process, to the explication of human behaviour, and then to an understanding of ourselves, the human being.  Today, AI programmes that have been created have proven capable of mimicking and even surpassing human beings at many tasks. In your view, how has AI science changed from five decades ago and now? 

Mr Nouri: AI is an interesting technology for I guess around four decades, it was sort of going with a pretty slow speed. AI research was happening, but there was not much adoption from the industry perspective. The applications were very, very limited in different sectors. But then if you think about it, like a hockey stick, around 10 to 15 years ago, something changed, something hugely changed. As we all know, having a lot of data available, the computing power, the algorithm, these three elements together helped the mass adoption, the application of AI and delivering value directly to many businesses. This is I guess, joke about it, AI runs in dog years. It is like one year is equal to 10 years, and things are exponentially growing in multiple different industries as you have heard in the health sector, many great projects that are changing the way that we are living, changing the way that we are coming up with the drug discoveries. Everything is just very interesting at this point of time that we are living on, in general sense, we are all super happy to be around these technology at this point. I don’t know if probably it is similar to the point that electricity was sort of discovered, or the first time a person landed on the moon, that we all felt the step forward. It is exactly this time for AI. I would say the next 5 to 10 years, we will see the exponential growth of the AI applications and disruption in different industries. It started from big tech sector enterprises, the ones that were accumulating data for a very long time. They had enough power and money to be able to bring together a lot of great minds, and I guess, scientists to develop it. But then future adoption is going to be much faster, cheaper and widespread.

Mr Tan: Just a few years ago, AI was a field that lived in academic research labs and science fiction films. The average person has the impression that AI is about building sleek robots that could think like humans, behave like humans, and do things that humans can do. Today, businesses thrive on leveraging on AI to boost profits and governments talk about harnessing technology with AI. AI now powers our favourite apps, websites and in years to come drive cars, manage financial portfolios and maybe even manage humans ourselves! There are people out there who have concerns about what AI automation will mean for our jobs and our sense of purpose. Is AI something to be feared? How we can change this fear into opportunities?

Mr Nouri: That’s a great question, because the reality is that we have these Hollywood movies and the perception of those movies that are kind of believed by many in the public sector. I want to give you an example. Let’s say at home you want to use a very, very sharp knife to cut meat, your spouse or parents will definitely be concerned. It’s like, be careful, this is very sharp. The same thing happens when you are using AI, it is really sharp. So we need to be careful, we need to be cautious, we need to be sometimes worried in some applications, but in general sense, the fear is not relevant. As long as the global communities are aware of the usage of AI, and the way AI is leveraged by the government, by the tech sector, by global enterprises, and if there is good supervision, there is no reason for the fear. I would say still, we need to be concerned because of the over centralisation, which will bring this centralisation of the power in the world, and it is not beneficial in many senses. But still, we are in the early days, so it is not doomsday, there is no big problem happening. Elon Musk is a little bit extra concern. Maybe he knows something that he is doing that we are not aware of it. But in the reality, AI is helping many different industries. Also, it is bringing a lot of facilities into our lives. Like when you are watching Netflix, you are bored, you don’t know what to watch, you will see the recommended system coming up with something that is interesting. I am buying something on Amazon and I get some recommendation about something that is definitely valuable and relevant. There is a business power behind it, and it should be. Well, expectations should be there is always a business driving the technology, but at the same time you are on the receiving end of it. It is something that we need to embrace, we need to learn about it, we need to be cautious, and also we need to be involved. That is where AI4Diversity is bringing the involvement of people to understand the global sector and don’t be left behind.

Mr Tan: AI adoption is increasing in most industries, but capabilities vary. Logistics is becoming an AI-driven industry, and in the medical industry, predictive analytics is used to explore patient datasets to have early forecast of the likelihood of onset of certain diseases. Home digital assistants such as Amazon Echo Dot, Google Assistant uses AI as well. Digital Twinning, a subset of AI that involves digital replica of physical assets to aid operations, inspection and maintenance in the construction and engineering industry. What do you think would be the three most revolutionary examples of how AI has been applied in any fields that has made a profound impact and changed the way things have been done forever?

Mr Nouri: I would evaluate the most important impact that is directly related to health and I guess AlphaFold’s recent innovation or invention, based on how you look at it, is one of the most profound AI innovations in recent years. They could predict the structure of the protein, and that will change the way that we are going to make drugs, drug discoveries, the impact, and also the efficiency of the drugs will not be the same. That’s one of the most profound, in my opinion. If you want to go with two others, I would say probably autonomous vehicles are the other ones that are going to disrupt a huge industry of the logistics and transportation. It will change everybody’s lives, we are all sort of involved in one way or another. It is something that we are still looking forward to it. I think that based on observations, and based on the technology, the technology is ready to be adopted. There are lots of other things that need to go forward, like the regulations and the adoption. So that would be another very important and interesting one. Recently, DeepMind announced, I think last week, they announced that they could innovate the way that matrix multiplication is being done. Matrix multiplication is the most important basic, I guess, algorithm in deep learning, or in any machine learning algorithm in general. By using AI, they could come up with a better way, more optimised and efficient way to do this. This is, for me, the beginning of some, I guess, bigger sort of revolution where AI is able to efficiently optimise itself. I am still looking forward to see how this will be applied in different ways, building the models or the current algorithms.

Mr Tan: In your point of view Steve, what would constitute a successful AI application?

Mr Nouri: It definitely depends on the context. So for me, the success is when the AI application adds value to humanity in general. But then you will definitely have different perspective from the business side and also from the research side. Not always the research side agrees with the business side, where it means that there should be a direct short time impact for a business. But the reality is, any project to be successful, you need to have a value that directly would help optimise something, make something cheaper, faster or more available. AI, like any other technology, needs to facilitate one of this, to be able to be adopted or be valuable. In some deep research, we will see that in short time, there might be no applications or there might be no adoption. Like deep neural networks, they came up with this, I guess idea a long time ago, maybe 30 years ago, 40 years ago, but then it wasn’t adopted in anything. So you could see that at that point, it was not a successful research, it was some concept. But then it changed everything in the future. So, it is very difficult sometimes to understand the success if you are having a sort of a mindset of getting value in a short time. Specifically, I can just relate it back to maritime industry where things are very costly, and there are lots of I guess, legacy systems. So if you want to change something, you might not get the value back in short time, and that means that a lot of the companies might not be very interested in doing that.

Mr Tan: That is a good lead in into the maritime sector, which we will talk about next. The maritime sector is a very global industry. It is made up of several major groups of stakeholders such as shipping companies, port authorities, terminal operators, the maritime logistics supply chain and a whole ecosystem of onshore ancillary services such as finance, ship management, classification societies, etc. What do you think are the top three areas where AI will make the most impact in the maritime industry? What kind of revolutionary changes that AI can create in the maritime sector?

Mr Nouri: That’s a billion dollar question. There is no easy answer to that. I can see lots of people are working in these projects, different interesting projects happening in Singapore and around the world. They are trying to optimise something or they are trying to make something cheaper, maybe transportation cheaper, or maybe something related to sustainability. These are all important, at the same time they might have different impacts based on the way that you look at it. Some people believe that sustainability is the most important part of it, the business might disagree and think that when things get cheaper, that’s where it has the most impact. So I would like to just think about it. Probably, based on the complexity of the whole ecosystem of the maritime industry, couple of things are very important to happen. One is cross-industry collaboration, and make sure that the stakeholders are agreeing on some long term goals that might not necessarily benefit one or the other ones in the short term. The second one is exchanging data, because data is the most important part of an AI platform. It is  pretty much the ingredient, maybe 60% of what we believe that needs to happen in order to have an AI platform is just getting the data in one place that can be leveraged by any model. Maybe some coalitions, some collaborations across the different countries and industries might be able to bring this data together. So I would say the first one for me is just discovering and gathering rich data that can be leveraged later. Secondly, I guess simulation, a very detailed and accurate simulation of the weather, in general the earth, would be very, very important to the maritime industry. Finally, reinforcement learning in terms of understanding, identifying the areas of optimisation without actually doing any sort of experiments, because like I said, it’s very costly to do an experiment. Imagine you need to send 1000 ships across the world and try to come up with the way to optimise them. Instead of that, just make sure that you identify this simulation environment and make it as accurate as possible, then run these simulation agents inside the environment to come up with the best practice. It can be in any way or shape. It can be policies, it can be the optimisation of the speed, the way to run the ship. I guess finally would be adoption of robotics and autonomous vehicles in general sense, in the cargo, in the ships, in the port and trying to have, as much as possible, less manpower involved, and then that would definitely save a lot of costs, make things faster and more efficient.

Mr Tan: The maritime industry has sometimes been criticised by some as being too slow to get on board major global developments, such as digitalisation and the use of AI. Capital investments in major ocean-going vessels and port infrastructures typically have life spans of 20 to 30 years in sunk costs. As we try to look ahead in the next 20 to 30 years, how can AI potentially disrupt the maritime industry if it does not embrace the AI evolution that is coming?

Mr Nouri: That is a really good question. All the industries in the world, they need to ask themselves this question every day: How, where and when we will be disrupted. Some of them are more agile, they disrupt themselves before being disrupted from outside. They are more smart and more agile. Some of them are more complacent. It just depends on different reasons. One of the reasons might be the cost of entry, which produces some safe nets for that particular industry, and they don’t feel that there would be any sort of any newcomer that will challenge them. But if they look at the reality in the world, like the car industry, out of nowhere there is Tesla coming. All the big manufacturers that have been there forever, like Ford, GM, Toyota, they were like, I am not saying they are complacent, but they were sort of relying on their traditional way of business. Every year they changed a little bit of the model, the new platform and they sent it to the world. Now they have this huge company that came out of nowhere, and now they are forefront of the value that they are generating. Sometimes it takes a lot of time for the others to actually get their, I guess, position if they are not the early adopters. So the same way, I think if the maritime industry doesn’t think about the tech that they have been accumulating for some years, there might be newcomers that they figured out that there is a huge value here which is not being harnessed by the companies. If they can put together some sort of a new mindset, we can drop the costs, you know, by fractions. SpaceX is another one. NASA has been sending people to outside the earth for very long time in this space, and they have been doing it in a very, I guess, pioneering way. They have been inventing and innovating, and they have been accumulating all these scientists from all over the world. And then still, you would see SpaceX coming in and they would make some new technology that will be able to be recycled. With a fraction of cost, they can send someone into the space. Now they need to think about how to either collaborate or to get out of the business. The good news for such companies like NASA, they are government related and they are able to survive because they are tied to the national securities and they will not be disrupted. That is one of the reasons that they sometimes might not feel the pressure. Maybe maritime doesn’t have the same sort of safe net that, you know, there might be someone coming up with a fully autonomous ship that would go very fast and very cheap and would run on, I don’t know, some nuclear fuel. That is just something that will be totally different. Nobody expecting that and out of nowhere, we have seen that difference will change the balance in the world.

Mr Tan: Okay. Now, earlier you mentioned about two key factors. One is the forming of a coalition of partners, and the other one, the gathering of datasets for AI applications. Are there other ways you think that we can further advance and accelerate the use of AI in the maritime sector?

Mr Nouri: As I said, these two definitely are very important. But at the same time, I guess talent is also another important one. Maritime sector is generally pretty complex, but at the same time, not on the forefront of the public view. So there aren’t a lot of information shared with the public, not that it is not available, it is just not a lot of interest in terms of bringing the talents into the industry, based on my own observation and I can be wrong. But I think there should be a little bit more effort put in this particular, I guess, movement to bring more talents across the world.  To bring their interests, give them the idea and the platform to be able to do the innovative work that they are interested. Also, if you have watched the Tesla AI Day a couple of weeks ago, the whole setup, the whole thing was actually a show for the talent. That is a talent show, like they already said that publicly, everybody knows about it. So they put together such a show, a huge show. Robot will come and walk a little bit, flimsy robot that is probably not fully-fledged and probably cannot do anything outside that particular steps that we have watched. But then the huge impact that it makes would bring a lot of interests across the world. Everybody would know about them, would be interested to be part of it. I think something in a similar fashion might be able to bring the maritime industry more in forefront of views of the people that are interested in this sector.

Mr Tan: So Steve, we know about the good that AI can do in many areas. Are there any downsides to AI that we should look out for?

Mr Nouri: I guess I touched a little bit on this front. Over centralisation is very tempting. Think about it, it is the only way that you can conquer the world without conquering the world. If somebody wants to really leverage that particular technology to do that, there aren’t a lot of barriers, there aren’t a lot of limitations. Big tech enterprises are collecting and accumulating a lot of data. So that is one of the major threats that I think we all need to be aware of it. The other one is another obvious one, it is bias. The bias will come into the data from different sources. Actually, tackling the bias is something that everyone in different level, doesn’t matter enterprise or very small company, they all need to be aware of it because that sort of would impact the accuracy, and would be also unethical to have it incorporated and not being aware of it. I think these are the major ones, responsible AI and also accumulation of power, are the major two threats that we all collectively need to tackle in the next 10 years.

Mr Tan: The potential of Artificial Intelligence is hard to ignore.  Artificial intelligence can deliver considerable benefits to the maritime supply chain, global shipping operations as well as port operations. Some advantages include reduced cost, reduced risk, improved forecasting, faster deliveries through more optimised routes, and much more. There will be continued digital transformation of the maritime industry. Those who take action and start incorporating AI into their business models, supply chains and operations can expect to reap not only a more efficient business model, but also a smarter and more intelligent way of doing things. This, in turn, will free up scarce and high value human resources to be deployed in an intelligent and targeted manner, where machines and algorithms are not able to do a better job than humans. So, any last thoughts Steve on this maritime AI topic?

Mr Nouri: This maritime industry has huge, huge potential. I think that in the next couple of years, it will definitely be the major fronts for investors and researchers. So I would totally encourage all the stakeholders to try to embrace it as much as possible. And thank you for having me.

Mr Tan: Thank you, Steve, for your time and sharing your thoughts on this series of SMI Horizon. Thank you very much.