Stuart McClure: Cyber superhero; using AI to head off the criminals

The World Economic Forum has been way ahead on the major technology trends. Fellow South African and Silicon Valley hero Paul Maritz was here talking about Cloud Computing long before it caught on. Ditto smartphones, twitter, youtube and other transformative technologies we now accept as mainstream. Blockchain is the next wave that’s about to become ubiquitous, enhancing the role that cyber security will need to play. Stuart McClure is at the forefront of using artificial intelligence and processing Big Data to protect the world against cyber criminals. And if you think you’re not in the sights of these miscreants, consider what Paypal CEO Dan Schulman told the Edelmans Trust Barometer event this morning: “There are companies that have been hacked. And companies who delude themselves that they haven’t been.” One day the world may salute Stuart McClure in the same way it shall the people who cure HIV/Aids and cancer. – Alec Hogg

Well, Iā€™m here with Stuart McClure. His business is called Cylance, something to do with cyberspace?

Yes, basically we are responsible for providing protection online in cyberspace.

Thatā€™s a big deal at the moment.

Itā€™s a big deal and no one seems to be doing it properly and so weā€™ve applied a brand new way of learning about how the bad guys think and work and apply it into technology to actually convict and prevent attacks before they start. Itā€™s a true way to actually use AI to predict whether or not something is going to be malicious and preventing it and stopping it, for example, the attacks on the DNC recently, the attacks at Target, Home Depot, global names around the world.

The Russians?

The Russians, the Chinese.

Is there anything in that Russian story that Trump keeps denying and the Democrats keep saying is on?

I always say, it definitely could be possible, but the problem is attribution in cyberspace is basically useless, why even bother because itā€™s so easy to make attacks look like anybody. I could make it look like you hacked into the DNC, definitively, with your fingerprints. If I really wanted to, I could, so whatā€™s the point in it? I really turn it around and say, ā€œWhat about the how, how did they get in, why was it so easy to get in?ā€

Well, why?

Well, because for the most part no technologies out there can actually predict whatā€™s going on, no process that we have, in the human process that we look for these bad guys, can we prevent these kinds of attacks in real time without the technology, so you really have to look to these next gen, AI-driven machine-learning-driven technologies and security, which are just starting to come out now. Weā€™ve been out for a few years so weā€™re sort of ahead again, but being able to predict and prevent before it ever starts, thatā€™s really the key, not just detect and respond, but prevent it.

Artificial Intelligence, you have robots around the cyberspace making sure that the bad guys donā€™t get in.

Yes, basically thatā€™s it, cyber robots, you can think of it on the computer running all the time, vigilant 24 by seven, watching everything that comes into the computer and then determining whether or not itā€™s malicious or not without ever checking to the cloud or asking for a signature somewhere or seeing if it was something in the past, but we can literally look at it a Priori, objectively and determine whether it is malicious or not.

You know something that confuses laypeople is the way that youā€™ve got this thing called the deep web and really bad criminals who operate in that space and they can do just about anything to you, steal your money if they want to. How much of the work that youā€™re doing is working towards addressing the deep web which to all intents and purposes, for the layman is something that is untouchable?

Yes, the deep or sometimes dark web is this area that is basically not indexed by any search engines and these are servers and computers and private networks that run all the time, 24 by seven by criminals or people that want to do bad things. Ultimately, what we will do is anything that comes out of that dark web, we can prevent against. So if one of the attackers or one of the groups on the dark web want to send you, for example, an attachment with a malicious payload and you open it, thinking maybe itā€™s coming from me because it looks like itā€™s from my email address and you start to open it, well we can actually determine even though weā€™ve never seen it before and the world has never seen it before, we can determine that itā€™s malicious in real time in milliseconds before it actually gets opened and we can block it.

So you would try and open this attachment like a .PDF file for example and it would block it, it would say, ā€œSorry, this attachmentā€™s maliciousā€, itā€™s not going to allow it. Now even though we have never seen it, the world has never seen it, so no antivirus has ever seen it, no attack victim has ever seen it, but we have determined that itā€™s malicious in real time using the learning of the past with AI to determine the future.

How many times have you been to Davos?

This is actually my first time, as part of the technology pioneers.

Did you have a good dinner last night?

Yes, it was a good dinner, it was a fun dinner.

Meet interesting people?

Many interesting people, there were folks from Bahrain as well as from our neck of the woods in San Francisco, there seemed to be a bunch of those folks too, a lot of startups there, but Iā€™m always keen to meet a lot of interesting people and to understand where theyā€™re coming from and how they’re thinking about the problem and if theyā€™re applying sort of new next gen type technologies like AI and machine learning. Iā€™m really curious to know how theyā€™re using it. For example, Iā€™ve met people that for example, use AI to predict based on the voice of an individual, whether or not they have early onset Parkinsonā€™s or early onset Alzheimerā€™s and they can get to the degree of confidence up to 94, 95 percent predictability based on their learning and training of the computers that they have.

What are they doing with this?

What they’re trying to do is start businesses and companies that help the health system provide early diagnosis so they can get early treatment. I mean the ultimate application of AI however, will be taking a human body scan of all of its cells and all the state of its cells and the chemistry of the body and actually learning what is toxic situations and malicious situations and being able to tell you ahead of time what changes in your life you can make to prevent early onset disease or illness. Itā€™s this kind of application of AI; I think is really the future of what weā€™re doing. Weā€™re just proving it here in cyberspace that it works. It works 99.9 percent of the time we can predict and prevent bad stuff from happening, but in the physical world, weā€™re just getting started with how we can apply AI into healthcare for example in diagnostics.

Itā€™s an exciting time to be alive when youā€™re exposed to all of these potential developments.

Itā€™s a lot of fun because I mean Iā€™ve been in cyber and technology for over 30 years now and Iā€™ve never seen so much hope for applying technology to really benefit mankind and human beings. I have four kids, all of them are on their vices constantly, you know they play video games left and right and so thereā€™s some benefit there in terms of entertainment or keeping them off the streets, but when it comes to actually applying technology for the real physical betterment of human beings, today is the time.

There are going to be lots of people kicking back against it, thereā€™s a whole group of vested interests. How does one play that card, how do you make sure that innovation doesnā€™t get thrown out because it is going to be hurting people?

Yes, thereā€™s always that worry that weā€™ll get too good at training computers to be like human beings. I think the ultimate fear is always that we train computers to eliminate the need for humans and if you understood sort of where the space is in machine learning and our capabilities of supervised versus unsupervised learning, you would know that weā€™re decades and decades away from ever really talking about that in a meaningful conversation. However, I understand the fears and so my recommendation is always to just think about, we can barely understand how our brain works today, how are we supposed to train a computer on how our brain works. I mean itā€™s just a big leap forward for us to even think that, thatā€™s truly possible, even to me.

Just unpack that a little, ā€œWe barely understand how our brain worksā€, where are we in that scale?

I mean, you know we donā€™t really understand how our brain works. There are many patterns in the human brain and behaviour that theyā€™re very predictable and no linguistic programming is one of kind of semi-science that proves that out, but ultimately, we have no idea how emotions are really derived and how logical thinking is really derived and so how are we supposed to train a computer to think like a human being and actually get better if we canā€™t even train them properly to know how we think.

Elon Musk

So I think weā€™re ways away from really having to worry about that, but itā€™s of good concern, we want to make sure weā€™re thinking about it as we develop the AI, make sure thereā€™s that kill switch, you know, which Elon Musk always tries to pull in and I donā€™t see why we wouldnā€™t, it seems silly. For the most part if you really know what AI does today and machine learning does, it just learns from the past to make a prediction in the future. Thatā€™s really it today, so this sort of like singularity and things of that nature, I think is long ways away, but what we can do today is apply AI in fantastic ways to make our lives better for good, so I call it AI For Good. A lot of people worry about AI, I donā€™t. I think AI can be tremendously powerful and helpful in everyday life for human beings around the world. Youā€™ve just got to find the right application for it.

No doubt. Stuart, why the breakthrough? Is it a combination of big data; is it a combination of people just joining the dots a little bit more accurately?

Itā€™s a few things. First off yes, the data is finally available and we have large amounts of it and we have long periods of time for it. We have the platforms now, the systems and the networks and the cloud to be able to actually handle and process the data because if you have data, thatā€™s one thing, but if you canā€™t process it, itā€™s useless to you right. So you have those abilities, but you also have the kind of emerging learning algorithms that are getting better and better at figuring out what lessons are there to be learned inside the data quicker. For example, the latest neural nets tend to be sort of what you think of as deep learning algorithms that go out and go find and go deep into large amounts of data basically in dimensional space to determine where that classification is, A or B, like is it a male or a female and pictures of people right.

How do you know itā€™s a male or a female? Well, the more data you give it, the more features of each male, female, the computer and the neural nets can sort of learn that. In the old school way, it would produce a lot of false positives, a lot of false negatives, but with a lot of the new emerging learning algorithms, it can actually get to that 99th percentile.

Cylance is the ā€˜lanceā€™ anything to do with freelance, do you use associates around the world?

Well, you know Cylance originally was Cybervigilance, but it also started a play on the word ā€˜lanceā€™ which is something that you use in weapons and war and so Cyberlance was also another play on us, but ultimately, Cylance and the way we pronounce it is meant to be ā€˜Silenceā€™, in other words, security has to be ā€˜silentā€™ for it to really be effective and used otherwise it just gets turned off or worked around or removed, so Cylance is what we produce on the end computer because we prevent the bad stuff from ever happening.

Great companies need a higher purpose, have you got one?

Yes, I mean literally to protect everybody in the world through prediction using AI. I mean it really is going to every user, computer, and thing out there in the world, so a big part of our future is also applying the same algorithms for math onto IOT devices and embedded systems, industrial control, critical infrastructure, so that we can actually predict malicious behaviour and prevent it before it starts. Yes, that is our passion across the company in every corner.

How far are you on the road?

Weā€™re getting there. Weā€™ve been running the company for about four and a half years, weā€™ve been selling the technology for two and a half, but we have over 3,000 customers.

As well as big ones.

Yes, big and small across the board, so we are making our way and we are doing the best we can as quickly as we can to get the word out about the technology that it is the real deal, it really does work exactly the way Iā€™m saying and you just need to try it yourself.

You mentioned Elon Musk, of course being South African. He was born there, he grew up there, he left at 17, but heā€™s quite a big deal in Silicon Valley now.

He is, heā€™s sort of iconic now in the Valley and heā€™s had numerous successes, but heā€™s had numerous failures too and I think thatā€™s what makes him so interesting as a person and somebody that you would look up to as a businessperson because everybody fails and the trick is how do you get back up and succeed after that?

Thatā€™s an incredible story. Have you read the Ashlee Vance biography?

No, I havenā€™t yet.

What do you read?

You know, I read all kinds of stuff, usually non-technology based. My wife is big into nutrition and holistic health and so I try to read works around that in terms of what makes our body toxic, what creates illness and disease based on what we ingest, our food, our fluids that we take in, but also the environment that surrounds us, so Iā€™m fascinated by that because I think that the human form and person is very much like a machine and itā€™s like what you put into it is what you get out of it and Iā€™ve just seen it time and time again, so Iā€™m really keen on learning a lot about that stuff these days.

Hence the discussion earlier about AI being used for that kind of better good in health.

humanoid realistic robotsThatā€™s exactly right, yes. I think we can, as I learn more and more about how the human body works and the chemistry set that we are, there are many ways to predict bad things that are going to happen in your body, but I want to prove it scientifically. I want to prove it from a data-driven approach and using AI I can do it, so someday maybe next.

So fix the computers, and then fix the people.

Thatā€™s what Iā€™m just working on, yes thatā€™s absolutely right.

Just to close off with, everybody in the world knows that there are these two schools of thought that are going on at the moment, the nationalism versus the globalisation, the letā€™s go back to letā€™s go forward, but industries like yours would be almost the targets of the new world Luddites who want to smash the machines so that they donā€™t take their jobs. How do you have a rational conversation, with those coming with that kind of emotional argument?

Itā€™s challenging when theyā€™re that emotionally charged on the topic, but this is what I usually anchor on, is you know, we are bound by Manouche here, weā€™re almost dragged down by Manouche as human beings and what we can use AI for, is yes, to replace human beings for some of the most mundane and laborious efforts in work, but honestly even if we were to get to 99 percent of effective application of AI into manufacturing or any job of any sort, that one percent, AI will never get to and that one percent is the hardest to find and catch. It really requires human beings period.

Take cybersecurity, for example, even if we get to 99.9 percent okay, on our ability to detect and prevent bad stuff, thereā€™s always going to be the 0.1 percent thatā€™s bias and thereā€™s no computer, thereā€™s no technology today or in the very near future that could ever detect that stuff. Human beings are absolutely required to be able to detect that. So I say focus the resources that you would be giving or letting go, turning them into more specialised deep dive experts into whatever problem that the machine is taking over. The machines cannot get to that 100 percent ever, whereas a human being can get much, much closer.

So love the change, embrace it.

Yes exactly, thatā€™s what I try to say.

Stuart McClure is with Cylance.

Visited 58 times, 1 visit(s) today
Categories WEF