Building an AI skills Workforce

Okay, so here's my agenda. There's only two bits to this whole presentation, the scary bit and the intelligent bit. But before we get into ai, I just wanna go back in time a little and talk about another person who changed the way we live and changed industry. Henry Ford. Now Henry Ford is famous for mass production and the production line, and of course the Model T Ford that he created.

But what a lot of people don't know about him, and which is kind of key to this presentation, is that we tend to think about data as being something that's modern and new and that we, it's only kind of come up in the last, you know, 30 years or so. But data has always been around, you know, not just in Henry Ford's time, but for forever. But what Ford was a pioneer in was looking at data and then analyzing it much as we do today.

And he discovered that from his processes, from all the bits that were left off when they made a car, he could actually create new products outta those cars or those bits rather. And that's what he did. But he also was constantly rationalizing his production line and constantly looking for savings to make his cars cheaper. And he was actually, I mean, Henry Ford was not a, a nice man in many ways, but he did actually treat his employees quite well.

I mean, he gave, he paid them well above the, the, the rate that most auto workers got in Detroit at the time. So, you know, good and bad. But what the point is that he understood the power of data that other people didn't. So let's move into today, or rather the future, and talk about ai. You know, this is the question that is asked, has been asked a lot in the last 12 months or so.

So there is kind of the traditional scary version of artificial intelligence, popularized probably by the film 2001, the Space Odyssey, for some of you probably too young to remember that film, but it's pretty famous, Stanley Kubrick. So there's the first connection. Stanley Kubrick, when he made this film, was playing with all sorts of themes, not least of which was artificial intelligence. And you all remember if you've seen the film, the computer that goes wrong, how when it refuses to do what the astronaut asks him to do, because it's developed its own sense of self.

But when you, when you asked about the ending of 2001, Stanley Kubrick said it was about a superior intelligence that took over the spaceship. Basically, he took over the robot, and then the guy at the end, when he's in that rather strange surroundings, which looks like the French sort of 14th century furnishings, is actually quite funny because he said that was the intelligence idea of what a human being would like to live in. So they, which is a lesson, therefore, you shouldn't trust everything that AI can do.

So even this super intelligent alien life force only had a best guess of what a human being from the 20th century would like to live. So the poor guy lived out the rest of his life in that strange surroundings and then was reborn as a baby, which is why a lot of people in the Catholic faith quite like the film because they think it's a spiritual tale. But let's go from that to a test of what you might call artificial intelligence.

Now, I'm gonna pick on someone in the audience now, and I'm gonna ask them a question and are there any Americans here, actually? Oh yes. Right. So we have two in, let, let's go down here. There's one at the back of it. I'll ignore that one. So can you name me all the US states in alphabetical order starting?

No, no, I cannot in alphabetical order, but Well, give me, give me some states. Alaska, Alabama, Idaho, New York, Florida, New Jersey, North Dakota, South Dakota, et cetera, et cetera. Yeah. Okay. California. This one I own.

Yeah, so that's a human being and it's pretty obvious that let's, so let's now ask a bit of a bit of artificial intelligence. So, hi Siri.

Hello, Paul. Hey Siri, how are you? I'm pretty good, thanks.

Okay, Siri, can you tell me all the United States of America in alphabetical order? I found this on the web.

Hey, Siri, can you read them out? Here's what I found.

Siri, can you list for me All on your list? Siri, be quiet.

Hey, Siri, can you tell me all the United States of America in alphabetical order? No.

Okay, So that proves my point. Okay, that is the Turing test, or my version of the Turing test, which stipulates that you couldn't tell a rope a computer from a human when you interacted with them. So clearly a human doesn't know all the states of America unless they've memorized them.

The chat, GPT or Syria in that case failed because the technology didn't work very well, although it did do it in all my rehearsals. It was brilliant last night, he was reading them out like, I, I should know them. He read 'em so many times.

So, so yeah. So this, this is the cheering Turing test of what is considered to be an intelligent computer, or when a, a computer develops sentient feelings or true intelligence. And to date, no one has passed that, certainly not this morning. But you know, the only way that the human could do that is with, you know, take a book, memorize all the United States in alphabetical order, whereas AI will use a generator in discriminator. So just to round off on the sort of scary bit of ai, these photographs that you see are quite convincing at first glance.

So these are completely utter, these aren't in done in Photoshop or anything like that. These are actually generated from scratch. And the guy who's done them has asked to produce some pictures of some women at a party. So if you look at that one, that's not too bad, but let's look at the next one.

Now, first glance, again, that looks quite good. I mean, you know, that looks like human beings having a good time. But have a look how many fingers that woman has got on the camera and then look at their teeth, right? Rather a lot of teeth there.

So again, it's scary that that can be generated, but it's still not quite there yet. So let's get to the intelligent bit, as in why you are here, rather than listening to all that stuff about Stanley Kubrick and what I believe AI in the tools that we have now, as in things like chat, GPT and all the others. How can they, I think this is on autopilot, this, so here are some things that AI can help us with right now. I think malware detection, prevention, phishing detection, all these things are able, or can be assisted by fairly simple AI tools.

Nothing to do with super intelligence, but actually just using these tools to help us do a better job of cybersecurity. Because the thing is, at the moment, AI can't think for itself.

It can't, you can't put an AI machine and say, make sure that no malware gets through. You have to make sure that the AI is given the right data. So AI can only learn from what's happened in the past and it can only predict on world's past patterns. And this is where the Henry Ford piece comes in because all of these tools that we have right now, Siri, when it's behaving, can only give me the United States because it's been learned those in the past and it's got the right data. It didn't give me the states of South America, for example, or, or the counties of England.

So if we use AI as it is now, there is no doubt, I think that it'll give you an advantage in your cybersecurity practice. And if you have that advantage, it's gonna give you an advantage in business. Just as Henry Ford sold more cars than his rivals, because he used the data better than him, if you use AI data and AI tools for cybersecurity, it's gonna be better for the business that you work for. Because we don't often talk about business in these forums.

We often talk about technology and, and criminal action and everything else, but the after, you know, it is actually about running a business and actually making that business safer first, safer first and profitable second, or maybe the other way around. So your advantage will, this will tick, like click off in a minute 'cause they're on some kind of auto queue.

This, so the advantage depends on, yeah, there we go. Unrelenting data collection. And to do data collection properly, you gotta have people, and this is the, the other great thing about AI is everyone's talking about jobs being lost and it's gonna take over the world and all that crazy stuff about, you know, the, the, we'll, we'll be controlled by robots, which is, I think apps a rubbish, frankly.

I mean, you, you could just see how good it is right now. So to use these tools, you need human intelligence. You need human skills to train cybersecurity in business policies and business outcomes. Because without those, they won't, it won't do anything for you. So you train your scientists. So it's creating jobs, it's not actually killing jobs, it's creating new jobs. And that's the people, I call 'em data scientists. They could be called anything in your company, but they're the people that are gonna make sense.

So they could be cyber data scientists, and then after you've trained them, then you can train whatever I ai you've put in place, whatever system it is. And don't forget that every single vendor has now got ai. And this is, this is quite a, a hard slide to read, but you can, this is actually from the Harvard Business Review, just to give it some credence, right? I didn't do this. The Harvard Business Review did this. And it talks about, this is part of a, a whole framework of developing AI in your company.

And basically it just sets something, the skill names, skill descriptors and proficiency levels. So if you use a framework like this, then you establish the rules for how you're gonna look for data, what you're gonna find in the data, and who is going to be responsible for applying that data using ai.

So again, it's about the human factor and the thought process that goes into using tools. So I say if you've got this, one of the, one of the problems with many systems right now is like, like sim systems and analytics and logs and things, is that we actually have too much information. We have so many things to look at that you can't actually see what's happening.

And in, in my presentation tomorrow, just a little plug for tomorrow, I'll be talking a bit more about the software supply chain and how most of it's in is invisible. But if you can't find, like, like yesterday, someone mentioned this new phrase that I've, I've just picked up and I quite like it. Pets and cattle, where the pets are something that you must treasure and must look after. Yeah. So I got it from you.

So, whereas cattle unfortunately is expendable and you don't care for them so much. So you need to find the pets and you need to find the cattle. So there's a lot that's going on in your organization that might seemingly feel like a risk or a vulnerability, but in actual fact it isn't. But you're spending valuable time trying to close it down. So that's why any tool that does the kind of thing that we've been seeing with chat, GBT is gonna help you. So coming right up to date. So Henry Ford's production line now looks something like this. There's not many people in there.

In fact, there is some robots, first generation robots that all car manufacturers now use to do the bulk of their manufacturing. So to, to wrap up, I've come up with this, what I call the cyber ai cyber positive feedback loop, which is really just kind of sums up everything I've been saying. So first you improve the data that you feed into your AI platform, whatever platform it is. Okay?

So this, this, this theory works for virtually any platform. You improve data and then the AI platform gives you better output and then you deploy that and then you get more data from that deployment. And so it goes on. So you should have a virtuous circle of AI and data working perfectly well with your humans. And that's it. Thank you.