Tech-IT-Out: Advancements in AI and Microsoft ChatGPT

In this episode of Krome Cast, Tech-IT-Out, we discuss the advancements in AI, and how Artificial Intelligence has seemingly taken a giant leap forward in 2023, with the development of Open AI ChatGPT and the other AI tools in the market. 

This episode features Krome's co-founders Sam Mager and Rupert Mills, discussing the evolution of AI, the effect of AI on industries and the security concerns of AI and Chat GPT, along with Microsoft's investment in ChatGPT and the exciting ways in which Microsoft ChatGPT will extend its capabilities further!

► ABOUT KROME: Krome Technologies is a technically strong, people-centric technology consultancy, focused on delivering end-to-end infrastructure and security solutions that solve business challenges and protect critical data. We work collaboratively with clients, forming long-term business partnerships, applying knowledge, experience and the resources our clients need to solve problems, design solutions and co-create agile, efficient and scalable IT services.

► KROME WEBSITE: https://www.krome.co.uk/

► SOCIAL MEDIA
• YouTube: https://www.youtube.com/@krometechnologies
• Linkedin: https://www.linkedin.com/company/krome-technologies-ltd
• Instagram: https://www.instagram.com/krometechnologies/
• Twitter: https://twitter.com/KromeTech
• Facebook: https://www.facebook.com/KromeTechnologies/

► CONTACT
• Telephone: 01932 232345
• Email: info@krome.co.uk

Transcript

Sam Mager: 0:24

Welcome to Krome Cast Tech-IT-Out. We're back once again in 2023 for the first podcast of the new year, and today we're talking about the advancements in AI. So Mr. Mill's, can you specifically tell us what we're talking about today?

Rupert Mills: 0:37

Yeah, I think the plan is, everyone else has had their say on ChatGPT so far, so we thought we'd have ours. We're going to talk around the advancements in AI, ChatGPT and others, a specific bit about ChatGPT. And then something about security and the implications of ChatGPT on the security market space.

Sam Mager: 0:52

Cool. So ChatGPT.

Rupert Mills: 0:57

Well, advancements in AI first,

Sam Mager: 0:59

Okay, so I'm just diving in to ChatGPT as to be honest, it sacres the devil out of me. Okay. The whole advancement in AI is obviously a big piece, like you say, everyone else has had their bit, but let's, let's stick our oar into that conversation.

Rupert Mills: 1:10

Yeah, I mean, there's, there's masses coming from it. It's coming really quickly. It's, it amazes me how it's been around for quite a while. But suddenly, it seemed to hit sort of critical mass just before Christmas. And everybody suddenly went, oh my goodness, look what this can do! And yeah, it can quite frankly. GPT 3.5 is out, and it's doing big things already.

Sam Mager: 1:30

It's doing lots of homework.

Rupert Mills: 1:31

Yes, it's doing lots of homework. Absolutely. Yep. colleges, universities, etc. Worried. But there's there's other things out there. I mean, there's there's people using ChatGPT in commercial implementations. The engine behind it is writing, copywriting, so marketing people using it for things like that. There's loads of people using it for things like, I don't know, generating code, all that sort of stuff. We'll come on to ChatGPT specifically in a minute. But there's other things out there. There's, there's DALL-E, which is the imaging one, we've talked around the fact that I mean, I've had conversations with people around the fact that there is concerns for copywriting, genuine imagery and stuff is being so accurately represented as, as an artist specific work, that it is causing challenges right now we, as you know, part own a design business, they've had to already update all of their T's and C's and everything to say anybody creating content using AI can't then try and sell it through the marketplace because they could be creating it based on someone else's style. And all of a sudden, you've got huge issues there. There's things out there. I mean, there's one out there that will do podcasts now, we're going to be replaced doing this fairly soon, it will ease us all over. So yeah.

Sam Mager: 2:39

That's not such a bad thing.

Rupert Mills: 2:40

Yeah exactly. You.. It creates avatars for you, tell it what you want to do a podcast on, it'll go off and create you a podcast. But there's there's editing ones, Kaija behind the camera, there's editing ones that are going to replace her soon, take long form video and edit it down into short clips and things, I mean everything with YouTube shorts and all the rest of it and Tik Tok.

Sam Mager: 3:00

Everyone in this room essentially becomes redundant.

Rupert Mills: 3:02

Yeah, we're all out of a job. Yeah. [Laughter] But there's all of that sort of stuff going on. There's there's ones that will take all your social media and manage those for you, there's one that'll go through LinkedIn, it will sit there and and try and build relationships with people for you. And then, if you're in sales, it will go out and say, right, I'm gonna build a relationship with all these people and the ones that have successfully managed to build relationship for it, I'll hand back over to you, there you go, you can now talk to this person, because they're happy to talk to you. Yeah, and all of that day to day human interaction. So you've got social media, or anti social media, as potentially should be known, is then going to be taken into all these other platforms and the ability to interact. Everybody's been out there pushing all these various different things you need to do to get your name out there. And all of a sudden, AI is going to mean that it's doing it all for you. Where's the genuine content?

Sam Mager: 3:48

So it's interesting, obviously, there's some very practical good uses of AI, especially in our industry, we set up things around security, and (coughs) excuse me, doing thousands of thousands of actions that human couldn't possibly do. And that's brilliant. But I find it quite terrifying. When you talk about the fact that we're joking about people in this room could be replaced, but legitimately 5-10 years, where did those jobs go? Is that a case of these jobs disappear or do they evolve?

Rupert Mills: 4:13

Yeah, yeah, absolutely. I mean, there was conversation years ago around, okay, robotics is coming for the blue collar jobs, the sort of bricklayers, plumbers, etc,

Sam Mager: 4:22

Truck drivers,

Rupert Mills: 4:23

Truck drivers, all that sort of stuff. Truck drivers, maybe, because of self driving vehicles, etc. But all of the stuff that require someone to do something, actually, that's the stuff that's not necessarily going to get replaced. It's the stuff that is task workers. So I mean they're talking about sort of contact center agents and all the rest of it, yeah, a lot of that being replaced by AI. Who knows? It's,

Sam Mager: 4:43

There's some contact centers I've called I reckon it'd be better.

Rupert Mills: 4:45

Yeah. Yeah Possibly. But, but yeah, all of those things out there that potentially could be impacted by AI in the very near future. And it's, it that I think what ChatGPT has done is bring that whole thing that was over here and make it suddenly here and everybody's suddenly paying attention because it's, it's very easy and very real to go out, and go out and do, or create. I remember showing it to you last year.

Sam Mager: 5:08

Yeah.

Rupert Mills: 5:08

And you saying,"Oh, it's fine. I'm not really interested", and I showed you what it could do and you're suddenly like "Oh, my goodness, hand on." So, and I think that's been the general reaction of the general public.

Sam Mager: 5:16

It's Skynet

Rupert Mills: 5:17

Yeah, it is, absolutely. Yeah. I mean, there's or if you look at Boston Dynamics, you've got the dog that Boston Dynamics has done some amazing stuff. There's a couple of guys in the US who have who strapped assault rifles to the back of them and gotten patrolling their houses on security duty. That worries me. The software, misidentifies you coming home from work, and you've got a problem?

Sam Mager: 5:38

Yeah. There's real dangers, obviously. Which, yeah, but it's going nowhere. That's the thing, it's not going to go, let's just not use it.

Rupert Mills: 5:47

It's not going to stop. We've we've gone past that point.

Sam Mager: 5:49

Yeah.

Rupert Mills: 5:50

Yeah. I mean, so if we, if we focus in on ChatGPT, which is, obviously Microsoft's main push to market at the moment

Sam Mager: 5:58

Its a huge investment

Rupert Mills: 5:59

They've invested in billions and billions, absolutely billions. And but they're starting to bring it in commercialise it, it's not been commercialised till this year, you can now buy it in Azure marketplace. So if you're using Azure to do various different things for you, you might be running your website, doing an analytics in the background, there's been the various different tools that you could use in the background to analyze things, even able to do speech analytics, you've been able to look at OCR, all that sort of stuff. But all of a sudden, you've got AI, full on AI in there, able to do things. And it's at a really, really, really cost effective rate. I mean, it's, it's fractions of a cent per 1000 credits, 1000 credits about 750 characters of text. So you can start to chuck it at what people are putting into anything, and read it and understand it and do things with it really, really quickly

Sam Mager: 6:47

It's that automation piece. So take this and automate, etc, etc.

Rupert Mills: 6:50

Yeah, absolutely. So that's coming really quickly. I mean, Microsoft are doing it with Teams, if you look at Teams premium, which is out there now. Essentially, they're taking minutes from your meetings, automatically creating tasks for, for people. So you can say out of this meeting, these are the tasks and these are the owners, etc. And they're doing a lot of other stuff with it, like creating templates for future meetings. So saying, Okay, this is all the stuff you discussed, and you've got a meeting coming up.

Sam Mager: 7:15

So this comes next. Yeah. And this comes next. And that sort of stuff is just, the first iteration of it. The pace of it, like you say, the pace that's happened, and that's brilliant. But you just think about the amount that.. I listening to that going, that's great. That takes X amount out of my day, but it also makes my PA redundant to a certain extent.

Rupert Mills: 7:34

Yeah.

Sam Mager: 7:34

But then it's obviously that's say the evolution, people's jobs, I guess will evolve. You've got to have a level of, of overwatch.

Rupert Mills: 7:41

Yeah.

Sam Mager: 7:41

So who watches the watchman type stuff, right? Someone's got to actually look at that and go, is that? Correct? So the day to day? Excuse me? I guess that the churn of the role, changes, but someone has to make sure that's apt and onpoint and doing the right things, right? So does it just, does it help by just taking out the the mundane elements of what we do, given us time to focus, laser beam on the really important bits, or should we all be terrified? It's a bit of both I think.[Laughter]

Rupert Mills: 8:06

I think, right now, there's a load of security built into it that is supposed to not let it do things. But we know that people are out there using it for malware generation, they're using it to, I mean, you can go and get it to create some code. If you don't tell it what you're doing, and you get it to create a piece of code here, and a piece of code here and a piece of code there. It can speed up software development massively by then saying, okay, someone wants to creates some malware and they take those three elements and put them together, they've just got ChatGPT to do most of it for them. I mean, we, our development team, I was chatting to them the other day, sometimes if they're kind of, okay, how can I make this work, they'll ask ChatGPT and it will come back with a piece of code, normally doesn't work first time. So you still need the skill, but they're okay, I can now take it and get it to do this, or I can get it to do that. And they tune that.

Sam Mager: 8:48

That's my point, it's saving the

Rupert Mills: 8:49

Takes two hours work, into 20 minutes

Sam Mager: 8:51

There you go, I cheated and used it for an Excel formula, very basic, the other day, should have been able to work out myself. But I chucked it in there because it gave me the answer immediately, rather than me sitting there trying to work out. if that's and others. It would have took me half a day.

Rupert Mills: 9:03

Yeah. That's it. And you look at what Microsoft is doing with it. They're about, I think they might even have done it already, adding it into Dynamics to try and take on Salesforce. They've got Google really worried because they're talking about adding it into Bing and search. It's the first time Google have seen a proper threat to their search market space in years.

Sam Mager: 9:19

They've not been challenged at all, ever, have they.

Rupert Mills: 9:21

No

Sam Mager: 9:21

We discussed this the other day, when you think about Google you chuck, tell me about, or find and you just get a plethora of page after page after page you then have to sift through and work out where's what I want to know. Yeah. Whereas ChatGPT you say tell me exactly. And it goes - this!

Rupert Mills: 9:33

Yeah

Sam Mager: 9:34

Job done.

Rupert Mills: 9:34

Exactly.

Sam Mager: 9:35

Okay, so let's, I think we probably need to, this is all quite scary. Let's delve more into the scarier element of it, how people are really leveraging that and what if anything can we do?

Rupert Mills: 9:45

Yeah I mean if you get into the security side of it, first of all, you've got whether the contents real so you've got what they call AI hallucinations, which is essentially you get to the point where, there was a case last week week before of an artist who basically asked ChatGPT if he was dead, and ChatGPT said yes. Because it is understanding what's out on the internet. And he tried to argue with it. No, I'm not dead. I'm actually talking to you right now. Well, yes, you are dead because here's an obituary I read of you. And another obituary over here. And there's some evidence here, here and here. And at the moment, AI is hallucinating about real things, because there's enough evidence out there in the internet. So that's dangerous in the first place. Because people will take whatever they read from these things as gospel. So you need to still use your your brain to filter it a bit or do your own research, even though it can speed things up and give you a lot of stuff very, very quickly. You've got to be able to say, is that real? Or is it an AI hallucination?

Sam Mager: 10:37

There's still the human element. Right? So that's, someone still has to stop and check that and go, are you? Correct?

Rupert Mills: 10:43

Yeah.

Sam Mager: 10:44

Someone with a conscience.

Rupert Mills: 10:45

Yeah, with a conscience. Absolutely, that's exactly it. And that going to, I'm sure, be filtered. And I think it's going to get better and better and better, they're gonna get better and better about finding out real things. Also linking up to real data sources, because there's only a certain amount of data in it at the moment, as they link it up to more and more real data sources, you can imagine government's opening up things like, so if someone's dead, well, let's open up a death register to it, it'll know. You'll be able to go, can I trawl this, can I find, anything, when you think about the data is open source, essentially, at the moment, there's so much data out there under the Freedom of Information Act, you start requesting all of that data and feeding it to the AI, it's gonna get infinitesimally smarter. But that brings its dangers with it. And that part of those dangers are things like, the people using it to for malware at the moment, for phishing, specifically, you can be really targeted with phishing, if you've got an AI engine that can go and learn about someone learn about what they're doing. And then specifically, okay, so I know you are this person, and you do this, rather than a mass would you like to or

Sam Mager: 11:44

You've just transfer 10,000 pounds to this account? Because Bob said so type thing

Rupert Mills: 11:48

Or the usual,

Sam Mager: 11:49

Way more targeted,

Rupert Mills: 11:49

Your sister told me that I need you to get me$100 Amazon card or whatever, I don't have a sister, that's... well I do. But in those situations, that's that's the situations where you can say, okay, AI will make that infinitesimally smarter, because it'll learn about people and go, your brother so and so who's traveling here at the moment, have contacted me and said

Sam Mager: 12:08

In this timezone, etc, etc.

Rupert Mills: 12:10

Yeah, and it'd be smarter than that. But people are using it for that sort of thing, which is a concern, along with the code generation, and then impersonation. ChatGPT is going to allow people that have relatively little knowledge to get out there and impersonate people, and to know about people and do a lot more very, very quickly. So there's a lot of good it can do, but there's a lot of bad it can do too.

Sam Mager: 12:32

Well, I guess in summary, that is, should we be terrified? Maybe! [Laughter]

Rupert Mills: 12:37

Yep.

Sam Mager: 12:37

But not too much as yet.

Rupert Mills: 12:39

Yeah, I think you've got to adopt these things and say that you look at social media and how fast it came into the world it exploded, and it made a change to everything everybody does every day.

Sam Mager: 12:48

Yeah.

Rupert Mills: 12:48

Is it all good? Certainly not.

Sam Mager: 12:50

No.

Rupert Mills: 12:50

Is it policed at the moment? Certainly not. AI is kind of the next, the next wave. If you think about the internet coming on in the 90's, social media coming on sort of 10 years later, then we're 10 years on from that, AI is the next phase, and we'll be looking back at it in 10 years going, okay, was that good? Or was that bad?

Sam Mager: 13:08

We won't be doing this in 10 years. It'll be automated. But let's step back in a year. I'd be interested in 12 months time to see how we're all using it, how it's adopted. Whether we are sitting here doing this or it's two avatars, whether Kaija is sat there or we've automated it.

Rupert Mills: 13:21

Yep.

Sam Mager: 13:21

It'll certainly save us some time on a Monday morning!

Rupert Mills: 13:23

Indeed!

Sam Mager: 13:25

Thanks very much.

Rupert Mills: 13:26

Welcome.

Sam Mager: 13:27

And thank you for joining us again on Krome Cast Tech IT Out. Please remember to like, subscribe and share and use ChatGPT to leave comments in the section below and we'll try to put them into the next episode. Tech-IT-Out