Picture this - what if you could have a deep and intellectual conversation about the intriguing world of Artificial Intelligence? That's exactly what we do this week as we engage with Dr. Chris Harding. This conversation unravels the role of AI in education, the power of AI and its application in various walks of life. Dr. Harding emphasizes the essence of maintaining a balance between using AI and understanding it. He helps us explore the vastness of AI beyond the realm of education.
Ever wondered about the sophistication of models like Chat GPT, the natural language processing system that uses billions of probabilities for its output? We get to hear Dr. Harding's insights on this marvel of AI, and the possibility of its evolution in the future. We also delve into the potential of AI in processing data structures as he shares details of his ongoing projects. Plus, we touch on the use of AI in web queries and the organization of search results. He underscores the indispensable human element in AI and suggests ways to balance individual and collaborative work for optimum productivity.
Are you ready to kickstart your journey in the tech sector? Dr. Harding has got some priceless advice for you. Tune in to hear his unique perspective on time-travel and his favorite invention - the computer. We also take a sneak peek into what's lined up for our Open Comments Community. So, buckle up for this intellectual roller-coaster ride through the fascinating world of AI with Dr. Chris Harding.
Copyright © The Open Group 2023-2024. All rights reserved.
Picture this - what if you could have a deep and intellectual conversation about the intriguing world of Artificial Intelligence? That's exactly what we do this week as we engage with Dr. Chris Harding. This conversation unravels the role of AI in education, the power of AI and its application in various walks of life. Dr. Harding emphasizes the essence of maintaining a balance between using AI and understanding it. He helps us explore the vastness of AI beyond the realm of education.
Ever wondered about the sophistication of models like Chat GPT, the natural language processing system that uses billions of probabilities for its output? We get to hear Dr. Harding's insights on this marvel of AI, and the possibility of its evolution in the future. We also delve into the potential of AI in processing data structures as he shares details of his ongoing projects. Plus, we touch on the use of AI in web queries and the organization of search results. He underscores the indispensable human element in AI and suggests ways to balance individual and collaborative work for optimum productivity.
Are you ready to kickstart your journey in the tech sector? Dr. Harding has got some priceless advice for you. Tune in to hear his unique perspective on time-travel and his favorite invention - the computer. We also take a sneak peek into what's lined up for our Open Comments Community. So, buckle up for this intellectual roller-coaster ride through the fascinating world of AI with Dr. Chris Harding.
Copyright © The Open Group 2023-2024. All rights reserved.
Welcome to Open Comments with myself, Ash, and me, Oliver, when we'll discuss things openly with our guests from a variety of backgrounds and from different walks of life. Through this podcast, we hope to give you an inside look into a variety of topics with an equal mix of humour and candour In this series so far, we have touched on the following topics healthcare, HR, diversity, access to technology, cybersecurity and lots more. We hope you enjoy our show and look forward to bringing more topics into the fold. Let's get started.
Speaker 2:Thank you, ash. Today we're very pleased to welcome back Dr Chris Harding. For those of you who have listened to the podcast, you'll know Chris from our very first episode. For those of you that may not know, dr Chris Harding is the founder and principal of Lakeabus Limited. He formed the company to provide services based on virtual data lakes and data-centred architecture. Chris developed the ideas that led to the formation of the company whilst working as director of the Open Group Forum, open Platform 3.0. So, chris, welcome back. It's great to have you back on the podcast.
Speaker 3:Thanks, oliver, it's a pleasure to be back.
Speaker 2:So today we thought we'd start on the subject of AI. It's a very trendy subject to talk about right now and, chris, you are a man great for the job.
Speaker 3:Oh, thanks, Oliver. Ai, I think, is a fascinating subject, and Lakeabus is working on deploying practical AI applications on its data platform, so I'm very happy to give you my two cents on that topic. Perfect Okay.
Speaker 1:So, to start things off, you recently wrote an article for the Open Group Blog called Natural Language Processing for Standards Development. Please can you explain natural language processing with the inclusion of AI in a nutshell?
Speaker 3:It's a lot easier to do that since chat GPT came on the scene. Chat GPT is essentially a natural language processing system. You ask it questions in natural language. It gives you answers in natural language. So the idea of associating this with Standards Development is that standards obviously are written in natural language and people working on standards need to understand what's in them to be able to manage the intellectual content of the standards, and that's where I think that AI can help them in a number of ways.
Speaker 1:Thank you, and speaking more about AI, how can you see AI being used in education or becoming a subject matter?
Speaker 3:I think AI is the generation of AI that is used in education and the generation of AI that we have now with chat. Gpt is very good at what you might call associative reasoning. It's very good at detecting patterns in things and it can be quite creative in building on those patterns. It's not good at the formal logic side of things, so it makes mistakes quite often with what people would regard as common sense deductions. In the field of education, it's proving to be a bit of a double-edged sword. On the one hand, it has the capability to help teachers organize their training material. It can be used, in fact, to generate exam questions and to mark exam questions and to mark the work of the pupils. On the other hand, we're finding that pupils can use it to produce that work, and one of the things that education is going to have to get to grips with is how to teach people to use this kind of natural language, processing artificial intelligence.
Speaker 3:It's not necessarily a new challenge in kind. I can remember the days when there was a lot of debate about should students be allowed to take calculators into their maths exams or should they be allowed them even in their maths lessons, and we've reached, I think, the resolution of that as to when it's appropriate for them to use calculators and when it isn't, after quite a few years, and we've got to reach, I think, the same resolution in the field of AR when is it appropriate for the students to use AI in their work and when should they be doing it all themselves? But putting that aside, I think there is a large amount of scope, as I said before, for the use of AI in organizing training courses and in presenting training material and in evaluating the degree to which the students have assimilated that material.
Speaker 2:So it's very much like your analogy with the calculators in maths. You can definitely tell that it will help with productivity and you may get things done quicker, but you need to know when to use it and when not to use it.
Speaker 3:Exactly. I mean, nowadays you would expect anyone in a business transaction who needs to do some adding up or taking away or whatever, to bring out their calculator. Actually, they probably do it on their mobile phone now, but so that's an accepted. It's a much more efficient way of proceeding for them to get out of pencil and paper and do it all by long multiplication, and they've got to be taught how to do that. But they won't understand what they're doing in doing that unless they've done a bit of long multiplication on paper first. And I think we've got to get to the same point with artificial intelligence, where people use it at the right time when it is the most efficient way of doing something, but they understand what they're doing when they use it.
Speaker 2:Yeah, you need to have the skills and knowledge first, otherwise I guess you wouldn't be able to tell if it's the wrong or right answer that you get.
Speaker 3:Right, and that is, at the present time, something that is very important. With AI, the natural language processing things like chat, gpt, they cannot be relied on to give you the correct answer every time. It still needs some human input and that's partly, I think, because they're very strong on the associative reasoning side. They detect patterns that perhaps aren't necessarily the appropriate patterns and they do still need human input to make sure that they get it right.
Speaker 2:So you mentioned that it's about adapting and learning how to use it. Do you see any way that we can use it in any walk of life? How can we adapt to embrace AI?
Speaker 3:Absolutely. I think it's quite routine now for people in some of the creative industries, for example graphic designers rather than starting by sketching out a lot of ideas, they might ask Dali or McJourney to sketch out the ideas for them, and then they might pick and develop from what looks like the best one. Equally, I think, in producing a written copy, I think quite often people and I've done this if I wanted to get something written, get chat GPT to give you its thoughts on it and take that as a starting point. It's also good at adapting what you produce for a particular audience, and that's I mean.
Speaker 3:The extreme example of that is something which is now quite commonplace, which is language translation. You know, you have a piece of text on a webpage in some language that you don't understand. You can copy and paste that into Google and it will come up with a translation. The less extreme form of that is you write something in the words that you understand, but you want to present it, say, to a different audience, maybe an audience of children or an audience of a particular kind of professional, and you can ask the natural language processing tool to revise it in that format. So I think there are a lot of ways that AI actually is being used at the moment, in the natural course of events, for practical purposes.
Speaker 1:And going back to the article that you wrote for the Open Group blog recently, please can you tell us a little bit about how chat GPT compares to Eliza from the 1960s?
Speaker 3:Okay, so that is a very good comparison, and I think, essentially, it is Eliza on steroids. I mean, eliza was a very primitive system that worked by capturing the input and kind of picking out key phrases and repeating them with variations and questions and building on them in a fairly simplistic way. Chat GPT is, I think, not really different in kind. What it has is a very sophisticated way of looking at the input that it's given and predicting what would come next after it. So, obviously, if you ask it a question predicting what comes next, you are predicting the answer, and a simple form of this technology is routinely used in mobile phones. You start typing a word in, and it comes up with suggestions for what you wanted to say, and those suggestions are based on probabilities of what will follow the letters that you've just put in, and chat GPT is similarly working on probabilities of what words would follow the words that have been put in.
Speaker 3:The thing that has distinguished it, though, is the scale of the model and the amount of training, so you're talking about, instead of, a sort of little probabilities look up table of a few hundred entries. You're talking about billions of probabilities, and you're talking about training to establish those probabilities with millions or billions of inputs and outputs, and in some ways I think it's surprising, given that it's really. It is a sophisticated model, but it's quite a simple thing really. I'm sure it's simpler than the way the human brain actually works, for example. It's quite surprising how good it is at getting something useful on that basis.
Speaker 2:So, in your opinion, is chat GPT the best of the best so far, and do you think that there will be a new version of chat GPT a few years down the line?
Speaker 3:There are new versions coming out all the time. I think we've reached a new level with chat GPT and I think that for the next few years. Obviously I could well be wrong on this, but my personal feeling is that we won't see a step change over the next few years, but we'll see incremental improvements. Chat GPT is actually less powerful than some of the other models around, but what I think has made it stand out is the degree of training that it's had on understanding the input. So it's better at understanding what the user is trying to do before it then goes and develops its probable output. So, but yes, I think it has reached a new level over what was there previously and I think we will see improvements in that level. But I don't see us getting to another, much higher level for another few years.
Speaker 1:And now moving on to lifelong learning, are there any projects that you are looking forward to starting on?
Speaker 3:There are a number of projects. I think I'm looking forward to exploiting this new level of natural language processing to produce practical, useful solutions. On the one hand, in Lake of Us, we're looking at using the technology to get an understanding, if you like, of the concepts and ideas in webpages or documents so that people can process them at a higher level than word processing. And in the open group, I'm involved in the data integration work group and we're looking there at using natural language processing to assist in the work of the group, and we're also exploring the possibility of doing that in conjunction with other parts of the open group. The open group has in fact got, I think, a much bigger data component than it used to. Things like the Face Consortium is kind of about data standardization across companies in the aircraft industry.
Speaker 3:Osdu is about data standardization and integration across companies in the oil and gas industry and, I think, most excitingly, open Footprint is about data standardization and data integration of footprint data across companies in all industries, and that's a particularly exciting area because it's so huge, it's obviously so important and because it's at such an early stage, the practical standards don't yet really exist and that's what the Open Footprint Forum is working on, so I'm hoping that we may have some collaboration with them on how natural language processing can assist and how their work is done, so how their work relates to data integration as a whole and is a good example of it.
Speaker 3:There's also, I should say, although I'm probably not going to get directly involved in this there's an exciting field opening up where people are using artificial intelligence to process not so much natural language but data structures, and that's an area where we may see AI help with data integration by doing the kind of job on data tables that chat GPT does on text documents. So there are a lot of exciting things, but what I'll be focusing on is, as I say, the practical application of natural language processing within web pages and text documents to try and enable processing them at the level of concepts and ideas rather than the level of just the words. Thank, you.
Speaker 1:Now the next question is a little bit similar, but it's can you tell us a little bit about what you're most looking forward to? I believe we just touched on that, so maybe expanding with you know what you just spoke about, but um, it could be within your career or just in general what are you most looking forward to?
Speaker 3:Uh, what I'm most looking forward to and I hope that that's not too far off now is launching a product which will, as I say, give people practical assistance with organizing um textual knowledge. We've done quite a lot of preparatory work in various places and got some prototypes um, but it's turning this into a product which is really quite an exciting process.
Speaker 2:How do you stay on top of current trends? You know, how do you keep yourself knowledgeable with an ever changing industry such as IT?
Speaker 3:Well, that's a very good question and of course, ai can help with this. So if you use uh, you could use um, the, the, the, the sort of new generation of a browser, certainly from Microsoft and um probably from other companies too in in the near future. Um, you can browse the web in an intelligent way. Uh, it, the, um. It uses um natural language processing to understand what you've asked it to do and to formulate uh web queries. That will um bring back search results, and then it uses natural language processing to organize those search results, so that actually speeds up the process of finding out about new developments.
Speaker 3:Um, I do tend mostly to work with web material, um for the latest stuff, um, I feel that um, it's going to take, if you, if there's an important new idea, it's going to take between one and two years before a good book comes out on it. Meanwhile, you've got the, you've got the web stuff which you have to digest, uh and uh. So search engines and now um AI driven search is is a good way to keep up with what's going on.
Speaker 2:Thank you. So, yeah, it's. It's just constantly researching and staying on top of everything is is how you're basically staying ahead.
Speaker 3:Well, I hope I'm. I thought about staying ahead. I hope I'm at least keeping up Um, it's it. But yes, there is. There is so much out there Uh, and I think that is one of the most useful things about AI, in that it can help you to cope with uh a large amount of information uh and extract what might be most relevant uh and help you to understand what that is. But the other aspect of it is um. I do gain uh a great deal from participating in um. What is uh, such as the data integration work group um, because uh AI is all very well Um, but there has to be, I think, a human component uh of of what you're doing, so that you, you, you, you get the, if you like. It all has to be, I think, in a human context. You can't um purely work in a um in a data context and then expect that to be useful to people. Thank, you.
Speaker 2:We've talked a bit about how you, how you keep up with uh trends and technology and and we know you have uh uh uh accomplished background within the technology sector. Um, what advice could you give to those just starting? Either either they've just graduated or they're looking to change fields into the area that you work within, Is. Is there any advice that you may have to someone that's looking to start out?
Speaker 3:What I would say is that Computing remains a very exciting field. It was an exciting field when I got into computing a very long time ago. There's been constant development since then in the power and scope of the work and the recent developments with chat, gpt, for example, show that those developments are continuing and I believe they will continue to develop. So for someone starting out, I would say there's a lot. There's a lot of interesting things there. Enjoy it. I think would be my first piece of advice. You know, get your teeth into it, try a few things. Don't be afraid to experiment with different aspects of of the technology. Keep learning. You have to keep learning because it's changing so rapidly.
Speaker 3:But you can do that. It's easier to do that with the amount of information that's now available over the web. But, as I said previously, do it in the context of human interactions with your peers and with people who've been in the, in the industry for a while. You know, and participate in discussions with them, maybe in discussions in standards bodies, in the open group and so on, maybe in just informal discussions. And, as I say, enjoy it. You know it's a great field. It's a great field to be working in and there's a lot that you can get out of it.
Speaker 2:Yeah, so the collaboration between you know, human to human, is a great way of sharing and gaining information and constantly learning by talking to new people.
Speaker 3:Absolutely. I mean, I think the way to look at something like chat GPT is it's a part of the team, it's a member of the team, but the team is essentially a human team and it's providing assistance and, yeah, that's the environment that you should be looking to work in.
Speaker 1:Thank you. What does collaboration mean to you and how important do you think it is to continually collaborate, and how does that affect you? Know how you work or what you learn from others, and you know how you see projects growing.
Speaker 3:Well, there's got to be a balance, I think, between collaboration and individual work. So, yes, you are, you are a part of a team, you are collaborating in a joint activity, but equally, you are providing your personal input to that team and there will be times when you need to explore an individual topic that you believe is important, that maybe doesn't, isn't something that the team as a whole is buying into. You've got to balance that. You will get, you know, you will get the. The end result will be derived from collaboration, but you may have to think individually as you contribute to that.
Speaker 1:Thank you Now, before we end, we'd like to start a short round of quick fire questions, so I can start things off with the first question. So if you could time travel anywhere right now, where would you go?
Speaker 3:It's a very good question. I don't know. I think I'd go. I think I'd go 100 years in the future so that I could see how people at that time are thinking about what's happened today and I could see how things that look to be important today, how they've shaped up and are they really that important over time?
Speaker 2:Thank you. What would you say is your favorite invention that you use?
Speaker 3:Oh gosh, no, probably the computer. I use that quite a lot. I don't really even think about it, but that's probably. I would say that's certainly the invention I use most and probably my favorite one.
Speaker 1:Nice. Thank you, chris, for joining our show again. It's been great catching up with you and also learning more about AI and the technicalities behind it, but also a bit about its history as well. Now to our listeners, also known as our open comments community. We look forward to bringing even more topics and special guests into the fold for you very soon. Please stay tuned for more to come. Thank you, stay safe.