We are going to jump straight into our presentation today. We have three of our lecturers with us today. I'm going to get them to jump on to their cameras, and turn themselves off mute one at a time, and introduce themselves. So I'll not talk too much. I let them do their thing. You're here to see. So we'll start with Bernard if that's OK. If we can pass you straight over, and then you can just jump through and pass on to the next as well.
Bernard Hi, everybody. So my name is Bernard Roper. Let me tell you a little bit about myself and my background. So I have been a data scientist for most of my adult life. I've got a highly checkered past where I had far, far too many day jobs. I think the longest one I had was probably as a timber fellow, working for the National Trust and Forestry Commission doing specialist storm damage clearance. So I had an accident doing that, and then went to University as a mature student. Then I trained as a teacher, and worked for a TV company and did freelance web development and all sorts of other things.
And then eventually taught my way onto a funded master's degree in computer science. I didn't have any programming skills before I went on that, and I had to learn very quickly. So I spent a year not sleeping very much. And now I do research at the University of Southampton in conjunction with the ordnance survey. And I work on a project to convert provenance data into from OpenStreetMap into a form, which can be understood by machine learning algorithms, which is what I do when I'm not teaching for the SDSA. So that's me.
Hi, everyone. I'm Manuel. And I'm a lecturer in the computer science department of the University of Southampton. I am also the head of learning of this venture of Southampton Data Science Academy. And I have research interests in data literacy and also learning technologies and online learning. And what I do for the SSA for the Southampton data Science Academy is a bit of Learning Design, put the courses together, and recruit and train the tutors, and design the way we deliver those courses and the way we prepare the materials for those courses. So this is me.
Hi. My name is Carlos. I'm currently a Marie Curie Research fellow in explainable AI and unfairness. I started my journey as a data scientist after I got my Bachelor in physics and I got interested in predictive model. I started working in the National Meteorological Agency of Spain. Then I went to Deloitte as a consultant, where I help companies to deliver the machine learning and then the AI products. Then I moved to the European Central Bank where we're working finance in the public policy sector.
We tried to apply machine learning. We were doing machine learning during the COVID crisis that was very interesting. And now, I'm back in the academia doing what I like the most, which is research in machine learning and its implications in society.
I think these three courses, I think they're probably one of the most important things they do is help you bridge the gap between people who are knowledgeable and people who are practitioners. So you've got the spectrum in AI and data science. So in data search, you have at one end, people who are developers and researchers like myself and Manuel and Carlos. And at the other end, you have people in organizations with business problems.
And they need those translating into something which the practitioners and researchers can understand to design the systems. And a really important part of that process are people who can bridge that gap. And so that increases your importance to your organization dramatically because it helps you have those meaningful conversations. So that is really vital. And the other thing I think is that we help you get the kind of curiosity and the confidence to learn more. It's a really gentle route into some quite advanced data science techniques.
And it dramatically increases your importance to your organization. So yeah, it's very important for your career, or profession I would say.
Yes. And I would add that there are some people who call it the hybrid economy where there is this need for a skilled workforce with mixed abilities or mixed backgrounds which is your specific background in the specific industry where you are in on the specific knowledge that you can bring in and in combination with data science knowledge. Of course, you are in these courses, this is it is not a degree in data science. You will spend some 40 to 60 hours. So you can only learn up to a point.
But the way we have designed this is that you-- mainly I call it you will speak the language of those data scientists and developers who develop those applications or who handle those applications. And yes, you in some of the courses you will even get your hands dirty with data and you may even type a few lines of code. But it is just only to empathize or to see what is involved in developing in any data science application or any stage of the data science process from handling the data, from obtaining it, from to analyzing it or even visualizing it and telling stories about the findings that you have achieved by going through the data.
So this it's basically this middle ground this transition where we will be-- that it's in very high demand. It's in very high demand, not only within your own organization for maybe, either promoting or changing or doing new things in your organization, but also to either change your organization or if you are job searching to demonstrate that, yes, you can speak that language.
First, the best way to distinguish these two is to define them. Data science, contrary to what many people believe, it's not only about data. The important word for me it's science is the application of scientific methodology to the data. And this change of paradigm, so people used to think to apply administrative process to data. Now data science has been so revolutionary, we're going to apply this scientific methodology. When applying this scientific methodology, one of the tools that one can use in order to extract insights and deliver value is AI.
AI is a prediction program. It's a learning program where the goal is to create machines that are able to take decisions to predict. So that's why I think that the math for me, the main difference is that science is the application of the scientific methodology, the data. And AI is just one technique, one tool that you can use in the science.
So yeah, building on that. So AI, we call it AI because it refers to types of computations that let machines display human like behaviors. So they do calculations in a human way. So for instance, you might teach a machine to tell the difference between a picture of a puppy and a picture of a turtle, but what you're not doing is turning that machine into something sentient. So the machine still has no idea what a puppy or a turtle is.
And a chat bot for instance, it will reply it'll have a conversation with you. But it doesn't really understand, it just has this massive database at the back which matches patterns and it has loads of speech features so it knows how to respond. But it doesn't really understand. So that's AI. It's a kind of very complex way of analyzing probabilities and matching patterns that can make it seem human.
And data sciences is something which includes that but you're looking at studying all the other things you can do with data. So the data gathering and cleaning and research and statistical analysis and how you manage your data, and collect it. And as important, if not more important than all the other things are the ethics in governance considerations. So ethics is increasingly becoming the bottom line in data science and AI just because when organizations get it wrong, it can be so disastrous. So yeah, it's a very broad topic.
Yes, and more specific to the courses we offer, you will see that there are different titles to our courses. One of them is AI in business. Artificial intelligence and machine learning for business. And then we have the fundamentals of data science, technical, and then we also have the fundamentals of data science non-technical.
What is the main difference? Basically, in the Artificial Intelligence course, what we entitled artificial intelligence, we will also consider data science topics and issues, yes? But the main focus is to have an overview of the latest and most common artificial intelligence methods to solve business problems.
So we will look this from a business perspective or from an industry-specific perspective, whereas, in the data science courses, the courses that have the data science title, we will look much more at the data science techniques that are used actually across the board. But it is not so in business-specific. It is more kind of opening the bonnet of the car. It's the analogy that I use usually.
It doesn't mean that the data say our data science courses do not consider AI. They actually do. But they consider AI from the data in itself, whereas, the AI-specific course for business, we tackle AI from the business problems. So these are the different perspectives that we use in the different courses. And this is the way we specialize the content of those courses.
So the skill set required will depend on the course that you join. So yeah, we will start with data science technical course. If you have been using, say, data science applications and of you have a bit of experience in coding, that will be an advantage for you in the technical course.
It does not mean, at all, it does not mean that without coding, you cannot do the technical course. Actually, there are participants who never wrote a line of code in Python. And then they succeeded in the course.
The only difference is that it will take you a little bit more time. You will have to go through a few more materials. You will have to watch the videos maybe twice, some of them. But other than that, the skills required is that you have been considering or you have been handling data in a on a day-to-day basis.
Then for the AI course, well, there are no coding skills. There are no technical skills required. However, it will be also good that you are working or you have been or you have been involved in data-driven projects where you have had to handle data. Maybe not yourself, but you have been involved in a team that they are considering business problems with any artificial intelligence method. Again, if you come from scratch, the only difference if you come from zero, if you have done nothing at all in that respect, the only difference is that you will have to devote a little bit more time to complete the activities, the exercises, and the assessments of the course.
And then we have the data science non-technical, which is the most introductory course. And in these courses, you will only need to bring your interest and be keen on, again, discussing data science, data-science-related issues with your peers, read quite a bit, and go through the materials that we offer. But again, the skills that you have to bring will determine the effort that you will have to do in the course. The more experience you bring, the less extra time you will have to spend in our course.
What we guarantee is that, no matter what your skill set is, you will learn something new. And see data science and artificial intelligence from new perspectives and from new angles. That's our experience. And that's the feedback that we have from all the participants in this course who come from very different, very diverse backgrounds and with very different levels and skillsets. Bernard, anything to add to this?
BERNARD ROPER: Yes, I think one of the things I would add, and it comes from my own experience of having managed to parachute myself into a computer science software engineering master's degree with no programming skills at all. And I can remember what a painful and frankly terrifying experience that was. And so I understand what people go through when they're confronted with code and how important it is to have that learning experience presented to you in a non-threatening way.
It's really, really important. And it's something we're very good at. We we've got a very good balance of a kind of gentle slide into some very technical stuff. So on the nontechnical courses, you can get away without doing any coding at all. But it's there.
And on the technical courses, it's a gentle easing into the process, which is very important. And the other thing is that you will have access to tutors, personally. We have online seminars where we talk you through these things .
And I'm trying to recall if anyone has ever failed. I don't think people don't. I mean, people sometimes struggle. But we're always there and we get people through. So the material arrives in a very nonthreatening way. And people surprised themselves by how much they learn at the end of it.
Myself, basically, I have learned the science after University. And I have seen many colleagues attend to these courses and then doing many things with data science with the knowledge that they get. I have seen people gaining insane amounts of money, getting promoted, changing companies.
For example, one of them, as an example that has happened in these courses, one of the students that was working in the pharma industry decided to pursue a project during the six weeks. And he was getting constant feedback from us, from the tutors. And this project was analyzing some data that had to work, depend on [INAUDIBLE].
The question that he has is like, did the positioning of the articles in the pharma, from the pharmacy, impacted the sales? And after some weeks learning on these courses and after he was able to get some insights and noticed that rearranging the articles that were in the pharmacy could improve the company in the business, with the help that with what he learned from these courses, he was able to provide this to his managers in the company. And I think that he's doing much better now.
MANUEL LEON URRUTIA: Yes, can it provide much more? Yeah. So basically, this question is better answered with examples. And a whole company, big company, yeah, no, I'm not going to say the name. But it's a big banking and insurance company.
They send some of their staff in the change program department. So they did the AI course. And there is an exercise, a big exercise that you have to do in this course where you have to consider all the aspects and all the challenges and opportunities of applying an artificial intelligence method in a department of an organization.
And the learnings that they took from there, they just reproduced them back there in the company, in their organization, and the whole structural change of and the whole purchase of artificial intelligence products was based on the exercise that they did in our course where you have to consider many things, not only the data that you have but also the maturity of your organization. How ready is the staff? What are the skills of the staff in your organization?
So basically, you look at very many different aspects of applying artificial intelligence for any transactions and for any things that you go in there in the company. And the reason is that our courses are very practical and very based in real-life examples. Of course, we don't do consultancy for you.
We will not we will not go through your whole company. But one of the exercises is that you have to come up with a problem that is close to you and try to find the solution. And we guide you all the way through.
The industry increasingly demands end-to-end solutions. So there are lots of companies coming up now that specify an AI pipeline from end to end, from the data gathering to specifying the problem, deployment of the model, the ethics and governance all are in one package. And there are some really enormous startups, what's it called? I can never pronounce it. [INAUDIBLE]
It ended up, Google bought a huge stake in it. And it's now worth something like 4 and 1/2 billions. It's getting really big. And that's one of the things we cover quite a lot in the course.
We have the Fundamentals of Data Science course, for instance. We have, and Manuel has already talked about it. The last assignment is the specification of an AI deployment from beginning to end. So we cover that. And it's a very important part of AI and the industry now.
There is the increasing importance of privacy and data collection. It's now so easy to harvest data. There are algorithms out there that can build a portrait of you really, really easily. And the privacy is becoming an expensive legal issue.
It's caused some quite big startups to fail when they get it wrong. There was, what's it called? inBloom, there's a big education startup in 2018 that failed. They tried to produce a data warehouse solution harvesting data of students, a lot of whom were minors. And they got the privacy wrong. And the whole thing collapsed and cost hundred million.
Yeah, I mean, I could go on. The rise in the importance of Python as a programming language is now blooming really large on the AI landscape. Security and adversarial machine learning, so creating data, which can be used by algorithms and deliberately creating data that can cause damage and for malicious purposes.
There's the phrase in machine learning, "garbage in, garbage out." Increasingly now, people are deliberately making garbage to feed to algorithms in an adversarial way. The sophistication of video handling now, I was at a conference in Singapore a couple of years ago where people were demonstrating algorithms that can describe in a textual way, they'll write a report about what they're seeing in the security cam footage. They'll flag up when they think a crime has been committed. It's really quite scary.
So there's an increasingly dark side to AI as well. And the legal and instruments and frameworks and governance are being quite slow to catch up with that. So yeah, there's a lot going on. And a lot of it's a very-fast moving, changing landscape.
I can take about my favorite one. It's natural language generation. And this is something that has happened in the last three or four years. It happens, for example, when you write an Gmail And it suggests to you what's the next sentence. Or you write the name and suggest to you what is the title, the topic of the email that you're sending.
It also happens with coding, [INAUDIBLE] codes. And whether even making grammar, Grammarly, it makes grammar mistakes. This is a topic that we cover in some of the courses. And for me, this is one of the most amazing things that have happened in AI.
I think the one that I see, well, there are two things I've seen recently, both of which are potentially worrying. One is the rise of deepfakes. It's increasingly becoming, well, it's hitting the news now. And just the sophisticated and adversarial nature of AI, it seems to be becoming more and more sophisticated. And that and just the sophistication with which AI can now handle video data, particularly.
I already spoke about what I saw in Singapore with the description algorithms. But then I suppose the other thing is, it was on the news this morning. They were talking about sentient AI. And people are beginning to say that we're making speech-generation algorithms now which are beginning to sound sentient.
And I say "sound sentient" because they really aren't. But the problem is, it's not so much, I mean, people are talking about the ethics of sentient robots. And we're nowhere near that. But we do have ethical issues coming up which are AIs that can behave like humans and lead us to think that they are human. And that raises its own unique set of ethical issues which we are still trailing behind the technology on.