Big Data and AI | SDL4 A3.1 |
Artificial Intelligence, or AI, is intelligence that is demonstrated by machines, rather than people or animals. In these videos, you'll learn more about what AI is, the history of AI, and some use cases.
After watching the videos, you're going to complete a series of tasks on Artificial Intelligence and machine learning.
When you're ready, click 'next step' to start the first task.
So we'll be talking about the history of AI because it in many ways is the history of computer science also on the history of programming, right. So the 1940's and 50's was the first generation of AI and here really this was just the use of logic in programming, so we consider this logic plus programming. So it was the development of algorithms that has conditions in them, tests, and those tests could be sensitive to information, external to the machine and the machine could make decisions based on that information and that is indeed something that human beings do with intelligence perhaps, but it is nevertheless a little too dumb for us today. So for example, like I say, you know, if the age of the user is more than or equal to eighteen we could allow them into the building, say, otherwise we could deny them. Now this is very intelligent system compared to human-, compared to the history of human inventions, you know.
If we go back into the Victorian era there is no system we can build to adapt to a user's information in such a way. So this is a real innovation. Right, okay. So it was-, you know, that was attempted but it didn't quite work out and so we have what was called an AI winter. We say winter, doing blue perhaps, and in-, during the winter, you know, the research projects dried up. So what came next? Well the 1980s was when we, sort of-, another little renaissance and here there are-, there are multiple approaches being tried.
One of the key ones, expert rules, expert rules and the idea here or at least the thought here was that well, you know, maybe what we-, what was going wrong with the earlier approach of just logic and programming was that the tests being used didn't have the right structure. Maybe if we asked experts about things, you know, ask the expert bike rider, 'What rules do you follow?' and if we take those perhaps we can build a machine which imitates the experts, you know. Behaves as if human, with human performance, right, or better than human performance. So when this approach seemed to be somewhat limited we move on to a new one, which is machine learning.
So again we have this winter, this AI winter and then we have in the late 2000s really those, you know-, really mostly we have machine learning. Now it was invented much earlier, though to be honest many ways it was invented in, in the Victorian era or even earlier than that, since the techniques are mostly just those of statistics.
So it's mostly just the use of statistics here somehow, to specialise the algorithms the machine will follow and, and therefore we're using statistics as that process of considering data and finding out the key features of it and then machine learning becomes this, this, you know, this activity which needs large amounts of data to make these high quality rules. Let's now then have a look at some of the ethical concerns around the use of AI.