WEBVTT 00:00:00.600 --> 00:00:03.608 Welcome to lesson 1.1: What is AI? 00:00:03.958 --> 00:00:06.305 How it works, and what it can and cannot do. 00:00:06.655 --> 00:00:11.007 AI is already part of your everyday life, often without you noticing. 00:00:11.357 --> 00:00:14.642 Maybe your phone suggested the next word in a message today. 00:00:14.992 --> 00:00:21.072 In this lesson you find out what artificial intelligence really means, and how machines learn from examples. 00:00:21.422 --> 00:00:23.790 You learn what AI can and cannot do. 00:00:24.140 --> 00:00:28.791 And you learn how to write a good prompt, which simply means asking AI a good question. 00:00:30.291 --> 00:00:33.341 Here is what you will be able to do by the end of this lesson. 00:00:33.691 --> 00:00:38.214 First, explain in plain words what AI is, and where you meet it every day. 00:00:38.564 --> 00:00:43.620 Second, describe how AI learns from examples, instead of rules written by hand. 00:00:43.970 --> 00:00:47.149 Third, list what AI does well and what it cannot do. 00:00:47.499 --> 00:00:51.552 And fourth, write a clear prompt, with a task, some context and an example. 00:00:53.052 --> 00:00:55.655 Before we go further, here are some key words. 00:00:56.005 --> 00:01:03.130 Artificial intelligence, or AI, is software that learns patterns from data, so it can predict, recommend or create. 00:01:03.480 --> 00:01:08.664 A model is the trained system that has learned from examples and can now handle new inputs. 00:01:09.014 --> 00:01:13.558 Generative AI creates new content, like text, images or audio. 00:01:13.908 --> 00:01:16.084 ChatGPT is one example. 00:01:16.434 --> 00:01:19.634 A prompt is the instruction or question you give an AI. 00:01:19.984 --> 00:01:23.845 A hallucination is when AI states something false as if it were true. 00:01:24.195 --> 00:01:27.715 And training data means the examples an AI learns from. 00:01:28.065 --> 00:01:31.329 Their quality shapes how good and how fair the AI is. 00:01:32.829 --> 00:01:35.069 So, what is artificial intelligence? 00:01:35.419 --> 00:01:37.403 First of all, it is not magic. 00:01:37.753 --> 00:01:42.084 It is a computer system that uses the data it gets to produce an output. 00:01:42.434 --> 00:01:44.290 AI does three main things. 00:01:44.640 --> 00:01:48.779 It predicts, like the next word you type, or next week's demand for a product. 00:01:49.129 --> 00:01:53.331 It recommends, like films to watch, routes to take or replies to send. 00:01:53.681 --> 00:01:58.780 And it generates new content, such as text, images or summaries, from a short prompt. 00:01:59.130 --> 00:02:06.490 This plain-language description is based on the OECD definition of an AI system from 2024. 00:02:06.840 --> 00:02:09.635 Can you think of one thing AI does in your own life? 00:02:11.135 --> 00:02:14.740 In fact, you already rely on AI, often without noticing. 00:02:15.090 --> 00:02:22.855 On your phone, voice assistants, predictive text, spam filters, map routes and bank fraud alerts are all AI making fast guesses. 00:02:23.205 --> 00:02:27.472 When you shop or stream, apps match your past choices to new suggestions. 00:02:27.822 --> 00:02:33.966 And at work, AI sorts emails, translates text, writes up meeting notes and powers customer chatbots. 00:02:34.316 --> 00:02:36.961 This week, try to notice the AI around you. 00:02:37.311 --> 00:02:39.999 Most people use much more AI than they think. 00:02:41.499 --> 00:02:44.123 Now let's look at how AI actually learns. 00:02:44.473 --> 00:02:49.145 Modern AI is trained on examples until it can handle new cases on its own. 00:02:49.495 --> 00:02:54.210 First, developers collect lots of examples, such as text, images or numbers. 00:02:54.560 --> 00:03:00.747 Second, during training, the system adjusts itself again and again until it links inputs to the right outputs. 00:03:01.097 --> 00:03:04.574 Third, the result is a model, which has captured those patterns. 00:03:04.924 --> 00:03:08.337 Fourth, you give the model something new, and it makes a prediction. 00:03:08.687 --> 00:03:10.949 Think of a child learning what a dog is. 00:03:11.299 --> 00:03:13.581 They see many dogs, they do not read a rule. 00:03:13.931 --> 00:03:18.219 The key idea: more good examples usually means a more reliable AI. 00:03:19.719 --> 00:03:22.919 This brings us to the big idea: learning, not rules. 00:03:23.269 --> 00:03:26.192 Older software followed rules that a person wrote. 00:03:26.542 --> 00:03:33.283 That person had to foresee every situation, so the software broke down on messy tasks, like understanding language. 00:03:33.633 --> 00:03:38.433 Modern AI is shown thousands of examples and works out the patterns by itself. 00:03:38.783 --> 00:03:42.517 That is why AI can now write, translate and answer questions. 00:03:42.867 --> 00:03:48.520 But remember, AI learns from data made by humans, so it can copy human mistakes and bias. 00:03:48.870 --> 00:03:53.286 It is only as good as the examples it learned from, so always check its answers. 00:03:54.786 --> 00:03:57.133 There are several kinds of AI you will meet. 00:03:57.483 --> 00:04:02.816 Recommender AI suggests what you might like next, based on what you and others chose before. 00:04:03.166 --> 00:04:09.182 Predictive AI forecasts an outcome, for example whether an email is spam, or the fastest route home. 00:04:09.532 --> 00:04:13.905 Generative AI creates new text, images or audio from a prompt. 00:04:14.255 --> 00:04:16.197 ChatGPT belongs here. 00:04:16.547 --> 00:04:21.603 Assistant AI combines the others to help you draft, summarise, answer and organise. 00:04:21.953 --> 00:04:24.342 You do not need the maths behind any of this. 00:04:24.692 --> 00:04:27.849 And most tools you use actually mix several types together. 00:04:29.349 --> 00:04:34.128 Let's take a closer look at generative AI and chatbots, like ChatGPT. 00:04:34.478 --> 00:04:40.686 A chatbot writes by guessing the most likely next word, again and again, based on huge amounts of text. 00:04:41.036 --> 00:04:45.900 Because it learned from human writing, it sounds fluent and human, even when it is wrong. 00:04:46.250 --> 00:04:48.234 But it does not truly understand. 00:04:48.584 --> 00:04:52.957 It matches patterns in language, and does not know facts the way a person does. 00:04:53.307 --> 00:04:55.846 So treat chatbots as tools, not oracles. 00:04:56.196 --> 00:04:59.268 An oracle is someone whose answers are never questioned. 00:04:59.618 --> 00:05:03.159 Used well, chatbots save hours, as long as you check the result. 00:05:03.509 --> 00:05:06.091 Remember: likely is not the same as correct. 00:05:07.591 --> 00:05:10.044 So what can AI do, and what can it not do? 00:05:10.394 --> 00:05:13.957 AI spots patterns in huge amounts of data in seconds. 00:05:14.307 --> 00:05:17.464 It drafts text, translates and summarises documents. 00:05:17.814 --> 00:05:23.873 It answers common questions, automates repetitive tasks, and works around the clock without getting tired. 00:05:24.223 --> 00:05:26.121 But AI has real limits. 00:05:26.471 --> 00:05:30.205 It cannot truly understand meaning, context or your intentions. 00:05:30.555 --> 00:05:32.965 It cannot guarantee its answers are correct. 00:05:33.315 --> 00:05:36.494 It cannot replace human judgement, ethics and empathy. 00:05:36.844 --> 00:05:39.468 And it cannot know anything it was never trained on. 00:05:39.818 --> 00:05:43.210 A good rule of thumb: AI drafts, a human decides. 00:05:43.560 --> 00:05:46.184 Can you add one example of your own to each side? 00:05:47.684 --> 00:05:49.817 Here is something very important. 00:05:50.167 --> 00:05:52.194 AI can be confidently wrong. 00:05:52.544 --> 00:05:57.259 It can state false information as if it were true, and this is called a hallucination. 00:05:57.609 --> 00:06:00.787 It happens because AI predicts text that sounds likely. 00:06:01.137 --> 00:06:02.737 It does not check facts. 00:06:03.087 --> 00:06:09.125 Ask a chatbot about a small local business, and it may invent opening hours that sound real but are not. 00:06:09.475 --> 00:06:17.112 So always verify facts, dates and numbers with a trusted source, and read and check AI answers before using them in important work. 00:06:18.612 --> 00:06:21.471 Now for a practical skill: writing a good prompt. 00:06:21.821 --> 00:06:25.895 A prompt is what you ask the AI, and better prompts get better answers. 00:06:26.245 --> 00:06:29.829 First, be specific: say what you want, for whom, and how long. 00:06:30.179 --> 00:06:36.601 Write a five-line friendly reply to a customer asking for a refund works much better than just write a reply. 00:06:36.951 --> 00:06:41.644 Second, give context: your role, the audience, the tone, and any facts to include. 00:06:41.994 --> 00:06:45.066 Third, show an example of the style or format you want. 00:06:45.416 --> 00:06:50.173 Fourth, refine step by step, asking for changes like shorter or more formal. 00:06:50.523 --> 00:06:52.785 If the answer misses, do not restart. 00:06:53.135 --> 00:06:55.140 Just tell the AI what to fix. 00:06:56.640 --> 00:06:58.795 Finally, here are some smart habits. 00:06:59.145 --> 00:07:03.475 Think of AI as a helpful assistant, not an authority you never question. 00:07:03.825 --> 00:07:05.916 First, verify before you trust. 00:07:06.266 --> 00:07:09.594 Check facts, figures and names against a reliable source. 00:07:09.944 --> 00:07:12.141 Second, protect private information. 00:07:12.491 --> 00:07:17.910 Never paste passwords, customer data or confidential files into public AI tools. 00:07:18.260 --> 00:07:20.031 Third, keep your judgement. 00:07:20.381 --> 00:07:23.751 You make the final decision, and you are responsible for it. 00:07:24.101 --> 00:07:27.984 Fourth, give it feedback, because going back and forth gives a better result. 00:07:28.334 --> 00:07:35.779 Free tools to try include ChatGPT and Google Gemini, Microsoft Copilot, and the translator DeepL Translate. 00:07:36.129 --> 00:07:39.884 Always check a tool's terms before entering personal or work data. 00:07:41.384 --> 00:07:43.667 Now it's time to check your understanding. 00:07:44.017 --> 00:07:46.790 Pause the video and try to answer these questions. 00:07:47.140 --> 00:07:50.276 In one sentence, how would you explain AI to a friend? 00:07:50.626 --> 00:07:54.551 Why can an AI chatbot sound completely confident and still be wrong? 00:07:54.901 --> 00:07:59.147 Name two tasks AI does well, and one important thing it cannot do. 00:07:59.497 --> 00:08:02.995 And what habit will you use every time you rely on an AI answer? 00:08:04.495 --> 00:08:07.567 If you want to learn more, there are two useful resources. 00:08:07.917 --> 00:08:11.139 Elements of AI is a free online course for beginners. 00:08:11.489 --> 00:08:19.809 The OECD explanatory memorandum from 2024 explains the updated definition of an AI system in more detail. 00:08:20.159 --> 00:08:22.804 You will find the links to both on the lesson page. 00:08:24.304 --> 00:08:26.821 Let's sum up the key takeaways from this lesson. 00:08:27.171 --> 00:08:30.179 AI is software that learns patterns from examples. 00:08:30.529 --> 00:08:32.343 It is a tool, not magic. 00:08:32.693 --> 00:08:36.895 It predicts, recommends and generates, but it does not truly understand. 00:08:37.245 --> 00:08:42.792 AI can sound fluent and confident yet still be wrong, so always check important outputs. 00:08:43.142 --> 00:08:47.409 Clear, specific prompts with context and examples get far better results. 00:08:47.759 --> 00:08:50.276 And most importantly, you stay in charge. 00:08:50.626 --> 00:08:54.743 AI drafts and suggests, but a human decides and is responsible. 00:08:56.243 --> 00:08:57.993 Now put it into practice. 00:08:58.343 --> 00:09:04.401 On the platform, start with the activity called Build a better prompt, where you turn a vague prompt into a clear one. 00:09:04.751 --> 00:09:10.533 Next, do Exercise EX 1.1, Map AI in your world, which takes about one hour. 00:09:10.883 --> 00:09:12.888 Finally, take the lesson quiz. 00:09:13.238 --> 00:09:16.694 It has five questions, and you need 70 per cent to pass. 00:09:17.044 --> 00:09:19.625 You can retake it as many times as you like. 00:09:19.975 --> 00:09:22.087 Good luck, and see you in the next lesson.