Daily Dose of English 109
AI
Daily Dose of English 109
Intermediate
Hey everyone, my name is Ben and you're listening to a Daily Dose of English. This is a short, simple podcast that you can listen to every day to improve your English. You can find the transcripts for all episodes and more on benslanguagelab.com. I'm glad you could make it today. In this episode, we're going to be talking about AI, a very big buzzword nowadays. Oh, I guess that's another good word. Buzzword is something that people say a lot because it's popular and people want to hear it. So right now it's AI. A while ago it was like blockchain, cryptocurrency. At some point it was like fintech. At one point it was like internet. There's all these different buzzwords that people use, usually to make themselves sound more relevant and interesting than they actually are. And definitely that's happening with AI. And I wanna talk a little bit about my experience with AI, how I use it and how I see it being useful. And then also how it's not useful, what it isn't, I think is a really important thing to know. And so we'll see how in depth I can get in a simple 10 minute podcast for English learners, which is not supposed to be about big topics like this, but you know what, here we are. So, first AI stands for artificial intelligence and technically the definition has been around for a long time and we've had AI for quite a while. Artificial intelligence is something that is supposed to be close enough to thinking in a way or being intelligent that it can make decisions in some way without actually being intelligent. And that's the big difference, right? An AI of any kind is not actually intelligent. It just seems intelligent and can do things that we would think of or consider as intelligent. And so an example is something like a, even something as simple as a, like a simple program rather. God, I couldn't say that. A simple like program that does the, if A then do B, if C do X, right? Like there's this like logic chain that can be considered an AI. So for example, if you set up a simple process for your, I don't know, your home smart lights, if you have them, right? At this time, turn on, if it's raining, turn on, like these sorts of things, that can sort of be considered an artificial intelligence because it's essentially making decisions based on information. However, and then we get more complicated, there's plenty more that are artificial intelligence, right? Algorithms often are a form of AI, like a lot of things like that, things that make decisions, whatnot. But nowadays, it started to be being used for things that are a little bit further along than that. The things that use neural networks to make more complicated decisions with any kind of input, or any kind within a specific kind of input, rather. I'm not doing a very good job at explaining this, so I hope that you're following along. And so the biggest example of the past couple years is LLMs, which are large language models. So specifically the biggest one is ChatGPT, and now Cloud AI is also pretty good. And large language models, essentially what they are, are very complicated programs in the background that have learned how to learn the language that we write as humans. And so this very smart, brilliant researchers created this way of learning, how to learn this for machines, and then they gave it a ton of writing, an absolute ton of text. And from that text, the machines processed that language and they figured out a way to create more language out of it, if that makes sense. And that's really all that a large language model is doing. It's taking an input and it's guessing based on all of its past experience with language on what should come next. And so that's why when you write things in to an AI and you ask it a question, it'll always give you an answer because it's not actually thinking, it's not saying, okay, what is the answer to this? It's going, what words should come after this, this word, what words should now come next? And so in the case of like questions, it can give the wrong answer. but it thinks it's right, because it's not actually trying to answer your question, it's trying to just create the next words. And so that's a big thing to understand about AI, is that it's not smart, it's not actually trying to answer your questions or do things intelligently, it is doing something else, which can be a little bit odd when you think about it. So the same thing happens with images. If you look at the image that I use for Daily Dose of English, it's actually generated with an AI, with an image AI. and I asked it for something relatively specific so that I got this sort of image, but you see that there's things that don't make sense in them. Like the headphones on the table look weird, like the board with knobs and stuff, like there's things that just look strange about it. Many of the cables go to nowhere, right? It's clear that this is not a real picture and this is not something that a person made, if you look at it. Because the AI doesn't know these things, it just has seen millions and millions of pictures of microphones and headphones, and it can guess something that looks close enough. To it, it's the same, but to us, it's clearly not. Because we're intelligent, it is not. And so that's something important to know. However, they are very useful tools, or they can be. They can also be absolutely terrible, because they are just guessing what comes next. However, since they're based on what they've seen before, that can be actually really useful, especially for language learning, and that's what I wanna get into. Because there's a lot of uses that are just dumb, that are not useful. A lot of this replacing people or replacing work or image generation, it's just not that good yet. It's gonna be better in five, 10 years for sure. Even in a year, it'll be way better. But right now, it's still just, a good like first step and it can be useful. But as language learners, it could be really, really useful because a lot of the stuff that we want to know about language is already written in many places and this AI knows that. And so this is how I use AI every single day, is with my Czech learning. Specifically as a good dictionary. Because Czech dictionaries are sort of annoying. Because one, Czech words change all the time. So think about words like change, changes, changed. Those are all different words technically, but when I look up any of them, I wanna see the definition for change, right? But Check does this with every single word all the time in like hundreds of different ways. And so when I search things that are real words, they appear in books, people say them, they just don't show up in the dictionary. But I can always ask an AI and it knows what I mean because it's able to just, not know, but it's able to guess what word that is because it's seen it used as that word so many times on the internet. And so when I ask it to give me a definition for a word that a dictionary can't, it goes, okay, here you go. And it just gives me a definition. I find that so, so useful. I actually say that when you ask a dictionary what a word means, you have to ask it like a robot, right? You have to know the exact root word, you have to know exactly what you wanna search, and then you can look it up, and it's great. But if you don't know that, and you don't wanna ask it like a robot, it's not gonna help. But you can ask a thing like chat and GPT, like a person, you can say, hey, what does this word mean in English? And it's gonna help you, because it, can understand that better than a dictionary can. Again, it's not actually understanding, but it's able to look at the other things that you say and figure something out. This is especially relevant when you're looking up multiple words that are kind of one word. If you ask an AI, what's an idiom? Um, Oh, like if you look up raining cats and dogs in a dictionary, you might just get the definition for raining. You're like, okay, it's raining. There's water falling. That's not very helpful. What does raining cats and dogs mean? But if you ask an AI, it knows that that's an idiom and it's seen the definition and it can just tell you. And so I really, really like it for that. There's also some really, really good AI-generated voices now, which is also useful if you want to hear some things said. You can hear, essentially, a native speaker saying those words. They're not always correct, but like 95% of the time, it's really, really good. Like, totally go for it, it's great. And so that's how I use AI for language learning specifically, and that's how I find it most useful. I occasionally use it to ask questions or to do stuff for work, especially like technical stuff, like what's called a Regex or Google Sheet stuff. But that's a couple times a week. But I definitely use it as a dictionary and grammar helper every single day. All right, well, I know I didn't cover all of AI, but I hope that you maybe learned some new words or learned a bit of something. And if you ever have questions about English that you don't, you can't ask a native speaker, I definitely recommend trying out an AI to see what its answer is. Maybe it's wrong, but also maybe it's right. And you just saved yourself a bunch of time. So yeah. Anyways, thank you so much for listening to this episode. I know the last couple episodes have been a little bit technology-based, but hopefully you've enjoyed. I'll see you again tomorrow for another very different episode. Have a good one. Bye-bye.
For email updates, subscribe to my blog via email or RSS feed.