I asked an AI Chatbot to write a podcast episode.
In this episode, I explain how I did this, the results of it, and explain what this means for the world of creative work.
[00:00:04] Hello, hello hello, and welcome to English Learning for Curious Minds, by Leonardo English.
[00:00:11] The show where you can listen to fascinating stories, and learn weird and wonderful things about the world at the same time as improving your English.
[00:00:21] I'm Alastair Budge, and today we are going to be talking about AI, Artificial Intelligence.
[00:00:28] This is partially a follow-up episode to the one last week, where we talked about Artificial Intelligence and language learning, but it is also a follow-up episode to a blog article and a video that you can find on leonardoenglish.com.
[00:00:44] In today’s episode, we are going to talk about how AI did when I gave it a very specific task, the task of writing a podcast, and we’ll ask ourselves how this technology might impact the world not just of podcasting, but the world of creation more generally.
[00:01:04] Let’s get right into it.
[00:01:07] Now, as you may remember from the last episode, a company called OpenAI released a chatbot powered by AI, where you can type almost anything into the chat, and you receive back a human-like response.
[00:01:23] You can ask it to respond to an email, you can ask it to tell you about a period of history, explain a scientific concept, give you a recipe, almost anything you could ask a human being, you can ask this chatbot.
[00:01:38] So, I thought, well, I wonder if it can write a podcast episode, I wonder how it will do if I ask it to write a script for English Learning for Curious Minds.
[00:01:51] Currently, as you may know or you can probably imagine, there are many hours of research, writing, editing and rewriting that go into writing a script for this podcast, and the process typically involves multiple talented people.
[00:02:07] It’s a lot of work, so I was very curious, somewhat scared even, if all of this could simply be replaced by an AI, replaced by a robot.
[00:02:20] If an AI could do as well, if not better than me and the Leonardo English team, well that would be quite scary.
[00:02:30] So, how did it do?
[00:02:33] I actually recorded the entire process of me doing this, so if you haven’t seen this already, you can watch it if you like, it’s over on the Leonardo English blog, and you can find a link below.
[00:02:45] It was a pretty amazing experience, and I’d certainly recommend checking it out if you haven’t done so already.
[00:02:53] Anyway, long story short, I told the AI a little bit about English Learning for Curious Minds, about who listens to the podcast, and I gave it some examples of topics.
[00:03:06] Then, I asked it to suggest some new ones.
[00:03:09] I have to say, the results, the topic suggestions, were pretty good. It suggested an article on spying, the history of cars, and it even suggested a couple that we have done already - the mysterious world of cryptids and one on the science of dreams.
[00:03:29] Anyway, one title particularly stood out to me: "The History of the Circus: From Ancient Rome to the Modern Day"
[00:03:38] Interesting, I thought.
[00:03:40] OK, let’s see if the AI can actually write a script.
[00:03:46] Now, I first asked it to create an outline for me, a plan of what it would cover in the episode.
[00:03:54] It did a really good job, following a structure that I might have chosen had I been writing the episode. I did need to give it quite specific instructions, quite detailed instructions, but the results were encouraging.
[00:04:10] But, that was the easy bit.
[00:04:13] How would it manage when it actually had to write in perfectly correct English, using a varied vocabulary, using interesting words and phrases, repeating harder words, and writing a script that would be spoken out loud, not simply a blog article or something that would sound unnatural?
[00:04:34] And how would it manage to fill the script with a mixture of interesting facts and trivia, combined with a compelling narrative?
[00:04:43] Ultimately, would it tell a good story?
[00:04:47] On this count, again it did surprisingly well.
[00:04:52] It wrote each section individually, I was able to give it feedback after each one, asking it to make some small changes, and the results were encouraging. There were some factual inconsistencies, some things that weren’t quite right or seemed a little strange, but overall it followed instructions well, and it did a better job than lots of humans would.
[00:05:19] I say this, actually, because I have seen the work of many, many human scriptwriters, researchers and writers who have applied to Leonardo English to help me write episodes, and this AI was a better writer than lots of them.
[00:05:35] We’ll move onto the consequences of this, the impact of this, in a minute.
[00:05:41] So, the finished product. This AI chatbot, this AI robot, did manage to write an entire script for an episode on the history of the circus.
[00:05:53] The script was quite good overall, I’d give it 6/10 perhaps.
[00:05:58] If you’d like to see it for yourself, you can read the finished script on the blog. I'll put a link in the notes (link here), but you can find it at leonardoenglish.com/blog.
[00:06:09] It was a pretty amazing experience to see this robot, this artificial intelligence, in action, and for me it was a kind of life-changing moment to see how human-like this AI was.
[00:06:25] Now, what does this all mean?
[00:06:27] Firstly, if you are worried that episodes of English Learning for Curious Minds are going to be written by robots, by AI in the future, fear not.
[00:06:38] That’s not going to happen.
[00:06:39] At the moment, this AI isn’t capable of original thought, it can’t really provide opinions or data like that. It can analyse lots of information and provide conclusions and summaries, it has access to the archives of history and knows fascinating stories, but it isn’t a particularly great storyteller.
[00:07:02] Yet, at least.
[00:07:04] So, don’t worry, our episodes will continue to be written by humans, nothing is changing there.
[00:07:11] But, let me use this example, the example of English Learning for Curious Minds, to ask you to take part in an interesting thought experiment.
[00:07:22] How would it change your opinion about a podcast, a journalist, a writer, if you found out that their writing wasn’t their writing, it was the writing of an AI?
[00:07:36] Let’s say you had been reading someone’s books, or listening to a podcast for years, really enjoying it, and then you found out that it had actually been written by AI, by a computer?
[00:07:50] If you didn’t know that it was until you were told, how would that change your experience, how would that change the way it made you feel?
[00:07:59] Feel free to press pause for a minute and think.
[00:08:04] It’s a really important question, because the line between what is created by humans and what is created by AI is going to be more and more blurred.
[00:08:16] Ultimately, does it matter who, or what, created something?
[00:08:22] Yes, you probably think, it does.
[00:08:25] If you listen to a heart-breaking story of young romance and you are led to believe that this is true, then you find out it was all invented by an AI, I imagine that you would feel tricked, you’d feel duped, you would feel misled.
[00:08:42] But if you never found out that it had been written by an AI, those feelings you felt were true, they were real, does it really matter if the entire story was fictional and created by a smart technology?
[00:08:57] Let me give you another example, one that might sound more basic.
[00:09:03] Most of the products we use, from the clothes we wear to the cars we drive to the food we eat is made in a great part by technology, by robots.
[00:09:14] If it was made exclusively by humans, without the intervention of robots, it would be substantially more expensive.
[00:09:23] Let’s take the example of clothes.
[00:09:25] Pure handmade clothes are expensive. Mass-produced clothes, made in large part via automated machines, can be very cheap.
[00:09:35] Most people around the world choose mass-produced clothes, because they are a utility, people don’t care about how they were made, they only care about the finished product.
[00:09:48] Sure, you can choose to shop in a more ethical way, to make sure that the people involved in the process of creating your clothes are treated well and paid fairly, but ultimately you care about the final product, not the process.
[00:10:06] In some cases, for example with cars, I imagine many people would probably prefer the idea of a robot making the car rather than it being “handmade”.
[00:10:17] So, why can’t an AI write a play, paint a picture, or dare I even say it, make a podcast?
[00:10:25] Maybe you’re thinking, well, I guess they can then, but let me give you my point of view.
[00:10:32] At the moment at least, we still care about the process of creation, and the more creative the process is, the more we care about the process.
[00:10:43] For example, if you visit a small village and go to a small workshop where a family has been making chocolate, let’s say, for 100 years, and you can see the intricate process of lovingly crafting this chocolate, you are probably more willing to buy it, and pay more for it, than you would if you went to the supermarket and bought a mass-produced chocolate bar from the shelf.
[00:11:10] It’s because you care about the process of creation, you have a connection with the human beings you’ve seen going through the process of creation, a much bigger connection than you have with a multinational company that makes hundreds of millions of chocolate bars every week.
[00:11:28] So, let’s bring this back to AI, and the title of this episode, “Can A Robot Write A Podcast?”
[00:11:36] Yes, ultimately, a robot can write a podcast episode, it can do a pretty good job, in fact.
[00:11:43] I’m sure that there will be plenty of podcasters, writers, journalists, who will use AI tools like this to write parts or the entirety of their work.
[00:11:54] In fact, we’ve seen this already. An early but less sophisticated version of this tool, this AI, was released in 2020, and there was a flood of products built that allowed entire blog articles to be written based of different prompts.
[00:12:12] You could tell the AI to “write a blog article about the best sandwich shops in London” or the “top 10 activities for young families in Paris”, and the AI would write out a basic article on the subject that sounded almost like a human.
[00:12:28] The purpose of this was purely to create content for websites, so that they would get more people to them and make more money, there was pretty much no care given to the quality of the content.
[00:12:41] And this is why you might have seen so much low-quality content if you searched Google over the past few years - all the AI was doing was rehashing, reformatting existing content, it wasn’t creating anything new.
[00:12:57] And this is one of the big criticisms, or at least limitations of the current version of AI. It's very capable of analysing existing information and providing summaries, but it isn’t yet capable of creating anything new or original.
[00:13:16] With the example of the “best sandwich shops in London” or “top 10 activities for young families in Paris”, it has no way of going to a new sandwich shop, tasting their sandwich and reviewing it, or taking a young family to an attraction in Paris and writing its own opinion about it.
[00:13:35] All it does is it takes the information that currently exists and gives it to you in a different format.
[00:13:43] So, at least right now, there are limitations to what Artificial Intelligence can do.
[00:13:49] But it is, to use the cliche, a game changer, it is going to fundamentally change society, and the world of work, for almost everyone.
[00:14:01] When robots started to replace humans in factories, people like journalists, lawyers and writers no doubt felt some sorrow about this, but felt that their jobs were perfectly safe. After all, a robot could never completely replace them, right?
[00:14:19] But it’s clear that, at least on the lower end, the more mass-produced and unoriginal end, yes they can.
[00:14:29] We’ve seen this already with generic blog content, but it has even spread to newspapers.
[00:14:36] Many news articles about things like lower league sporting fixtures or company reports are completely written by AI, because people don’t really care about who wrote it, they just want the information about what happened, they want the detail, the facts, the figures, who scored, did a company make or lose money, and so on.
[00:15:00] And an AI is perfectly capable of creating this kind of content, indeed, most people don’t know they are actually reading AI-written content.
[00:15:11] As of yet, opinions, analysis and original thought has remained the preserve of human beings, there isn’t a robot columnist or a best-selling AI non-fictional author.
[00:15:24] And, there aren’t yet, as far as I’m aware, any real podcasts written and presented exclusively by AI.
[00:15:32] And this podcast isn’t going to become the first, so please do not worry about this; this is not a coded message from me informing you that a robot has taken over control of Leonardo English, and that from now on, an AI will be writing every episode.
[00:15:51] But, an interesting thought that I want to leave you with today is, what if it had?
[00:16:00] OK then, that is it for today's episode on AI, and specifically, whether an AI can write a podcast.
[00:16:08] If you haven’t done so already, I would recommend you check out the episode before this, on AI and language learning, and you read the blog article where I wrote about my experience trying to get an AI to write an episode, and you can read the full script with what it came up with.
[00:16:25] I think it’s absolutely fascinating to think about how this technology is going to change the world, and I’d love to know what you think as well.
[00:16:33] What impact will AI have on you, and your job?
[00:16:37] Do you think a robot could ever replace you at work, and if so, how?
[00:16:43] If not, are you so sure about it?
[00:16:46] I would love to know, so let’s get this discussion started.
[00:16:49] You can head right into our community forum, which is at community.leonardoenglish.com and get chatting away to other curious minds.
[00:16:58] You've been listening to English Learning for Curious Minds, by Leonardo English.
[00:17:03] I'm Alastair Budge, I’m not a robot, and I'll catch you in the next episode.
[END OF EPISODE]
[00:00:04] Hello, hello hello, and welcome to English Learning for Curious Minds, by Leonardo English.
[00:00:11] The show where you can listen to fascinating stories, and learn weird and wonderful things about the world at the same time as improving your English.
[00:00:21] I'm Alastair Budge, and today we are going to be talking about AI, Artificial Intelligence.
[00:00:28] This is partially a follow-up episode to the one last week, where we talked about Artificial Intelligence and language learning, but it is also a follow-up episode to a blog article and a video that you can find on leonardoenglish.com.
[00:00:44] In today’s episode, we are going to talk about how AI did when I gave it a very specific task, the task of writing a podcast, and we’ll ask ourselves how this technology might impact the world not just of podcasting, but the world of creation more generally.
[00:01:04] Let’s get right into it.
[00:01:07] Now, as you may remember from the last episode, a company called OpenAI released a chatbot powered by AI, where you can type almost anything into the chat, and you receive back a human-like response.
[00:01:23] You can ask it to respond to an email, you can ask it to tell you about a period of history, explain a scientific concept, give you a recipe, almost anything you could ask a human being, you can ask this chatbot.
[00:01:38] So, I thought, well, I wonder if it can write a podcast episode, I wonder how it will do if I ask it to write a script for English Learning for Curious Minds.
[00:01:51] Currently, as you may know or you can probably imagine, there are many hours of research, writing, editing and rewriting that go into writing a script for this podcast, and the process typically involves multiple talented people.
[00:02:07] It’s a lot of work, so I was very curious, somewhat scared even, if all of this could simply be replaced by an AI, replaced by a robot.
[00:02:20] If an AI could do as well, if not better than me and the Leonardo English team, well that would be quite scary.
[00:02:30] So, how did it do?
[00:02:33] I actually recorded the entire process of me doing this, so if you haven’t seen this already, you can watch it if you like, it’s over on the Leonardo English blog, and you can find a link below.
[00:02:45] It was a pretty amazing experience, and I’d certainly recommend checking it out if you haven’t done so already.
[00:02:53] Anyway, long story short, I told the AI a little bit about English Learning for Curious Minds, about who listens to the podcast, and I gave it some examples of topics.
[00:03:06] Then, I asked it to suggest some new ones.
[00:03:09] I have to say, the results, the topic suggestions, were pretty good. It suggested an article on spying, the history of cars, and it even suggested a couple that we have done already - the mysterious world of cryptids and one on the science of dreams.
[00:03:29] Anyway, one title particularly stood out to me: "The History of the Circus: From Ancient Rome to the Modern Day"
[00:03:38] Interesting, I thought.
[00:03:40] OK, let’s see if the AI can actually write a script.
[00:03:46] Now, I first asked it to create an outline for me, a plan of what it would cover in the episode.
[00:03:54] It did a really good job, following a structure that I might have chosen had I been writing the episode. I did need to give it quite specific instructions, quite detailed instructions, but the results were encouraging.
[00:04:10] But, that was the easy bit.
[00:04:13] How would it manage when it actually had to write in perfectly correct English, using a varied vocabulary, using interesting words and phrases, repeating harder words, and writing a script that would be spoken out loud, not simply a blog article or something that would sound unnatural?
[00:04:34] And how would it manage to fill the script with a mixture of interesting facts and trivia, combined with a compelling narrative?
[00:04:43] Ultimately, would it tell a good story?
[00:04:47] On this count, again it did surprisingly well.
[00:04:52] It wrote each section individually, I was able to give it feedback after each one, asking it to make some small changes, and the results were encouraging. There were some factual inconsistencies, some things that weren’t quite right or seemed a little strange, but overall it followed instructions well, and it did a better job than lots of humans would.
[00:05:19] I say this, actually, because I have seen the work of many, many human scriptwriters, researchers and writers who have applied to Leonardo English to help me write episodes, and this AI was a better writer than lots of them.
[00:05:35] We’ll move onto the consequences of this, the impact of this, in a minute.
[00:05:41] So, the finished product. This AI chatbot, this AI robot, did manage to write an entire script for an episode on the history of the circus.
[00:05:53] The script was quite good overall, I’d give it 6/10 perhaps.
[00:05:58] If you’d like to see it for yourself, you can read the finished script on the blog. I'll put a link in the notes (link here), but you can find it at leonardoenglish.com/blog.
[00:06:09] It was a pretty amazing experience to see this robot, this artificial intelligence, in action, and for me it was a kind of life-changing moment to see how human-like this AI was.
[00:06:25] Now, what does this all mean?
[00:06:27] Firstly, if you are worried that episodes of English Learning for Curious Minds are going to be written by robots, by AI in the future, fear not.
[00:06:38] That’s not going to happen.
[00:06:39] At the moment, this AI isn’t capable of original thought, it can’t really provide opinions or data like that. It can analyse lots of information and provide conclusions and summaries, it has access to the archives of history and knows fascinating stories, but it isn’t a particularly great storyteller.
[00:07:02] Yet, at least.
[00:07:04] So, don’t worry, our episodes will continue to be written by humans, nothing is changing there.
[00:07:11] But, let me use this example, the example of English Learning for Curious Minds, to ask you to take part in an interesting thought experiment.
[00:07:22] How would it change your opinion about a podcast, a journalist, a writer, if you found out that their writing wasn’t their writing, it was the writing of an AI?
[00:07:36] Let’s say you had been reading someone’s books, or listening to a podcast for years, really enjoying it, and then you found out that it had actually been written by AI, by a computer?
[00:07:50] If you didn’t know that it was until you were told, how would that change your experience, how would that change the way it made you feel?
[00:07:59] Feel free to press pause for a minute and think.
[00:08:04] It’s a really important question, because the line between what is created by humans and what is created by AI is going to be more and more blurred.
[00:08:16] Ultimately, does it matter who, or what, created something?
[00:08:22] Yes, you probably think, it does.
[00:08:25] If you listen to a heart-breaking story of young romance and you are led to believe that this is true, then you find out it was all invented by an AI, I imagine that you would feel tricked, you’d feel duped, you would feel misled.
[00:08:42] But if you never found out that it had been written by an AI, those feelings you felt were true, they were real, does it really matter if the entire story was fictional and created by a smart technology?
[00:08:57] Let me give you another example, one that might sound more basic.
[00:09:03] Most of the products we use, from the clothes we wear to the cars we drive to the food we eat is made in a great part by technology, by robots.
[00:09:14] If it was made exclusively by humans, without the intervention of robots, it would be substantially more expensive.
[00:09:23] Let’s take the example of clothes.
[00:09:25] Pure handmade clothes are expensive. Mass-produced clothes, made in large part via automated machines, can be very cheap.
[00:09:35] Most people around the world choose mass-produced clothes, because they are a utility, people don’t care about how they were made, they only care about the finished product.
[00:09:48] Sure, you can choose to shop in a more ethical way, to make sure that the people involved in the process of creating your clothes are treated well and paid fairly, but ultimately you care about the final product, not the process.
[00:10:06] In some cases, for example with cars, I imagine many people would probably prefer the idea of a robot making the car rather than it being “handmade”.
[00:10:17] So, why can’t an AI write a play, paint a picture, or dare I even say it, make a podcast?
[00:10:25] Maybe you’re thinking, well, I guess they can then, but let me give you my point of view.
[00:10:32] At the moment at least, we still care about the process of creation, and the more creative the process is, the more we care about the process.
[00:10:43] For example, if you visit a small village and go to a small workshop where a family has been making chocolate, let’s say, for 100 years, and you can see the intricate process of lovingly crafting this chocolate, you are probably more willing to buy it, and pay more for it, than you would if you went to the supermarket and bought a mass-produced chocolate bar from the shelf.
[00:11:10] It’s because you care about the process of creation, you have a connection with the human beings you’ve seen going through the process of creation, a much bigger connection than you have with a multinational company that makes hundreds of millions of chocolate bars every week.
[00:11:28] So, let’s bring this back to AI, and the title of this episode, “Can A Robot Write A Podcast?”
[00:11:36] Yes, ultimately, a robot can write a podcast episode, it can do a pretty good job, in fact.
[00:11:43] I’m sure that there will be plenty of podcasters, writers, journalists, who will use AI tools like this to write parts or the entirety of their work.
[00:11:54] In fact, we’ve seen this already. An early but less sophisticated version of this tool, this AI, was released in 2020, and there was a flood of products built that allowed entire blog articles to be written based of different prompts.
[00:12:12] You could tell the AI to “write a blog article about the best sandwich shops in London” or the “top 10 activities for young families in Paris”, and the AI would write out a basic article on the subject that sounded almost like a human.
[00:12:28] The purpose of this was purely to create content for websites, so that they would get more people to them and make more money, there was pretty much no care given to the quality of the content.
[00:12:41] And this is why you might have seen so much low-quality content if you searched Google over the past few years - all the AI was doing was rehashing, reformatting existing content, it wasn’t creating anything new.
[00:12:57] And this is one of the big criticisms, or at least limitations of the current version of AI. It's very capable of analysing existing information and providing summaries, but it isn’t yet capable of creating anything new or original.
[00:13:16] With the example of the “best sandwich shops in London” or “top 10 activities for young families in Paris”, it has no way of going to a new sandwich shop, tasting their sandwich and reviewing it, or taking a young family to an attraction in Paris and writing its own opinion about it.
[00:13:35] All it does is it takes the information that currently exists and gives it to you in a different format.
[00:13:43] So, at least right now, there are limitations to what Artificial Intelligence can do.
[00:13:49] But it is, to use the cliche, a game changer, it is going to fundamentally change society, and the world of work, for almost everyone.
[00:14:01] When robots started to replace humans in factories, people like journalists, lawyers and writers no doubt felt some sorrow about this, but felt that their jobs were perfectly safe. After all, a robot could never completely replace them, right?
[00:14:19] But it’s clear that, at least on the lower end, the more mass-produced and unoriginal end, yes they can.
[00:14:29] We’ve seen this already with generic blog content, but it has even spread to newspapers.
[00:14:36] Many news articles about things like lower league sporting fixtures or company reports are completely written by AI, because people don’t really care about who wrote it, they just want the information about what happened, they want the detail, the facts, the figures, who scored, did a company make or lose money, and so on.
[00:15:00] And an AI is perfectly capable of creating this kind of content, indeed, most people don’t know they are actually reading AI-written content.
[00:15:11] As of yet, opinions, analysis and original thought has remained the preserve of human beings, there isn’t a robot columnist or a best-selling AI non-fictional author.
[00:15:24] And, there aren’t yet, as far as I’m aware, any real podcasts written and presented exclusively by AI.
[00:15:32] And this podcast isn’t going to become the first, so please do not worry about this; this is not a coded message from me informing you that a robot has taken over control of Leonardo English, and that from now on, an AI will be writing every episode.
[00:15:51] But, an interesting thought that I want to leave you with today is, what if it had?
[00:16:00] OK then, that is it for today's episode on AI, and specifically, whether an AI can write a podcast.
[00:16:08] If you haven’t done so already, I would recommend you check out the episode before this, on AI and language learning, and you read the blog article where I wrote about my experience trying to get an AI to write an episode, and you can read the full script with what it came up with.
[00:16:25] I think it’s absolutely fascinating to think about how this technology is going to change the world, and I’d love to know what you think as well.
[00:16:33] What impact will AI have on you, and your job?
[00:16:37] Do you think a robot could ever replace you at work, and if so, how?
[00:16:43] If not, are you so sure about it?
[00:16:46] I would love to know, so let’s get this discussion started.
[00:16:49] You can head right into our community forum, which is at community.leonardoenglish.com and get chatting away to other curious minds.
[00:16:58] You've been listening to English Learning for Curious Minds, by Leonardo English.
[00:17:03] I'm Alastair Budge, I’m not a robot, and I'll catch you in the next episode.
[END OF EPISODE]
[00:00:04] Hello, hello hello, and welcome to English Learning for Curious Minds, by Leonardo English.
[00:00:11] The show where you can listen to fascinating stories, and learn weird and wonderful things about the world at the same time as improving your English.
[00:00:21] I'm Alastair Budge, and today we are going to be talking about AI, Artificial Intelligence.
[00:00:28] This is partially a follow-up episode to the one last week, where we talked about Artificial Intelligence and language learning, but it is also a follow-up episode to a blog article and a video that you can find on leonardoenglish.com.
[00:00:44] In today’s episode, we are going to talk about how AI did when I gave it a very specific task, the task of writing a podcast, and we’ll ask ourselves how this technology might impact the world not just of podcasting, but the world of creation more generally.
[00:01:04] Let’s get right into it.
[00:01:07] Now, as you may remember from the last episode, a company called OpenAI released a chatbot powered by AI, where you can type almost anything into the chat, and you receive back a human-like response.
[00:01:23] You can ask it to respond to an email, you can ask it to tell you about a period of history, explain a scientific concept, give you a recipe, almost anything you could ask a human being, you can ask this chatbot.
[00:01:38] So, I thought, well, I wonder if it can write a podcast episode, I wonder how it will do if I ask it to write a script for English Learning for Curious Minds.
[00:01:51] Currently, as you may know or you can probably imagine, there are many hours of research, writing, editing and rewriting that go into writing a script for this podcast, and the process typically involves multiple talented people.
[00:02:07] It’s a lot of work, so I was very curious, somewhat scared even, if all of this could simply be replaced by an AI, replaced by a robot.
[00:02:20] If an AI could do as well, if not better than me and the Leonardo English team, well that would be quite scary.
[00:02:30] So, how did it do?
[00:02:33] I actually recorded the entire process of me doing this, so if you haven’t seen this already, you can watch it if you like, it’s over on the Leonardo English blog, and you can find a link below.
[00:02:45] It was a pretty amazing experience, and I’d certainly recommend checking it out if you haven’t done so already.
[00:02:53] Anyway, long story short, I told the AI a little bit about English Learning for Curious Minds, about who listens to the podcast, and I gave it some examples of topics.
[00:03:06] Then, I asked it to suggest some new ones.
[00:03:09] I have to say, the results, the topic suggestions, were pretty good. It suggested an article on spying, the history of cars, and it even suggested a couple that we have done already - the mysterious world of cryptids and one on the science of dreams.
[00:03:29] Anyway, one title particularly stood out to me: "The History of the Circus: From Ancient Rome to the Modern Day"
[00:03:38] Interesting, I thought.
[00:03:40] OK, let’s see if the AI can actually write a script.
[00:03:46] Now, I first asked it to create an outline for me, a plan of what it would cover in the episode.
[00:03:54] It did a really good job, following a structure that I might have chosen had I been writing the episode. I did need to give it quite specific instructions, quite detailed instructions, but the results were encouraging.
[00:04:10] But, that was the easy bit.
[00:04:13] How would it manage when it actually had to write in perfectly correct English, using a varied vocabulary, using interesting words and phrases, repeating harder words, and writing a script that would be spoken out loud, not simply a blog article or something that would sound unnatural?
[00:04:34] And how would it manage to fill the script with a mixture of interesting facts and trivia, combined with a compelling narrative?
[00:04:43] Ultimately, would it tell a good story?
[00:04:47] On this count, again it did surprisingly well.
[00:04:52] It wrote each section individually, I was able to give it feedback after each one, asking it to make some small changes, and the results were encouraging. There were some factual inconsistencies, some things that weren’t quite right or seemed a little strange, but overall it followed instructions well, and it did a better job than lots of humans would.
[00:05:19] I say this, actually, because I have seen the work of many, many human scriptwriters, researchers and writers who have applied to Leonardo English to help me write episodes, and this AI was a better writer than lots of them.
[00:05:35] We’ll move onto the consequences of this, the impact of this, in a minute.
[00:05:41] So, the finished product. This AI chatbot, this AI robot, did manage to write an entire script for an episode on the history of the circus.
[00:05:53] The script was quite good overall, I’d give it 6/10 perhaps.
[00:05:58] If you’d like to see it for yourself, you can read the finished script on the blog. I'll put a link in the notes (link here), but you can find it at leonardoenglish.com/blog.
[00:06:09] It was a pretty amazing experience to see this robot, this artificial intelligence, in action, and for me it was a kind of life-changing moment to see how human-like this AI was.
[00:06:25] Now, what does this all mean?
[00:06:27] Firstly, if you are worried that episodes of English Learning for Curious Minds are going to be written by robots, by AI in the future, fear not.
[00:06:38] That’s not going to happen.
[00:06:39] At the moment, this AI isn’t capable of original thought, it can’t really provide opinions or data like that. It can analyse lots of information and provide conclusions and summaries, it has access to the archives of history and knows fascinating stories, but it isn’t a particularly great storyteller.
[00:07:02] Yet, at least.
[00:07:04] So, don’t worry, our episodes will continue to be written by humans, nothing is changing there.
[00:07:11] But, let me use this example, the example of English Learning for Curious Minds, to ask you to take part in an interesting thought experiment.
[00:07:22] How would it change your opinion about a podcast, a journalist, a writer, if you found out that their writing wasn’t their writing, it was the writing of an AI?
[00:07:36] Let’s say you had been reading someone’s books, or listening to a podcast for years, really enjoying it, and then you found out that it had actually been written by AI, by a computer?
[00:07:50] If you didn’t know that it was until you were told, how would that change your experience, how would that change the way it made you feel?
[00:07:59] Feel free to press pause for a minute and think.
[00:08:04] It’s a really important question, because the line between what is created by humans and what is created by AI is going to be more and more blurred.
[00:08:16] Ultimately, does it matter who, or what, created something?
[00:08:22] Yes, you probably think, it does.
[00:08:25] If you listen to a heart-breaking story of young romance and you are led to believe that this is true, then you find out it was all invented by an AI, I imagine that you would feel tricked, you’d feel duped, you would feel misled.
[00:08:42] But if you never found out that it had been written by an AI, those feelings you felt were true, they were real, does it really matter if the entire story was fictional and created by a smart technology?
[00:08:57] Let me give you another example, one that might sound more basic.
[00:09:03] Most of the products we use, from the clothes we wear to the cars we drive to the food we eat is made in a great part by technology, by robots.
[00:09:14] If it was made exclusively by humans, without the intervention of robots, it would be substantially more expensive.
[00:09:23] Let’s take the example of clothes.
[00:09:25] Pure handmade clothes are expensive. Mass-produced clothes, made in large part via automated machines, can be very cheap.
[00:09:35] Most people around the world choose mass-produced clothes, because they are a utility, people don’t care about how they were made, they only care about the finished product.
[00:09:48] Sure, you can choose to shop in a more ethical way, to make sure that the people involved in the process of creating your clothes are treated well and paid fairly, but ultimately you care about the final product, not the process.
[00:10:06] In some cases, for example with cars, I imagine many people would probably prefer the idea of a robot making the car rather than it being “handmade”.
[00:10:17] So, why can’t an AI write a play, paint a picture, or dare I even say it, make a podcast?
[00:10:25] Maybe you’re thinking, well, I guess they can then, but let me give you my point of view.
[00:10:32] At the moment at least, we still care about the process of creation, and the more creative the process is, the more we care about the process.
[00:10:43] For example, if you visit a small village and go to a small workshop where a family has been making chocolate, let’s say, for 100 years, and you can see the intricate process of lovingly crafting this chocolate, you are probably more willing to buy it, and pay more for it, than you would if you went to the supermarket and bought a mass-produced chocolate bar from the shelf.
[00:11:10] It’s because you care about the process of creation, you have a connection with the human beings you’ve seen going through the process of creation, a much bigger connection than you have with a multinational company that makes hundreds of millions of chocolate bars every week.
[00:11:28] So, let’s bring this back to AI, and the title of this episode, “Can A Robot Write A Podcast?”
[00:11:36] Yes, ultimately, a robot can write a podcast episode, it can do a pretty good job, in fact.
[00:11:43] I’m sure that there will be plenty of podcasters, writers, journalists, who will use AI tools like this to write parts or the entirety of their work.
[00:11:54] In fact, we’ve seen this already. An early but less sophisticated version of this tool, this AI, was released in 2020, and there was a flood of products built that allowed entire blog articles to be written based of different prompts.
[00:12:12] You could tell the AI to “write a blog article about the best sandwich shops in London” or the “top 10 activities for young families in Paris”, and the AI would write out a basic article on the subject that sounded almost like a human.
[00:12:28] The purpose of this was purely to create content for websites, so that they would get more people to them and make more money, there was pretty much no care given to the quality of the content.
[00:12:41] And this is why you might have seen so much low-quality content if you searched Google over the past few years - all the AI was doing was rehashing, reformatting existing content, it wasn’t creating anything new.
[00:12:57] And this is one of the big criticisms, or at least limitations of the current version of AI. It's very capable of analysing existing information and providing summaries, but it isn’t yet capable of creating anything new or original.
[00:13:16] With the example of the “best sandwich shops in London” or “top 10 activities for young families in Paris”, it has no way of going to a new sandwich shop, tasting their sandwich and reviewing it, or taking a young family to an attraction in Paris and writing its own opinion about it.
[00:13:35] All it does is it takes the information that currently exists and gives it to you in a different format.
[00:13:43] So, at least right now, there are limitations to what Artificial Intelligence can do.
[00:13:49] But it is, to use the cliche, a game changer, it is going to fundamentally change society, and the world of work, for almost everyone.
[00:14:01] When robots started to replace humans in factories, people like journalists, lawyers and writers no doubt felt some sorrow about this, but felt that their jobs were perfectly safe. After all, a robot could never completely replace them, right?
[00:14:19] But it’s clear that, at least on the lower end, the more mass-produced and unoriginal end, yes they can.
[00:14:29] We’ve seen this already with generic blog content, but it has even spread to newspapers.
[00:14:36] Many news articles about things like lower league sporting fixtures or company reports are completely written by AI, because people don’t really care about who wrote it, they just want the information about what happened, they want the detail, the facts, the figures, who scored, did a company make or lose money, and so on.
[00:15:00] And an AI is perfectly capable of creating this kind of content, indeed, most people don’t know they are actually reading AI-written content.
[00:15:11] As of yet, opinions, analysis and original thought has remained the preserve of human beings, there isn’t a robot columnist or a best-selling AI non-fictional author.
[00:15:24] And, there aren’t yet, as far as I’m aware, any real podcasts written and presented exclusively by AI.
[00:15:32] And this podcast isn’t going to become the first, so please do not worry about this; this is not a coded message from me informing you that a robot has taken over control of Leonardo English, and that from now on, an AI will be writing every episode.
[00:15:51] But, an interesting thought that I want to leave you with today is, what if it had?
[00:16:00] OK then, that is it for today's episode on AI, and specifically, whether an AI can write a podcast.
[00:16:08] If you haven’t done so already, I would recommend you check out the episode before this, on AI and language learning, and you read the blog article where I wrote about my experience trying to get an AI to write an episode, and you can read the full script with what it came up with.
[00:16:25] I think it’s absolutely fascinating to think about how this technology is going to change the world, and I’d love to know what you think as well.
[00:16:33] What impact will AI have on you, and your job?
[00:16:37] Do you think a robot could ever replace you at work, and if so, how?
[00:16:43] If not, are you so sure about it?
[00:16:46] I would love to know, so let’s get this discussion started.
[00:16:49] You can head right into our community forum, which is at community.leonardoenglish.com and get chatting away to other curious minds.
[00:16:58] You've been listening to English Learning for Curious Minds, by Leonardo English.
[00:17:03] I'm Alastair Budge, I’m not a robot, and I'll catch you in the next episode.
[END OF EPISODE]
