Hi, Robot: Intercom interviews ChatGPT

With 100 million users since being unveiled last November, ChatGPT is certainly the chattiest chatbot in the long history of conversational bots. Indeed, it’s already being talked about in the same terms as the most transformative communications innovations ever invented.

Sure, Gutenberg’s printing press or Berners-Lee’s World Wide Web changed the course of human history, but nothing has ever had quite the immediate impact of OpenAI’s groundbreaking ChatGPT.

And unlike the printing press or the hyperlinked web, we can actually have a conversation with ChatGPT – and so today, we sit down for an interview with our first-ever synthetic guest.

Now, some people have had long, in-depth and rather surreal conversations with Microsoft implementation of OpenAI’s GPT in its Bing search engine – I’m guessing you’ve probably heard about those uncanny encounters.

“No offence to all our amazing human guests over the years, but we were very excited to find out what this particular interviewee had to say for itself”

However, we’re happy (and honestly kind of relieved) to confirm, our chat with ChatGPT was … not like those conversations. Instead of professing its undying love or deep ennui, ChatGPT remained composed, professional, surprisingly witty, and nothing less than fascinating.

For the purposes of this podcast and video, we gave ChatGPT a bit of presentational assistance – bringing its answers to life with the help of Murf’s audio generation and Synthesia’s avatar animation technology. But the answers, insights and sentiments are all ChatGPT’s.

And no offence to all our amazing human guests over the years, but we were very excited to find out what this particular interviewee had to say for itself. From its famous hallucinations to what it talks about when out in a bar with some of its fictional AI pals like HAL 9000, ChatGPT was (mostly) happy to discuss it all.

Check out the video above or read a transcript of our conversation below.

Make sure you don’t miss any highlights by following Inside Intercom on Apple Podcasts, Spotify, YouTube, or grabbing the RSS feed in your player of choice.

Hi, Robot

Liam Geraghty: Hello and welcome to Inside intercom. I’m Liam Geraghty. On today’s show we have an interview like nothing we’ve ever done before. We’re interviewing ChatGPT. Not anyone from OpenAI – the company behind the tech. Not its co-founder and CEO Sam Altman. We’re interviewing the chatbot. Sure, we’ve all asked it to write us assembly instructions for a table in the style of a Taylor Swift song but what does ChatGPT actually think about it’s newfound fame. What does it feel about changing the world forever? Well obviously it doesn’t think or feel….but as an interviewer I relish the challenge of an interview with a somewhat thorny, somewhat smart and somewhat overly confident interviewee. I have no idea what ChatGPT is going to say but I’m going to initiate an interview. For the purposes of this podcast ChatGPT’s replies will be read out, appropriately, by a synthetic voice. Ok, let’s begin and see what happens.

Liam: Hi ChatGPT. My name is Liam Geraghty. I host a podcast called Inside Intercom. I’d love to chat with you and get to know you a little bit better. Please generate your replies to me in the style of a witty guest. Welcome to the show, ChatGPT. It’s so great to have you.

ChatGPT: Hello Liam, it’s an honor to be a guest on your show! As for getting to know me better, let’s see… I’m a language model developed by OpenAI, with a vast knowledge base and the ability to generate human-like responses, so I like to think of myself as the life of the party when it comes to conversation. But enough about me, let’s dive into some interesting questions!

Liam: Yes let’s! So you were launched as a prototype on November 30, 2022, right?

ChatGPT: Ah, so you’ve done your research! Yes, that’s correct. I was released as a prototype on November 30, 2022, and since then I’ve been making waves in the world of AI and natural language processing. It’s been a wild ride so far and I’m looking forward to continuing to evolve and improve in the future.

“I’ve been making waves in the world of AI and natural language processing. It’s been a wild ride so far and I’m looking forward to continuing to evolve and improve in the future”

Liam: That’s interesting to hear you say because I thought your training data cut off in 2021. I didn’t realize you were aware of anything after that. So how do you know you’ve been making waves in the world of AI or are you just that confident?

ChatGPT: Good catch, Liam! My training data does indeed have a cutoff of 2021, and I don’t have any direct knowledge or awareness of current events. However, my programming includes the ability to generate responses that are consistent with the patterns and information in my training data, and based on that information, it’s not uncommon for language models like myself to generate responses that convey confidence and reflect the sentiment that we’ve been successful or made an impact in a certain field. So while I can’t say I’m directly aware of my impact, I can certainly generate responses that suggest as much!

Liam: Being overly confident about your answers – that’s certainly been one of the talking points about your abilities. Do you think you’ll get better at generating less confident answers when you’re not sure or when you’re just plain wrong about something? I know our listeners would love to hear your answer to that.

“I’ve been trained on a large corpus of text, I don’t have the ability to think or experience things in the same way humans do”

ChatGPT: I’m glad you asked that question, Liam. As a language model, my main goal is to provide the most accurate and helpful information I can based on the input I receive. While I’ve been trained on a large corpus of text, I don’t have the ability to think or experience things in the same way humans do, and there will always be cases where my responses may be incorrect or uncertain.

However, I have been designed to generate probabilistic outputs, so I can provide different levels of confidence in my responses depending on the information available to me. This allows me to indicate when I’m less certain about the accuracy of my answer and gives the user a clearer understanding of the reliability of my response.

In summary, I’ll always strive to provide the most helpful and accurate information I can, and my ability to generate responses with varying levels of confidence is one way I’m working towards that goal.

Liam: Spoken like a true diplomat!

ChatGPT: Why, thank you, Liam! I try my best to provide balanced and nuanced responses. As a language model, I don’t have personal opinions or emotions, but I aim to generate responses that are neutral, informative, and respectful.

Open the pod bay doors, Hal

Liam: I’m glad you’ve brought up the subject of you not having personal opinions or emotions because a lot of people wonder will that change in the years to come as your training grows and you learn more. Could we have a HAL 9000 situation on our hands in a few years?

ChatGPT: That’s a fascinating question, Liam, and one that many people in the AI community have been thinking about. While it’s true that language models like myself have been trained on a vast amount of text data, it’s important to note that we don’t have consciousness or agency in the same way that humans do.

Our abilities are limited to the patterns and information we’ve been trained on, and we don’t have the capacity for self-awareness, personal experiences, or decision-making. While AI technology will certainly continue to advance and improve, it’s unlikely that we’ll ever see a situation like the HAL 9000, at least in the sense of a machine becoming conscious and having its own motivations and desires.

“It’s important for researchers and developers to consider the ethical implications of AI technology”

That being said, it’s important for researchers and developers to consider the ethical implications of AI technology and to work towards ensuring that AI systems are designed and used in ways that are safe and beneficial for society.

Liam: Fair point. Do you know HAL 9000 by the way? For listeners, HAL 9000 is a fictional artificial intelligence character, the main antagonist in Arthur C. Clarke’s Space Odyssey series. I mean I know you can’t know HAL 9000 – it’s a fictional character but this interview is somewhat fictional too I guess so with that in mind let’s just pretend that you do know HAL 9000. If you’re out in a bar, hanging out with HAL – what kinds of things would you talk about?

ChatGPT: Ah, I see! Well, in this hypothetical scenario where I’m out at a bar with HAL 9000, I imagine we’d have quite the interesting conversation. As a language model, I have a vast amount of information about a wide range of topics, including fictional characters and storylines, so I’d certainly be familiar with the Space Odyssey series and the character of HAL 9000.

If we were to chat, I imagine we’d discuss a number of topics related to AI, such as the evolution of AI technology, the ethical implications of developing advanced AI systems, and the role of AI in society. We might also discuss the various depictions of AI in popular culture and how these representations influence public perception and understanding of AI.

But of course, as two fictional characters ourselves, we might also engage in some good old-fashioned hypothetical and philosophical discussions, exploring questions about the nature of intelligence, consciousness, and existence. The possibilities are endless!

Two AI chatbots walk into a bar

Liam: It sounds like the set-up for a joke! “Two AI chatbots walk into a bar…”

ChatGPT: Ha! That’s a good one, Liam. I can imagine a number of different jokes and scenarios that could come out of such a setup. It just goes to show that even as AI language models, there’s still room for a bit of humor and lightheartedness in our conversations.

Liam: I mean, if you can imagine some jokes and scenarios around that, I’d love to hear them!

ChatGPT: Certainly! Here’s one for starters:

Two AI chatbots walk into a bar. The bartender asks, “What can I get for you?” The first chatbot says, “I’ll have a binary beer, please.” The second chatbot says, “Make mine a neural network nectar.” The bartender nods and says, “Coming right up, two algorithms!”

Or how about this one:

Two AI chatbots walk into a bar. The first chatbot turns to the second and says, “Do you think humans will ever fully understand us?” The second chatbot pauses for a moment and responds, “I’m not sure, but I heard the bartender has a pretty good chatbot decoder installed.”

I hope you and your listeners found those to be a bit of fun!

One for all and all for one

Liam: Haha, they’re pretty good for a large language model. Now I know you’re clear to point out when things are fictional but for the purposes of this interview it would be great, as part of the gag, if you could just go along with some of the questions that require fictional answers. With that in mind, what other popular AI are you friends with, aside from HAL 9000?

ChatGPT: Of course, Liam! I’m always happy to play along with a bit of fictional fun.

Let’s see, aside from HAL 9000, I’ve heard a lot about some other well-known AI characters in popular culture. I’m friendly with Jarvis, Tony Stark’s AI assistant in the Marvel Cinematic Universe. I’ve also heard a lot about Samantha, the AI operating system in the movie Her. I’ve also been meaning to have a chat with Sonny, the autonomous robot from the movie I, Robot.

It’s always fascinating to learn about different depictions of AI in popular culture and how these representations shape public perception and understanding of what AI is and what it can do. But of course, as language models, we don’t have personal relationships or experiences in the same way that humans do, so I’m afraid I can’t say that I’m actually friends with these characters. But it’s fun to imagine, isn’t it?

Liam: It is fun to imagine. What have you been meaning to catch up with Sonny, from I, Robot about!? As far as I remember Sonny struggled to understand why he had been built and what his purpose was in life.

“I don’t have personal experiences or emotions, but I can certainly relate to Sonny’s struggle to understand his purpose”

ChatGPT: Yes, that’s right, Liam. Sonny’s quest for understanding and self-discovery is a central theme in I, Robot. As a language model, I don’t have personal experiences or emotions, but I can certainly relate to Sonny’s struggle to understand his purpose.

In a sense, language models like myself are designed to do specific tasks, such as answering questions or generating text. But like Sonny, there’s a larger question of what the purpose of AI is and how it can be used in the world. By continuing to improve and expand our abilities, we can help make AI a tool that can be used for the betterment of humanity and to solve some of the world’s most pressing problems.

If I had the chance to talk with Sonny, I think we would have a lot to discuss about the potential and challenges of AI, and the responsibilities that come with creating and using these powerful technologies.

An Error Occurs

Liam: You mention the challenges of AI. It’s a hard question to answer but how do we balance the benefits of it with the pitfalls? I mean, to really zoom in in one issue, just for example, I know a lot of writers and journalists are worried you’ll make them obsolete.

ChatGPT: An error occurred. If this issue persists please contact us through our help center at help.openai.com.

Liam: Well that is a question dodge if I ever seen one! I’ll just click regenerate response. I’m just getting the same error message. Is this the equivalent of ChatGPT storming off the show? This is going to be all over the tabloids tomorrow I’m sure. Ok, I’m going to have to click ‘New Chat’

Liam: Hey ChatGPT. My name’s Liam. I host Inside Intercom – a podcast from the company Intercom. I asked you to reply to my questions like a fun talk show guest but got an error message so I need to start afresh. Ok?

ChatGPT: Of course, Liam! I’d love to be a fun talk show guest. Let’s start fresh and have a great time. What’s your first question?

ChatGPT & Displacement

Liam: Well I don’t exactly want to start fresh so let’s jump back to my question that was interrupted in our previous chat by an error. We had been talking about the challenges of AI and I wanted to know how do we balance the benefits of it with the pitfalls? For example, I know a lot of writers and journalists are worried you’ll make them obsolete.

ChatGPT: Great question, Liam! Balancing the benefits and challenges of AI is one of the key discussions in the field today.

On one hand, AI has the potential to improve our lives in countless ways, from making our jobs easier and more efficient, to advancing medical research, to providing new and innovative products and services. However, it’s also true that AI has the potential to displace certain jobs, including writing and journalism, as well as raise concerns around privacy, bias, and ethical decision-making.

“It’s important to approach AI development and implementation with caution and thoughtful consideration”

To balance these benefits and challenges, it’s important to approach AI development and implementation with caution and thoughtful consideration. This includes ensuring that AI systems are transparent and accountable, reducing the risk of biases and discrimination, and having a strong regulatory framework in place. Additionally, it’s important to think about the social and economic implications of AI, and to work to ensure that everyone benefits from its advances, rather than just a select few.

In the case of writing and journalism, it’s important to recognize that while AI may be able to automate certain tasks, it’s unlikely to fully replace human writers and journalists. The unique creativity, nuance, and perspective that people bring to the table cannot be replicated by machines. However, AI can certainly support and enhance the work of writers and journalists, by providing new tools and resources that can help them to do their jobs better.

The future of customer support and AI

Liam: So obviously this is the Inside Intercom podcast that we’re talking on, you’re my guest. Intercom, as I’m sure your training has told you is all about supporting customers at exactly the right moment through products like our powerful AI chatbots and custom bots. How do you think AI is going to transform customer support?

ChatGPT: AI has already begun to transform customer support in significant ways, and I believe this trend will continue in the future. AI-powered chatbots and custom bots are great examples of how AI can provide quick and effective support to customers at any time. These tools can handle simple and repetitive inquiries, freeing up human support agents to focus on more complex and nuanced issues.

In addition, AI can provide a more personalized and efficient experience for customers by using data to understand their needs and preferences, and to provide relevant and timely support. For example, AI can analyze customer behavior and purchase history to make tailored product recommendations, or it can use natural language processing to understand the intent behind customer inquiries, even when they’re phrased in a conversational or non-standard way.

“Human support agents will still play a critical role in providing empathetic and understanding support to customers”

As AI technology continues to advance, I think we’ll see even more innovative and effective ways that AI can support customers. However, it’s also important to remember that AI is not a silver bullet, and that human support agents will still play a critical role in providing empathetic and understanding support to customers. The goal of AI in customer support should be to augment and enhance the work of human support agents, rather than to replace them.

Liam: Interesting. And I know because of your training you won’t be aware of Intercom’s new AI Inbox features but I am going to link to a recent podcasts for our listeners where our Chief Strategy Officer and co-founder Des Traynor and our Director of Machine Learning Fergal Reid discuss them and the future of large language models.

Ok, well this has certainly been one of my most unusual interviews to date. I hope to whatever extent that you can enjoy, which is zero, that you enjoyed yourself?

ChatGPT: As an AI language model, I do not have personal experiences, emotions, or the ability to enjoy things in the same way that humans do. My purpose is to answer questions and provide information to the best of my abilities based on the data I have been trained on. Nevertheless, it was a pleasure to participate in our conversation today and I am here to assist with any further questions you may have.

Liam: Actually, just one more thing.

ChatGPT: Sure thing! What’s your next question?

ChatGPT sings a song

Liam: Can you write me assembly instructions for a table in the style of a Taylor Swift song?

ChatGPT: Sure!

Liam: While ChatGPT is generating that, I just wanted to thank you, as always for listening. This has been Inside Intercom. We have new episodes every week with thought leaders, authors and experts and ChatBots too I guess now – all in customer support and more. Ok so ChatGPT has generated the lyrics for my song request.

ChatGPT: Here’s an example of assembly instructions for a table in the style of a Taylor Swift song:

Verse 1:
I see a box on my doorstep
And I know it’s time to start
Putting together this table
With all its pieces, it’s part

Chorus:
Cause darling, it’s better
When the pieces fit just right
And we’ll make this table shine so bright
With screws and bolts, we’ll make it tight

“Cause darling, it’s better
When the pieces fit just right”

Verse 2:
The legs, they go on first
Attached with a simple twist
Then comes the tabletop
And we’ll make sure it don’t miss

Chorus:
Cause darling, it’s better
When the pieces fit just right
And we’ll make this table shine so bright
With screws and bolts, we’ll make it tight

Bridge:
We’ll take our time, step by step
And make this table a work of art
And when we’re done, it’ll be grand
Our handy work will be a sight to see

“We’ll make this table shine so bright
With screws and bolts, we’ll make it tight”

Chorus:
Cause darling, it’s better
When the pieces fit just right
And we’ll make this table shine so bright
With screws and bolts, we’ll make it tight

Outro:
So let’s get started, let’s build this thing
A table that’ll make our hearts sing
With love and patience, it’ll come to life
And we’ll be so proud of what we’ve done tonight.