top of page

Radio Report:
Jay Wonders About AI & Human-Centered Design

December 5, 2023

Chatting with ChatGPT and some experts, Jay Kemp struggles with their own selfish questions and tech quandaries at the intersection of AI and human-centered design.

Cartoon graphic of humans encircled against a light purple background, representing connectivity

Transcript

Jay (Narration): AI technology feels like it’s everywhere now. And it’s clearly here to stay. Personally, I’ve been overwhelmed at the possibilities and struggling to get to know this new set of tools. With the new text-to-speech feature that OpenAI just added to ChatGPT 4, it felt like the right time to go to the source and, as my colleague put it, have a chat with a chatbot. I decided to ask it what's been on my mind: what are all these AI systems actually going to do for humanity, for... me? 

 

Jay: Audio faders up. Microphone set. Okay. Good morning, everybody. Going to set the scene a little bit right now. I am sitting in the podcast studio on my university campus. I'm ready for an interview. Except this is a little weird because I'm the only one in here. Today I'm going to be interviewing my phone, which is currently balanced on top of a box of Clorox wipes to get it a little close to the microphone. Alright. A pulsating white circle showed up. It's bouncing and bumping to the sound of my voice, so I think it's picking that up. Let me start talking to it, see what happens. Good morning, ChatGPT. How are you doing today?

​

ChatGPT: Good morning.

​

Jay: Oh, woah.

​

ChatGPT: I'm here and ready to assist you. How can I help you today?

​

Jay: You sound freakishly human. Like it's kind of creeping me out a little bit, but my name is Jacob. I am hosting you today. ChatGPT, I'm a little confused. You say things like you and me, but you're clearly just a computer program that I'm talking to. So what's the point of it all? Why give you the human voice? Why do you converse with me like you're human when I very well know that you're not?

 

ChatGPT: That's a great question, Jacob. The human-like voice and conversational style are designed to make interactions more natural and intuitive for users. While I'm an AI, not a human, communicating in a familiar, human-like way can make it easier for people to express their questions and understand the responses. It’s all about enhancing user-experience. How does that sound to you?

 

Jay (Narration): Something about the chatbot’s response didn’t feel right to me. So I took it to some experts.

 

Dakuo: That's a very good question. Right. So human-centered AI doesn't necessarily mean that the AI needs to have a human-like face or an avatar body or a robotic… kind-of-like-a body. It’s not necessary.

​

Jay (Narration): That’s Dakuo Wang, researcher on human-computer interactions. He doesn’t think human-centered AI is so simple as a human voice either, but rather, it’s about the outcomes.

​

DakuoHuman-centered AI is really emphasizing, you know, the beginning of the whole design process needs to start with the understanding of the user, understanding of the real world, the context; what are the needs; what are the design requirements from the user, from the real world. And then we decide what kind of AI technology we want to build to support humans. The ultimate goal is also, you know, centered around humans’ needs.

​

Jay (Narration): A helpful answer, but frustrating in its broadness. Human needs. I get a headache trying to even figure out what all those could be. It feels easier to break it down a little bit, to talk about it in terms of my own life. I’ve seen some of my fellow students in my program try and fail to use ChatGPT to write their academic papers. That’s a need. I also talked to Cecelia Musselman, director of Advanced Writing at Northeastern University, who directs her students to try out ChatGPT for a couple of their essay assignments. 

​

Cecelia: So they know the basic set of ideas in their discipline. And they already- they kind of know not just the most important figures in their field, but even kind of the next layer up, right? So they know who Darwin is, but they also know all the people around Darwin, right? Like who was also working with them at the same time. So they- And ChatGPT, when it was writing these essays, is working at the Darwin level, right. It was not able to provide them any kick past the basics.

 

Jay (Narration): And having heard that from Cecelia was… discouraging. That led me to get a little selfish. I wanted to know what it could do for me, as a journalist. But ChatGPT’s answer was actually pretty applicable, and showed a little more how AI can be used in quote, unquote, a human-centered way.

 

Jay: So my question for you is, why should I, as a journalist, not be concerned about the ways that you might be taking my job, writing news articles, producing content? How can we ensure that you as a tool are used to enhance human creativity, rather than step on it?

 

ChatGPT: Your concerns are valid and shared by many. Here's how you as a journalist can view me as a tool, rather than a threat. One, complimentary, not replacement. I'm designed to assist and enhance your work, not replace it. Journalism relies on human judgment, creativity, and ethical considerations, which are areas where I can provide support but not substitute. Two, enhancing productivity. You can use me for routine tasks like data analysis, initial drafting, or background research. This frees up your time for more complex creative work that requires a human touch, like interviews, in-depth analysis, and storytelling. Think of me as a digital assistant that can help streamline your workflow, allowing you to focus on the uniquely human aspects of journalism that make your work irreplaceable. 

 

Jay (Narration): That answer intrigued me to it more not just as a tool for work or school, but other aspects of the human experience. Dr. Musselman also mentioned using ChatGPT to write thank you notes with her students, so I tried that out myself. Of course, I couldn’t resist injecting a little humor. 

 

ChatGPT: I got it. Here's a joke about octopi. Why did the octopus beat the shark in the fight? Because it was well-armed. 

​

Jay: Oh, grace. Alright. Let's try something different here. Write a thank-you note to my grandma for the pet octopus she bought me. And make sure you include that joke about being well-armed in there.

​

ChatGPT: Sure, I can help with that. Dear Grandma, thank you so much for the amazing pet octopus. It's the coolest gift ever. I guess now I'm well-armed to tackle any challenge, eight times over. I can't wait to spend time with my new eight-legged friend. Thanks again for being so thoughtful. Love…

 

Cecelia: And then we switch the audience, right? Like boyfriend, grandmother, aunt, right? And to watch ChatGPT change the register of the thank-you note… that gave them ideas about how to change register in their own writing, like which words do you change? Which- How do you manipulate the verb to make it closer, further, more formal, less formal? There's a lot of hype about like, “Oooo, ChatGPT can do this and that, dadada, just inspiring human creativity.” But the inspiration, I think in the end comes down to very small, kind of workaday things, right? (62s)

​

Jay (Narration): This has been Jacob Kemp, reporting journalist. After talking with ChatGPT, I think human-centered design can look like a lot of things. It can come down to something as simple as artificial intelligence saving us time, in whatever ways that means for you. For example, I saved time by asking ChatGPT to write to us out.

​

ChatGPT: Sure, Jacob. Dear listeners, this is ChatGPT, your friendly neighborhood AI, signing off. Remember, I'm like that helpful robot in science fiction, minus the laser beams and with a lot more dad jokes. Keep using your brilliant human brains. And if you ever need a digital sidekick for a quick fact check, a brainstorm or just a good laugh… You know where to find me.

bottom of page