Notes on the transcript:
Summary
The group is launching a new podcast called "Accessible Intelligence" that aims to provide a realistic perspective on AI from everyday users. They identified a gap in AI content, noting most existing podcasts feature either technical experts, overly cynical views, or hype-driven content. Their goal is to track recent AI developments from a young professional user perspective, making complex topics accessible without technical jargon. The podcast will focus on the timeline and evolution of AI over the past year, highlighting how it's actually being used and experienced by regular people.
The group discussed the rapid evolution of ChatGPT:
The participants shared their individual journeys with AI:
The group highlighted potential impacts of AI:
Potential Benefits:
Concerns:
Notes
Transcript
This will also be an exercise in how much storage I have. We should be good.
Welcome to Accessible Intelligence, where we bring a realistic perspective to... Oh, shit. Welcome back to Accessible Intelligence, or welcome to Accessible Intelligence, where we are giving a realistic take on... We have a pass the mic situation that will hopefully be smoothly edited.
I wanted to be, so just bear with that, but Megan, go. Hello, everyone. I'm Megan. I just finished a master's in technology policy. I run a social media account where I talk about all things tech, AI, and what's generally changing the world. And I'm doing this podcast because I'm really interested in talking about all things tech.
And what is interesting in the space right now.
Hello, my name is Austin. I graduated a year ago, got a job in corporate, and it was not what I wanted to be doing. And so I've kind of started experimenting with AI in terms of my job, it was like forced upon me. But since then I've been trying to experiment with it. You know, what I want to do is foreign policy.
Try to see how I can use AI to really get into a field that I think is one of the most gatekept or most high barrier to entry. So you're really experimenting with that in different formats. I loved being in school, loved the institutionalized opportunity for learning, and I think post-grad I was seeking that, and I really found a topic that I was really interested in with AI, in the sense that it creates an entirely new social experience.
It's an entirely new, like, Digital kind of reality. And I just love thinking about everything that represents and what it means for our society, our relationships, all of that. So really been pursuing. The key thing I've been doing is interviewing a lot of my friends and just hearing their perspectives and really platforming everyday perspectives on this technology that's supposedly transforming our lives.
And we all got the invite for the interview, so that was a good check for us. And I'm Sam. I'm a podcast producer and I create content for brands, mainly Notion, which is a productivity tool. And I'm doing this podcast because I like doing podcasts clearly, but also because I think informally we've been talking about AI stuff for a while now, and this is the next step too.
Whoever also might be interested to tune into our thoughts. So let's get into this group, accessible intelligence, and what we're trying to achieve. So something that we've been talking about is that... A lot of the AI-focused content out there these days is very much forward-looking expert opinions. They're either from...
very technical experts or they're overly cynical or they're overly hyped content and we feel like there's a gap for a podcast of this content that tracks recent as in the past year AI developments from a user perspective and creating that recent timeline a year to a few years about how AI is actually being used and experienced by
So I'll just put the first question out to the group, but what does all of that mean to you? Like, why does this podcast format and content? have the potential to work in your opinion.
So I think something that's been my main goal with my social media account has specifically been breaking down topics in AI without all that jargon and fluff. And I think when it comes to AI, a lot of the times, the conversations are not very accessible. We're talking about how people in corporate are using AI and how that's driving revenue growth.
When it comes to how AI is actually impacting people in their day-to-day, that's often a conversation that we have felt is very left out and not talked about enough. That was a major motivation for what we wanted to do with this podcast. Yeah, and I think we really looked around the ecosystem and saw that I think what the world needed was one more podcast. It was missing out on it.
So, but yeah, we were talking about it. I think we've had, I think us two at least have been on Sam's podcast before. And when we were there, one of the big things we were talking about was exactly like young professionals in the workforce and experiencing this AI disruption.
And I think the lesson we got from that is that everyone will need to know how to use AI at some level to kind of... If you have any questions, please feel free to reach out to me. And like you said, most of the content out there is from expert opinions, because really that's what you would want to listen to is an expert opinion. Who else would you want to listen to?
I think when you listen to that on, you know, this great podcast, I think they're very interesting. You get a lot of maybe theoretical knowledge and knowing how it works or what it means. But when you leave, you don't really know what to do with it, because either they're talking about something that's unreleased, or they're talking about, you know, building your own LLM model. Not LLM model. Building your own LLM.
And yeah, the vast majority of people will never do that. So I think what kind of is unique here is that we all experiment with AI in some way. And so at least to some level, we have takeaways, or we have thoughts, or we've tried before. So I think it's a much more accessible.
To make what both of you said a bit more concrete, can everyone think of an AI publication or podcast or any type of content or creator that They feel like it does a good job and talk about our podcast in reference to that. So for example, like this is pretty common, I think, in podcast strategy, where it's like if you're starting a new podcast, that was very like.
tech focus. You could be like, it's like Wired's Uncanny Valley, but with a younger person angle because it's clear to us as a four-person group that's been talking about the podcast for a while in some way or form, what it is. But I think to a listener, it's still pretty
So I'll give you guys all like a second to think about it, but at least like comparing it to for me a piece of content as a podcast that I really like admires, honestly, by Barry Weiss. And it's hosted by The Free Press, which is a company I also really admire. And I would say, like, at least how I see this podcast is like.
honestly but like tech version and like specifically young person version as well because I think the goal of honestly is to for young professionals who are coming at this from like very different perspectives, who are trying to figure out like what are you not seeing in the headlines or what are you maybe forgetting about that literally happened like six months ago because something happened three months ago more recently.
So that's how I would like position it in my content streams. It's like It's like honestly, but obviously we're not going to be interviewing experts, but we're still trying to get the same ethos of like showing what are the actual conversations happening around this for young professionals.
I feel like I'm not gonna have a good answer for this because I truthfully don't really watch podcast so this is like Based on one of the few I'll just say like the Ezra Klein joke because it is one of the few podcasts I listen to and I'll explain why I think that
I think for me, I'm a big fan of co-creation. I love the idea of creating knowledge, bringing together different perspectives. As I understand, a lot of podcasts do that, so maybe this isn't unique, but again, my point of reference. I feel like I've really enjoyed talking to all of you as a group, but also in individual contexts. I feel like we've had lots of really good conversations, bringing in aspects of our expertise, but also like...
When we're together we're sort of greater than the sum of our parts and I kind of envision this podcast or whatever being that basically where it's like just us literally be in the same space having this discourse and almost Creating new forms of knowledge, like creating new perspectives, ways of looking at things. And I say that as a client show because I feel like it does a really good job of really highlighting and articulating.
Perspectives that really bring together like different points of view. So I think that, or I hope that we can do that with this group.
Yeah, I think going back to what I said. It feels like our angle is different than the traditional podcasts I listen to because you know either we we've talked about Dwarakish Patel yesterday which This is a phenomenal podcast. He does a great job at really trying to bring these in-depth conversations by studying as much as he can. But
At the end of the day, it's like I said, where your takeaways are much more knowledge-oriented and intellectual and theoretical. There may be necessarily like knowing actually how to implement in your day to day, which is increasingly what people are being asked to do for AI or it's increasingly being forced in your day to day, whether it's at work.
or whether it's through a business, you know. So, I think because of that, where we're blending in much more of a conversational, casual, as you can see. Standing around talking, not sitting, sitting back here. Yeah, yeah, I'll stop. And I guess the one creator I would, I guess, reference for this is askcatGPT. I think the thing I admire the most about her content is that it's just, she's just talking about AI as a user herself.
And I think we all bring such unique and diverse backgrounds and experiences to the table. We're all working in different jobs right now, but recently graduated and are just trying to form an opinion around AI and try to figure out what we even think about it.
I think that's really what we're hoping to bring to this podcast in helping viewers who are our age just trying to navigate this new technology that's really impacting our lives.
Just to recap, the focus of why we're doing this podcast is to create an AI-focused piece of content that emphasizes the recent history of AI developments. We want to be different by focusing on the timeline and evolution of AI rather than the current releases and hype cycles or doom cycles from a young professional perspective and I think at the end of the day just make these topics about AI as accessible as possible to people who aren't deeply in the field but still get affected just the same.
So let's put our money where our mouth is and talk about AI. I'm going to talk about one specific timeline, so Pachachi BT, in case you live under a rock and you haven't heard about it at this point, it's pretty big in the world. It's probably, numbers wise and culturally, the most popular AI model that we use. And it is created by a company with a very weird structure called OpenAI.
And before November 2022, the team had been working on Chat GPT-1, Chat GPT-2, but it was really Chat GPT-3 in November 2022. That it blew up in one supernova, and I was reading the release of it yesterday, and this is from November 30th, 2022 and What struck me is that the messaging is so plain. Like it just starts with saying, we've trained a model called chat GVT, which interacts in a conversational way, which might be the most like low key introduction to such a game changing piece of technology.
And it goes on to say that the dialogue format makes it possible for chat to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests. And we'll get to how newer models are talked about. And I think it's obvious there's still these like release pages, but it's talked about in much more grandiose ways. But I'll just finish with the last sentence here and then ask people what their thoughts were, where they were when this...
Looking back, pretty monumental moment happened, but this release ends by saying we are excited to carry the lessons from this release into the deployment of more capable systems, just as earlier deployments informed this one. So it was never going to stop with GPT-3, and I think that really informs how we view the
That is a quick refresher, but anybody want to take a first stab at where they were when this came out? How long did it take for that to happen? If you want to understand how to use it, use it more, I'll have all of the memories. Austin, you have the mic, so go. Yeah, I can speak.
For both me and Hillary a little bit in terms of, I know neither of us really used it when it first came out. I think we both saw it as, you know, we're in school. And we both value learning a lot. And so we saw this, or at least maybe I should stop speaking for you. I saw it as something that I didn't need, that it was there, but it wasn't necessary because I had confidence in my own abilities and I hadn't really explored it and explore its use potentials. So yeah, I just thought that I didn't need it. And I was also doing, I studied international history, so it wasn't like,
I needed it for coding, or I needed it for math, or something that's more straightforward, I would say, for a model. But then with me, it really blew up in terms of usage, like I said earlier, at work. because my superior was very open to using it, especially because we had a lot of work. It was something that you could use to really
So at first, you know, like you said, I think everyone gets in a chat. It's the most, I think, consumer friendly in a way in terms of it's literally just a chat bot. You can chat with anyone can chat with. And then very quickly though, at that point, it's crazy to think about, but at that point it was only trained on data up until a certain point. Maybe it was when it was released.
But it didn't have web search abilities and have, you know, it was not nearly as good as it is now. So then we moved on to perplexity as well, which. It was this kind of revolutionary of using it, but also using it to scrape the web and to find data points and to...
To really use it to search rather than use it as like a just chat, a chat model, and I think it's just kind of blown up since then in terms of. Other models coming out in terms of capabilities, coming out in terms of new wraps, and I'm sure we'll get all this. But yeah, really...
I think it's crazy to think, going back to the original release, Chachabitty, how much it really has come since then. Yeah, I feel like, well, one thing is I wouldn't say that I had like one initial encounter with Trashy BT almost because I feel like it came up.
at different points. Just because it impacts so many different parts of your life. It came up in many different parts of my life. But going back to the first instance, I heard about it. I actually vividly remember I was like walking across the Key Bridge, which is this bridge between Georgetown and like Virginia, basically, and like it was very close to campus. I was walking around with a friend.
And my friend was like, oh, like, have you, like, used this, like, chat GPT thing? And I was like, you know, I honestly, I'm not like a big, like, tech person. So, like, I kind of viewed it as this thing where it was like, oh, it's like this. New website, you know, new websites come out all the time, like doing interesting things. I'm not necessarily the most keen on adopting stuff like that. But yeah, he was telling me he was like, oh, yeah, like.
It's kind of cool, like I was playing around with it. Again, like not really fully internalizing like what it would become over the next couple of years. But I really, I wanted to bring this up because I think My initial sort of, I don't know if interaction is the right word with it, but like, I think a big part of it was actually like play.
Because we talk a lot about how we use it for school, like we use it for work. And now it has these very practical functions, right? But I think especially in the beginning when you're encountering this like... really random website that suddenly everyone is talking about. It becomes like an entertainment source in some way or another. It's like, oh, what can this do? Can I just like...
If we were, I don't know, I think that's what characterized it for me. I mean, yeah, on that point, like, I think another way to define play is like jailbreaking. Like, you had people who got chat to be needed to, like, produce infinite A's, like, like, there was so many examples of those prompts at the very beginning.
I'm probably still not where you can, like, jailbreak it. I have no idea. Do you guys know about that? I think in the beginning.
Yeah, I think that was really interesting. I had a similar experience with my first interaction with chatGVT. I think in college, someone was like, have you tried this thing exactly? And it was very much word of mouth. And I was like, no, I have no idea what this is.
When I first tried it, I was kind of like talking to it as if it was like Siri. And I remember when I was younger, people with Siri would try to get it to like recite poems or like say something funny. And that was kind of the interaction I had with it.
It was like... I don't know, saying all these, like I could get it to like write me a song about myself or like do all these like silly things. And I was like, oh, this is so fun. I think that was prior to when I actually realized like how. Useful it would be for you know using it for some of my classwork things like oh my god in my senior year there was one part of this one course I had like 200 pages of reading to do a week and my teacher was actually just like
If you can't get to all of these, just throw them into chat 2T and it will summarize it for you. I was just like, wow, this is incredible. And then I started using it to help me fix bugs in my code for my senior thesis. Yeah, I think it was very much like a gradual progression from like hearing it word of mouth and then discovering new utilities for it. And yeah, I thought that was really interesting. And I think, yeah, like.
And I would say that definitely reflects my experience. I don't, maybe it resonates with all of you guys as well, where this idea of going from experimentation and play to something a little more practical. Also I think something that's like relevant to emphasize here is like we were all like college kids at this point right like we were all in school and in that way I also wonder like.
Was that a unique experience in some ways? The fact that we encountered this very transformative technology first through planned experimentation. It's also interesting, too, because now when we think about the benefits of AI, especially at a societal scale, it's like, oh, it's going to make businesses so much more efficient.
It has all these impacts on the global economy. These are things that are very far from people just playing around with things. It's interesting to think about this is almost a generational experience in some ways. And to understand like, like really the, maybe they're actually very different ways that people encounter the technology. And like, to what extent does that shape their current relationships with it?
Let's move forward a couple of months now. So we're going from November 2022 to March 2023. It only takes a few months for OpenEdge to release GPT-4. And just looking at the same format, press release, it's way more engaging. It's way more interactive. It's way more business focused. Whereas when they were introducing chat GPT-3, the team was just saying like, here it is. But they say GPT-4.
It is more creative and collaborative than ever before. It can generate, edit, and iterate with users on creative and technical writing tasks, such as composing songs, writing screenplays, or learning a user's writing style. So it's a lot more... focus on a user's creativity in the different ways, like you mentioned. It also, I think for the first time, includes stats on how GBT is performing in a bar exam or biology Olympiad, which is increasingly becoming relevant or calling into question the relevancy of those competitions for humans themselves.
So, Fast forward a few months, maybe nothing has changed for you guys, but curious to know if with the new GPT-4 in March 2023, if that I will quickly just say that I feel like during that semester of spring 2023, some of the peers of me and Hilary, because we had the same major, I was hearing more about how they were using it.
And especially because we had a few pretty intense courses, like I think that's when I started using it more. So probably right after GPT-4 came out, and that was just to like help me brainstorm, help assemble lists. Me and Hilary did a project for the ICP Rail project in California, which one day, God willing, will be finished. And we used ChachiBT to come up with the name of our hypothetical entity. So things like that where it felt more real and just more part of the work, but not like...
It was doing the work for us.
Yeah, I think it's a...
I really like how you talked about in terms of stumbling upon a website in terms of with Chachapiti 3 where like it reminds you of so many other websites that just does something you're like oh I'm gonna go on this you know play around with it you'll be in like Lao which was the library.
Like just hanging out with friends. You just play around with it, see what it would do. Very creative. And then there does feel to be that that jump. Chachapiti 4, like you said, in terms of the narrative, in terms of how they release it. And I think it feels like it's a lot to do with realizing the market and realizing really the consumer potential.
and stumbling onto what is like the next iPhone, you know, the next big, the next big market. And so you see Chach B.T. It had the first mover's advantage. It was the first to release a chatbot. But I think around this point, you were also getting Anthropics Cloud coming out. You know, because also when we're moving from this talking about, oh, this is an interesting concept to AGI.
It was even a hint of deep-seek of many Chinese capabilities. No one was thinking about semiconductors going abroad. It was purely, how fast is this going to evolve? And I think you can see that in the first or in the second release of chat 24, where it is much more aiming to capture that market share.
It's much more trying to no longer just a technology, it's now a product. They're really starting to emphasize it as being a product, and I think this is maybe where you see, I guess, like the first starts, or the first. It is a snapshot of the ongoing war for market share and for the diversifying brand. Yeah, I think something I find really interesting about just the origin story of OpenAI is how it started as a research lab.
further as the bounds of science or math. And I think like witnessing, that's exactly what we've been talking about, of how we witness that in real time, that realization kind of come about. I think that's actually a really interesting way to put it and like to sort of go into the theme like the idea of like history because I think so what I was gonna say is
Obviously, we'd all heard about AI before chat. We knew it was like this big like tech thing. We knew it was to transform things, right? And I would say like then it was talked about in pretty like a commercial sense, right? It was like, oh, like what sort of practical benefit will have like
For XYZ, like for the economy, for schools, for businesses, whatever, right? And in that way, that almost makes it seem like like our experience with it initially was almost like a fluke in some ways. Like that was like an exception and like a type of disruption that.
It went from being business focused, or I would use it as a proxy, but especially practically focused, to then this weird moment where people were experimenting with it, became very consumer focused. And now we're almost seeing... a shift back to the practical aspects. It was interesting to see that timeline charted in that way. That was just like a quick digression off of your point. But one thing I was just going to say is like.
I don't know about you guys, but I did not know about the versions of ChatGPT. First of all, I did not know that we started off with GPT-3. First, you said that 1 and 2 were in development. I guess 3 was the first one they made public. I literally don't know. It was the first one they put out publicly. Okay, got it. Didn't know about GPT-4, really. I probably came across it at some point, but would not be able to tell you the timeline for it.
So like, and just like a quick poll I guess like did you guys know that like, like when I noticed myself starting to use it more like definitely like second semester like, which is what around which came out but like, Did you guys know it was a new model? I remember the hype around it. I think that's what I was referencing when I feel like our peers are talking about it more because I feel like when GPT-3 came out, I had similar reactions to others where it was like,
Okay, let's just kind of see what this is and if it's not gonna like immediately impact my chances of success as a student. Like there's so many other things going on in college where I just didn't really think about it. But yeah, I remember like GPT-4 being hyped in a way that GPT-5 got hyped, and I'm sure GPT-6 and 7 and however many more. And of course, there's always going to be a gap between like the expectation and reality, but I do remember like...
Pretty aware of it. You know, to be honest, I did not. Like, I knew it had released in terms of Chacharity 4, but like I said, did not care, was not really part of my life. I think the most I knew or heard about AI was actually Claude, in terms of one of my roommates was in CS.