8: Journalism, Mainstream Media, and the Rise of AI

Hi, I’m Jesse Hirsch and I’m here with my friend Alan Greg for the Red Tory podcast. And this is where, you know, our history as a podcast uses a bit of AI. We got Che Guevara golfing, we got an AI audience clapping along.

And today we’ve got Eric Archer, an AI expert to help us understand the bleeding edge of creativity, both where the cultural industries, how and why they’re using AI as part of their creative process, but also how creative professionals like Eric are using AI to really do things that certainly when I was a kid, I could only dream of, but I think even now as creative professionals, our pushing the boundaries of what we might be thinking of and being able to imagine. But of course, we start every episode of Red Tory by kind of checking in with the news and current affairs.

And Alan, this is where I always love to throw to you and say, what have you been watching? You know, what’s caught your attention in the last few days?

Well, today, I mean, you have to watch the Trump challenge to the Supreme Court to rescind an earlier decision by lower court that basically stopped their firing of the head of the, what was it, the special counsel office. We talk about how important the division of powers are in the United States to ensure democracy.

But what has happened since the founding fathers came up with the Constitution is you’ve got, I think some 11 to 15 now, you know, independent regulatory agencies that also formed by Congress are there to stop abuse of systems. This particular instance was to protect whistleblowers.

If the Supreme Court sides with Trump, it’s going to give him full license to go after all these other independent regulatory agencies.

And that given how compliant Congress is right now, given it’s questionable where the judiciary is right now, we’re waiting to see this could be very, very important to, you know, developments going forward.

Well, and, you know, we are fortunate enough to have a law professor, fairly esteemed legal professor, coming up as a guest in a little less than a fortnight. And I keep joking that I hope the rule of law lasts that long.

But they will be here certainly to help us weigh in on, you know, Eric, I gotta throw to you on this subject only briefly, and I’ll give you an opportunity to dodge it with a condition.

Have you been following the kind of legal challenge, the constitutional challenge that’s been arising around the executive orders coming out of the White House.

And if you want to dodge that, your dodge on that is what’s your take really on the larger political scene and how the public, how your friends and Family are kind of reacting to it or conversely tuning out.

Yeah, it’s, it’s a, it’s a tricky, difficult thing, you know, and it’s one that I feel personally pretty strong about, but try not to get too involved in publicly. But my day job is running a public television station, as you’re aware.

And you know, our funding comes from a federal mandate and it’s under the current administration. Anything funded in that way is in jeopardy.

And you know, it’s kind of, we used to, we have these 10 year contract negotiations with cable providers and it, we, we track the sort of like the way that the FCC leans depending on the administration.

You know, like I’ve been in this job for 12 years, so I’ve seen different administrations come in and it’s always like, okay, well who’s gonna, who’s the chair gonna be? You know, is it gonna be a jeet pie who comes after net neutrality or is it going to be someone who’s more pro public media?

And now it’s like, well, maybe he’s just going to get rid of the whole damn fcc. Are we even going to have an fcc?

It’s one of those independent reg agencies that this case will bear on for sure.

Yeah. So that’s a wild thing that we had never even considered being a factor that is now actually a consideration. So it’s pretty wild.

Let me throw something back to Jesse, because Jesse, you wrote on the weekend a very interesting substack piece.

I mean, you’re known for your unorthodoxy, but here you’re basically alleging that the legacy media, something that I’ve defended virtually my entire adult life, that Eric represents somehow can be, if not a threat to democracy, certainly not a strong bolsterer of the health of our democracy. Tell me more about that.

Well, on its core, I’m trying to separate journalism from the institutions that claim to represent it. And I’m doing this in an era where journalism can be practiced by anybody, that it’s more a methodology.

And I’m doing that partly because right now we tend to judge journalists based on their organizational affiliation. And I’d rather judge them on their track record. I’d rather judge them on kind of how they do their job.

But to your larger point, I am critical in anticipating where the kind of mainstream media in the United States are going to go and whether they are going to maintain the kind of critical independent voice that we expect.

And here in Canada we have a slightly different environment because we don’t have really the same market size to sustain the kind of commercial operations that exist in the U.S.

but I’m also coming at this, Alan, from a very selfish perspective, which is I’ve always been rejected by the mainstream media, public broadcasters aside.

And my larger argument here is one of diversity, that I think the reason that the maga, the Trumps, the right has been successful is because they have captured the alienation that many people feel from the mainstream media. They’ve turned it into a dangerous conspiracy where now they’re saying they’re going to go after journalists.

And that’s why I’m trying to separate journalism from the institution, so that we can protect journalists.

But I also think we need to be honest with people, that there has been a kind of policing of what’s acceptable, a policing of who gets to speak and who doesn’t speak. And for everyone who’s on social media, where everyone gets a voice and anyone can say anything, that’s a peculiar contrast.

But that also flips a lot of conventional wisdom on its head, because the conventional wisdom says that what has caused polarization, what has caused the erosion of belief in science and evidence, is the disintermediation of authority in, in media. Because every crackpot in the world can have their own podcast, witness us right here.

Whereas, you know, journalism has certain standards of professionalism, traditional journalism, they have, you know, editorial content control. And that, in fact, was the unifying fourth estate factor that made the, you know, the comforted the afflicted, and afflicted the comfortable.

So I reject that wholeheartedly for a bunch of reasons, but I’m gonna answer the metaphor because I just finished episode two of the Vietnam documentary you got me watching that’s on Apple tv, and it is amazing. And we are gonna talk about it over the next few episodes as I make my way through it.

The issue here is the war changed, the war for minds, the war for ideas, the media landscape changed, and the guerrilla warfare, right? The media, Viet Cong are the digital media.

And had the public health officials, had the institutional authorities, had the people who relied upon the traditional institutions adapted, then we could easily science could still be standing toe to toe with conspiracy, if not winning. But unfortunately, the conspiracists filled the vacuum.

They went to the battleground, they went into the digital culture when a lot of the traditional institutionalists refrained. This is where, tangentially, I’ll say, given how wingnuts like us can’t have a podcast.

I love your idea of us inviting institutional authorities as guests, as inviting people who have not waded into the digital sphere. And us giving them a way to do that and converse with that.

And this alludes also to what Eric will talk about later in terms of AI and how AI can make media more accessible and allow for organizations to advocate and communicate their missions more accessibly. But as it stands, it’s the conspiracists and the extremists who have really embraced the tools and that’s why they’re winning.

Well, you know, we should get our friend Bill Fox on because he wrote Eric this brilliant book about the influence of social media on journalism. He’s a journalist.

He headed up the Toronto Star bureau in Washington and Toronto, and as kind of testament of our times, how the media works, he wrote this book. It took him three years to write it. For his efforts, he got an invitation to speak to the Rotary Club in Bingham, Ontario.

I mean, it was just so depressing. But to your point, this guy is absolute authority on this book. He’s got a PhD and nothing but.

No, no. He is getting an invite to the Red Tory podcast.

That’s it.

Eric, I do want to give you a chance if you want to weigh in this sprawling topic because you could pick your pieces and still not cross any lines of neutrality, because I suspect you do have a lot to say about this.

Yeah, yeah. I mean, nothing currently. I’ll jump in if I see the spot.

Fair enough. Fair enough.

So one more part of this story, though, that I is very intriguing is what’s happening because it borders on both the role of the traditional legacy media and what’s going on in terms of the disruptions in Washington.

Is that ap, the Associated Press has been its access restricted to both the Oval Office and Air Force One, for at least on the surface refusing to call the Gulf of Mexico the Gulf of America.

But much more broadly, because there is an Associated Press style guide where it talks about how you’re supposed to deal with race, how you’re supposed to deal with gender, how you’re supposed to deal with a lot of the subjects that get the MAGA folks and the wokes on the other side crazy. And again, it’s one thing to attack Congress. It’s one thing to attack your enemies. It’s another to attack independent regulatory agencies.

Attacking journalists, I mean, is just a whole new realm of activity. I’ve known dozens of politicians who hated journalists, but that was always verboten. That was a no go zone.

I think, you know, we are at a point in which we’re trying to see who’s going to push back and he’s going around and pushing everyone he can.

You know, Keith Olbermann, who has a podcast, made a very strong argument, which is every other media organization should have joined AP in solidarity and said, we are not going to go into the White House press corps. We are not going to go, you know, not because we don’t want to say Gulf of America, but because you can’t do this. Right.

And that would have happened at a time there used to be that kind.

Of earlier on, part of the changing of the, the, the seats in the Oval Office press room was to make room for new media, to ensure, and this is the Trump administration, to ensure that those alternative voices, you know, are heard. And it’s not just traditional legacy media anyway. We’re, we’re not using Eric’s time to its, its max.

Well, and, and allow me then to, to pivot that.

To what extent do you see, we’ve been talking about the old world of mainstream media, of legacy media, which was a media, a world of scarcity, right, where only a few people could get access to the ability to write for a newspaper, to be part of a radio station, be part of television.

And not only has the Internet changed that, where now anyone can have a voice, but I think we’re arguably on the cusp of an entirely new level of accessibility, of potential diversity in voices because of the creative tools that AI offers. Assuming that you correct, assuming that you agree with my hypothesis, can you unpack that a bit?

Especially someone like Alan, who gets a sense of how these things could be accessible, but doesn’t really understand how it is impacting the creative process and the way in which people can make really powerful media with minimal resources.

Well, it definitely changed who could participate.

You know, you used to, in, in my short career, you needed, used to need proximity to large markets, you needed proximity to industry and people and equipment and money and distribution and so many things that would stand in your way of ever getting something made. And none of that’s really true anymore. Like, none of it. You don’t need to be near anywhere, anyone.

Look at us, we’re all in different parts of the world and work looks like that now, obviously. And you don’t need cameras to make visual content. You can just generate it. You don’t need microphones, you can just generate audio of all kinds.

And these LLMs are just incredible thought partners and strategists and can really help you fill in a lot of knowledge gaps where you might be weaker and help you have a better chance of success with any project, whether it’s from the ideation stage or the distribution stage or the production or whatever, suddenly anyone, anywhere, can, based on their ability, participate in a real way, give themselves a shot to be noticed and maybe make some money and at a minimum, scratch their creative itch in a way that they never could before.

Well, and as an example, let’s take midjourney, because you’ve previously described kind of your workflow in midjourney, and I found it both really quite fascinating, but also illustrative of how these tools work and what they’re capable of.

Yeah, it’s a pretty amazing technology where you can pull in and assume that.

This is the first time that Alan’s heard the word midjourney. Potentially.

Okay, it is. It’s an image generation tool and it gives you a lot of different ways that you can approach generating an image.

And in my experience, that’s the hardest part of making an AI video, because you’re going to try to make the images first. And getting this software to understand the visual that you would like it to output is the challenge. And you can do that in a number of ways.

You can write a text prompt and say, I would like a picture of a beautiful mountain scenery and see what it gives you back. And you can try to get good at that prompting and, you know, you might start to figure out what words will get you certain things.

And it’s got codes for different aspect ratios and you have to think about your camera angle and your color and your characters and all the things that go into this picture. But you can also ingest reference imagery and do different things with it.

So like earlier today I was playing around trying to literally blend two different pictures together. You can do that.

You can use an image as a style reference, so it will look at things like composition and color and lighting, or you can look at things for character reference.

So it will pull in the person and how the person looks, and you, you can try all these different things to generate a new image and then bring that new image out of mid journey into some of the image to video generators where you can make that image look like it’s a video. And so it’s just a really interesting process of trying to generate an image.

Now, copyright holders understandably, are hysterical about this because even though it generates images, it’s not making it because it’s got a creative mind, it’s using data mining that it’s got. Where does the rights of the copyright holder fit in in all this?

Well, so there’s how it got trained and that I can’t really speak to, but it, it may or may not have been trained on assets it did not have permission to train on. But, you know, I think if you use it in a creative way, you get to fair use really quickly. For example.

Explain that.

Sure. So, for example, I was making a piece last year where I wanted to ingest an image that I was inspired by, which was a copyrighted image.

It was Annie Leibovitz’s Vanity Fair cover of the cast of Game of Thrones.

Okay.

There’s this beautiful shot there on a mountainside, looking cool. And that was the vibe I was kind of going for.

So I screenshot that picture, that exact picture, and I literally pulled that picture in and then I did probably 50 different things to it. You know, I. I brought it in as a style reference. Right. So I’m just trying to understand the color tones, the composition, the lighting.

How would you describe these characters and then taking that image so I can say, okay, Mid Journey, describe this picture as you see it. And it will say, well, it’s this very cinematic looking cast of characters on a rocky outcropping with mountains in the background.

And it’s like, okay, well now generate that picture for me. Now, based on it, looking at this picture, it’s now going to generate a new picture based on the description of that picture.

Now I have Mid Journey’s version of this based on the text interpretation. Then it’s going to give me four versions of that to choose from. I might say, well, I like number three. Give me four versions of number three.

Okay, well within those I like number two. Give me four versions of number two and then blend it with this other picture that I like because I want to mix in a little bit of this style too.

And so by the end, if you could tell me where I originated from, I would pay you. Like, there’s no way you could tell me, oh yeah, you ripped this off from Annie Leibovitz because with the equivalent.

Me looking at a Picasso and looking at a Van Gogh and having that inspired, inspire me to do my own painting under fair, essentially. Yeah, fair use.

Yeah. It would never be considered a one to one replacement of the original.

As Jesse said. I mean, I’m not an avid user of AI Follow the news, I follow the business news, I follow the copyright news, I follow the societal news.

But Jesse took a couple of speeches, I had given a couple of articles I’d written, put it into AI Google, Google Notebook, Google Notebook, and said, do a podcast about Alan Greg. 22 minutes later, a man and Woman doing this. The first half of this was so spot on, it was scary.

The third quarter was made up based on the kind of ideas I had. They were projecting. People who have that kind of ideas have these additional ideas and therefore Alan, Greg must have them.

And the last was a whole bunch of shit that I adamantly disagreed with that they were attributing to me. And so I went from kind of holy shit, this is remarkable to what the fuck are you doing?

If you do that again, I’m going to come through here and punch you in the nose. And again, for a novice taking this in, it was just head spinning.

I just didn’t know what to make of it or how I would ever deal with something if Jesse wasn’t a friend but chose to be an enemy.

Now, Eric, as an aside, have you used Google Notebook A little bit?

Yeah, I’m familiar with it and it’s pretty, pretty remarkable what it could do. Yeah. So I can imagine.

What do you think of Alan’s emotional reaction though to the accuracy, the close to accuracy and the full hallucination that these systems are capable of?

Yeah, I mean it’s, it’s where we are.

You know, it’s, it’s so easy to get these tools to do this deep web search and pull all this stuff together and since synthesize it into someone’s voice.

And you know, recently I made a custom GPT that was a mashup of my favorite analysts so that I could take today’s news and just have these three or four people sort of work on it.

Unpack that a little, explain what you mean and how you did it.

Sure.

So custom GPTs are just wrappers, little wrappers that you can put on Chat GPT. So Chat GPT, the big LLM, lets you just say, you know what, I only want you to do this narrow little thing.

And so I’ve made a bunch of these little tools and they’re great. And so I wanted to. I just started getting the $200 a month plan to try out these, the new features there.

So I wanted to try out the new search feature. So I’m like, all right, let me come up with a nice prompt to do search on today’s trending topics around AI and video.

Something I would search in the morning anyway, like what’s going on in my industry? And now I get this great search result back from this fancy search tool.

And then they have a deep research tool which doesn’t have access to the Internet. But if I take the output from the Today’s news search.

And I put it into Deep Research and I say do a deep dive and contextualize today’s news for me with Deep Research. It’ll go away for 7, 8 minutes and come back with the most incredible report with now today’s news contextualized.

And so I have this report, but I’m not an analyst, you know, and I have favorite analysts that I read about and are very publicly well known.

And so it doesn’t take too much of a session with CHAT GPT to be like, you know what, Go research these three people and help me understand what makes them so great at what they do.

And, and you gotta name, drop one.

Of them stores like Scott Galloway.

Okay.

You know, Scott Galloway, Prof. GPT or. Yeah, I don’t want to name too many. He’s pretty.

One’s enough. We just wanted an example. So, so you ask, you ask ChatGPT then to kind of take the role of Scott Galloway and.

Well, not quite, but.

Okay, go on.

You know, help me understand how he sees things. How does he break it down? He’s got a great writing style. It’s very clear thinking through the noise. Like, help me understand.

Here’s another example of someone who I like that’s a little bit different. And here’s a two more examples. So here’s like four people first.

Before we get started, I want to understand like, what makes them uniquely good at what they do. And then maybe try. And then, okay, now we have that.

Let’s try to make a super analyst taking the best parts from each of those and crafting a unique fifth analyst. And so it gives you that.

It’s like, okay, well, this analyst would start with a great hook and then they would hit you with this, this information here. And then they’d, you know, and, and so I just made that into a custom GPT that’s sort of always ready to analyze.

So I can take today’s news, have it contextualized pretty radically, and then pop it in and have it analyzed and then have that output every day.

And to Alan’s point, do you ever find it, get it horribly wrong? Like there are ever times where you go, wow, okay, no, that’s not what I was looking for.

Or do you find because of the effort you’re putting in ahead of time to basically configure it, that you’re getting results that you’re happy with?

Well, that is kind of the beauty of custom GPTs is that once you get it to a place where it’s pretty reliably outputting things you’re happy with. Then it’s a good time for a custom GPT because it’s like, okay, I can pretty well trust this.

Now maybe I’ve given it 10 examples of successful outputs. So I want it to, you know, but I would always, always, always recommend a human in the loop strategy.

You know, I, I think I get, I don’t automate anything for that reason.

You know, I think of my approach as like a semi automatic approach where like I can take an output from one thing and walk it over to the next thing while I check it to make sure that I’m not giving it some garbage on the way to the next part of the process. You know what I mean? So always kind of checking the work.

Right off the top. I mean, you talked about, you know, how AI could, you know, generate income for those, you know, in a very kind of independent way.

They don’t have to be proximal geographically, they don’t have to. The boosters of AI all talk about this tremendous impact it’s going to have on revenue and productivity, yet we really haven’t seen that yet.

I mean, where do you stand on that whole argument?

Well, I mean, I’ve had the good fortune of making extra money with it over the past year because of the opportunity it afforded me in the sense that I could on at night and weekends, on weekends I could on my computer make content that people wanted to buy.

You know, I no longer had to go get it, get my camera equipment and a budget together and clear up a weekend and get a babysitter and like get all my camera crew together and go shoot stuff.

That, that wasn’t gonna happen, but, and, and you know, you gotta fly on location or go wherever and you just, it’s, it’s very difficult to do that kind of work. Obviously.

That’s a very interesting perspective that this generate independent income. I mean, the boosters and advocates all talk about, you know, how important it will be to boost institutional income.

You know, law firms will become more efficient because they’ll be able to, you know, scope all of the legal precedents and do it in one second rather than having, you know, legal clerks coming out of their, their yin yang. So this notion of a decentralized.

As more and more people start adopting the use of AI as individuals and that network, you could have a whole, the equivalent of the whole explosion of corner grocery stores.

Well, and I think the, the tension you’re, you’re describing, Alan, is whether organizations are capable of learning or whether individuals like Eric will Inherently outpace them. Right.

And I say this in the sense that where Eric’s background is in working in large, you know, showbiz, entertainment productions and companies now, as, you know, a one man band. Right now, as someone who is far more nimble, you know, Eric is the vanguard. Right.

There aren’t as many people, I feel, Eric, you can correct me if I’m wrong, who are engaging in his level of learning and creativity. Because again, you’re doing a lot of learning as part of your work. There’s a constant learning curve that you’re on.

I’m not sure a lot of organizations have that kind of culture. If you wouldn’t mind, Eric, to what extent do you practice your own R and D?

If productivity is a byproduct of investment in research and development, how are you as an independent creator, as an independent media professional, how much effort time are you spending on your own R and D? Because I’m willing to bet it’s a decent amount.

Yeah, I mean, it’s every waking moment, really. I think learning is the biggest job of everyone today. And that’s certainly how I treat it.

I think, I think one of the things that really helped me a couple years ago was I had started to kind of get into continuing ed. Anyway, I was like, let me just, I gotta get back in the hang of like taking some classes and learning some business stuff. So I was kind of like.

And man, the first one I took after a few years of not taking a class was I had to kick some dust off. You know, it’s like, God, I gotta get the hang of learning again. But now it’s like it, it’s just, it’s just part of every day.

It’s, it’s just more than anything, it’s learning. Every, every project I make is a proof of concept, not a finished piece. So it’s just this rolling, you know.

Thing talks about your, your, your work with artists. I mean, I asked about copyright holders, but artists are afraid of this too.

Because I can say, do Leonard Skynyrd’s Freebird using Celine Dion’s voice and all of a sudden they can do it. Now, does that undermine Celine Dion’s livelihood? I don’t know, but it’s worth a conversation, I think.

Yeah, I mean, there’s definitely a lot of concern around copyright and ownership and things.

But even to refine the question a little, is this a kind of sink or swim moment for artists and creatives where you clearly are like, I’m going to surf, I’m getting up there and Riding this wave versus to the point, if they’re not learning as zealously as you are, if they don’t have your level of curiosity, is there a danger for artists who even regardless of their position on copyright, kind of just get left behind because they’re not able to be as efficient, as productive as those who employ these tools?

Yeah, it’s one of my big concerns, to be honest. And that’s not just artists. It’s anyone who doesn’t really understand how to get their arms around this fast enough.

Because whether you’re an aspiring artist or a senior citizen or a young person or whatever, this can really help you. You know, I’ve. I’ve felt the power of, of this as you grab it and let it pull you forward.

And I worry about the people that won’t because it’s going to be a huge and increasing divide. So that’s part of the reason why I’ve tried to talk a lot more about it this year is. Is for that reason.

Well, that’s very interesting too, because it sounds to me like what, without putting words in your mouth, I mean, you’re kind of thinking we needed one at the same time, probably more literacy on this so more people can participate so that divide isn’t there, but also possibly more regulation so we ensure that people who don’t get hurt on, on the periphery. Jesse and I talk a lot about politics, especially American politics. And we saw the Stargate announcement.

Half a trillion dollars dedicated to AI data centers. Bill Gates is investing in Chernobyl to bring it back on site in order to know Three Mile Island. I’m sorry. Yeah, yeah, yeah, Three Mile Island.

If only Chernobyl.

Not even close. But because of the demand on electricity, and this is again, we need billions of dollars. We need three times the amount of electricity were here.

How much did this deep seek announcement the other week rattle the AI industry and people who are users of AI?

It seems like a lot. I mean, it seems like it upended this whole assumption that more money was the answer when Deep Sea came around and for 5, 6 million bucks.

So Chinese excellence. And that that assumption was wrong, it looks like anyway, from what I can tell, it looks like, you know, open I.

If open AI is not going to be the runaway winner here. Nvidia is not going to be the runaway winner here. If anything, maybe this AI stuff will become a public good.

Well, Jesse knows all about this. It’s open source. What does that mean?

Well, it’s. It’s open in the sense that they’re very transparent about their research methodology.

So the others like OpenAI will be able to replic their research methodology. And Eric, this may be a little too breaking news for you to comment, but Grok 3 was released today and Alan, Grok is Elon Musk.

So X what used to be Twitter, they have their own LLM, their own AI, their own large language model called Grok that ostensibly has been trained off Twitter data. So all the years and years of Twitter data have gone into training this and just today they made Grok versions three available.

So if the idea that power means something, that compute means something that the more you have, the better their model, then this new version of Grok is going to kick everyone’s ass and it’s going to be the new news in AI as of now, or if Grok is not that impressive, if it’s just, oh well, it’s not that great, then all of a sudden everyone’s coming like, hey, how do you do this for much cheaper with less resources? Eric, do you want to be as.

Good as Musk’s autonomous taxis?

Exactly. Eric, do you want to weigh in?

I saw him say that he, he, there were moments where he felt like it was scary smart, but who knows if that’s just press or whatever.

But I wouldn’t, I mean, I think both are probably true, that, that throwing money at these things is, is a winning strategy and there’s other ways to do it.

I mean, right, like, so that’s why I feel like the individuals are probably going to be the ones to benefit, at least in my hopeful scenario that there isn’t enough way to gatekeep this stuff for any one company, you know.

And I think, you know the level of curiosity, especially amongst people like yourself who are creating, but even more so amongst developers who are turning these tools. Because you know the other side to this, Alan, that I’m, I’m is a little over my head.

And Eric, I don’t know if you want to comment on it, but these models that these companies release, they can use them to train each other. So meta releases, all their llama models. So Facebook has open sourced all their models and so everyone uses them to train their models.

So it’s a little bit of like standing on the shoulders of giants in real time where every AI is a new giant, and every AI goes on the shoulders of that giant because they just want this to keep developing as fast and fast and fast. And to Eric’s point, that may make it hard for anyone to control it, right?

That anyone who has the know how will be able to install these tools and use these tools. You know, Eric, we have already talked about text in terms of using AI to generate text, and we talked about using AI to generate images.

Is there another workflow that you could offer as an example that helps round us out? And I guess an inverted way to ask that question.

Is making stuff with AI all the same process, or does it vary depending on the application you’re using and what you’re using to create?

I’m very interested in exploring some of the gray space in terms of what it can create. Can it create humor or insights? You know, like the.

The workflow I was talking about earlier using the tools of search and deep research and custom analysts. You know, can I have new insights delivered to me every morning based on, you know, these tools and combinations of these tools?

Can I, in a similar way, pull in today’s news and then apply a layer of, like, humor processing to get, like, jokes and, you know, raw material for humor and different things like that that are, like, really interesting in the sense of, like, if you can figure out how to produce new value, you could monetize that, you could, you know, create content with that. So I like to play around there. I mean, there’s. There’s very templated ways to do things. I’ve made 150 GPTs with the exact same dial of instructions.

So there’s a very repeatable part of AI, but there’s also like, this fluid space that I’m really interested in because you can prompt these tools any way you want. You can converse with them, and so it makes it more fluid.

And in that way, it’s interesting to me, like, how can I go into the places that standard prompts aren’t going, how can I? How can I die? I mean, this thing is trained on the entirety of the Internet, like, you could go anywhere.

So, you know, your first few responses are just going to be what everybody gets. That’s going to be all the everyday stuff. But how do you navigate down to the gold, you know?

Yeah, I mean, you’re like kind of Alice through the looking glass saying, how do I hack the AI to really give me the kind of creativity and wisdom you’re looking for? Alan, any last questions or final thoughts before we start to wrap up no.

Eric’s been really enlightening for me. I mean, I. As I said, I read a lot on this subject, but I’m not a user and I worry.

I did use AI to do some work on the Canadian census, because what I tried to do is I tried to do a statistical analysis of the history of immigration in Canada, and it was impossible to do physically. I mean, so I got my $10 a month chat GPD and was able to do it in two and a half hours, and it blew my mind. Unlike the. The exercise that Jesse has.

And I mean, am I right to think that if the source that AI is using to do its analysis, in whatever form that might take, is legitimate and authoritative, the quality of the work is unreal? Like, as I said earlier, legal precedence, it’s all written down.

You know, give me the four case studies of mothers who drowned their babies in the bathtub, you know, and they’ll come on. But if it isn’t, if you’re using, as you said, you’re exploring new layers of application.

Bring some humor to the news of this morning that you might get into whole other areas that if not dangerous, then certainly kind of iffy. Is that fair?

Yeah. Or just fascinating. I mean, there’s so much more knowledge to know. You know, like, we can all.

But we can all handle a lot more knowledge, and there’s a lot in there, and I just enjoy going and getting it.

To your point, though, Alan, I think where the general premise is fair, I think where the example starts to fall apart is these systems don’t understand fundamentally the content they’re dealing with. And even in legal briefs, the nuance of language is such that there are lots of words that have double meaning, even triple meaning.

And AI will not always understand that context, because as a language nerd, I encounter this all the time when I’m playing with AI. And to Eric’s point, I will double down and press the button to see what kind of response I get.

And that’s where you break the system, because it fundamentally doesn’t understand, it doesn’t think. It’s statistics. Right. It’s just a mathematical model. And language is almost living in the sense that there is multiple meaning, there’s nuance.

So I think your premise is correct, that the greater the credibility of the information you’re basing it on, the greater reliability of the outputs you will get. But to Eric’s point, fundamentally you still need a human in the loop.

You still need a human who’s doing the critical thinking, who’s doing the editing and you know, there’s a guest we’ll have on in the future, this guy Jason Willis Lee who does language work and translation work and his point is it’s just making us all editors. Right. Because you have to edit the output of what it puts out just to make sure fundamentally that it’s correct. Eric, any final comments, thoughts?

If you’d have any advice for Alan in terms of getting his feet wet other than using it for social science research and statistical research, where would you encourage him to kind of kick the tires?

I mean I can’t say enough just about regular old chatgpt and just going there whenever you’re thinking about something, you know, bringing it the, your challenges and your, your big thoughts, for lack of a better word.

Like I think that’s a great place to start because when you are bringing some interest or subject matter expertise to the table and, and meeting the, the power of the tools there, that’s where I think the magic starts to happen.

So like whatever, whatever you’re interested in, whatever the you know, real intersection of your knowledge is, go, go to ChatGPT with that sometime and just, just see where it leads you and you know, or, or challenge that you’re thinking about. You know, I think that’s just a great place to start.

And you know, not, not to promote ChatGPT in particular, but point out that this is true with Chat GPT and true with almost all AI platforms. It’s pretty much free to use all of their features you can use for free. They will charge you based on usage.

So like any good drug dealer, they want to get you hooked and then they’ll start charging you the money that you feel. As Eric’s case, you’re up to 200amonth now Eric on just Chat GPT alone. I’m sure the other platforms too are starting to rig it up, but that’s.

Interesting for me is that that’s an investment on Eric’s part. I mean that isn’t a hobby.

Yeah, you’re looking as you elevate the skills that in the tools that you’re using, your ability to generate income goes up at least proportionately if not exponentially.

So yeah, yeah, $2,400 a year for the gains over that year. It’s a no brainer to me.

And this is where I think Eric is being humble and wanting to be accessible. Eric is really high up on that learning curve.

There are very few people that I’ve encountered in the AI space who are as curious and as active as a learner as Eric so we’re grateful for you coming and spending time with us, Eric. Thank you very much.

Do you want to tell our audience where they can learn more about your work and perhaps even follow and learn more about your learnings as you learn them?

Sure. I put all my learnings on my portfolio site.

It’s called cgacreative.com and you can check out my generative AI video work there and all my custom GPT work too. And definitely appreciate another chance to come back and chat with you, Jesse. And definitely had a good time. Good to meet you both for sure.

Right on.

Well, great to have you, Eric.

Now, Alan, I got to mention before we go, that one of our producers at large, David Fingroot, just finished the Margaret Atwood Massey lectures as per your prescription and loved them, so wanted to send his gratitude because he thought it was really quite fantastic. For anyone listening, we do have a WhatsApp chat group, the Read Tori WhatsApp chat group, which Miriam, our other producer at large, is part of.

And we’ve also got on Signal, which is David Finger is at. So get in touch if you want to join any of them.

And we’ll be back back on Wednesday with Vasiliki Bednar talking about policy, accessibility and other great subjects. So thanks again, Eric. Thanks again, Alan, and we’ll see everybody soon.

Take Carol.

Hi guys. Good to see you.