.png)
Pressing Matters
Pressing Matters
4th Anniversary Episode with Jon Swartz & Emilia David
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Welcome to our special fourth anniversary edition of Pressing Matters!
If you do the math, you'll quickly realize that I interviewed CNBC's John Fortt for our debut episode two months before ChatGPT changed the universe. So, yes, our anniversary podcast will be about AI, again. AI – AI in the world and AI in journalism for the third straight year. And I was feeling a little guilty about that, but really... it's not like it's even close to the same conversation we had last year, or for that matter, the years before. AI changes everyday and maybe every hour.
So to guide us through the past year of change, we asked past guests Jon Swartz of Techstrong and Emilia David of the ISMG Group to come back and talk about AI in China, Hugging Face (who goes IPO next year), and what they predict for a year from now, when we'll be marking the fifth anniversary of not only this podcast, but a little technology called ChatGPT2, for this special anniversary episode of Pressing Matters from Big Valley Marketing, the podcast that brings you conversations with the top media and influencers in B2B tech.
I'm Dave Reddy, head of Big Valley Marketing's Media and Influencers Practice, and I'm your host. Through research and good old-fashioned relationship building, we've identified B2B tech's top 200 media and influencers, including Emelia and Jon. Here's our chat with them. Enjoy.
Big Valley is hosting two events during SF Tech Week in October. We would love to see you there.
Marketers' Hot Take Happy Hour: An Evening of Agree or Disagree | October 5
Join fellow B2B marketing and communications pros at a rooftop bar for a happy hour built around the topics we love to debate — what’s working, what’s overrated, what needs to disappear, and where marketing and comms are headed next. Grab a drink and your AGREE and DISAGREE stickers, then make your way around the room weighing in on some of marketing’s most debatable hot takes that we’ll have on posterboards around the room. Register here: https://partiful.com/e/RxkakL6ey0GFQdeeTCGF
Make Me Care: The 60-Second Story Slam | October 9
Step up to the mic and tell your company or product story in just 60 seconds to a panel of people who hear pitches for a living — tech reporters (Don Clark, enterprise tech reporter from The New York Times and Chris Velazco, personal tech reporter), an industry analyst, and a venture capitalist. If you're a founder, start-up, builder or established company, this event is for you. Register here: https://partiful.com/e/6mK2jtOD2ggQrsLgv7YI
Hey everybody, this is Dave Reddy, head of Big Valley Marketing's Media and Influencers Practice and host of Pressing Matters. Before we get started with our special anniversary episode, fourth anniversary, by the way, of Pressing Matters, a big heads up. San Francisco Tech Week is coming up and Big Valley is showing up with two events. On Monday, October 5th, we're going to have Marketers Hot Take Happy Hour, an evening of agree or disagree. And then on Friday, October 9th at lunchtime, we're going to have Make Me Care, the 60 second story slam. If you want more information and please do look for more information, and please come to both events, check out the Big Valley Marketing LinkedIn page. Thanks, and now on to Pressing Matters. Welcome everyone to our special fourth anniversary edition of Pressing Matters. Now, if you do the math, you'll quickly realize that I interviewed CNBC's John Fort for our debut episode two months before ChatGPT 2 changed the universe. So yes, our anniversary podcast will be about AI again. AI, AI in the world, and AI and journalism for the third straight year. And I was feeling a little guilty about that, but really, it's not like it's even close to the same conversation we had last year, or for that matter, the years before. AI changes every day and maybe every hour. So to guide us through the past year of change, we asked past guests John Swartz of TechStrong and Amelia David of the ISMG group to come back and talk about AI in China, hugging face, who goes IPO next year, who doesn't, and what they predict for a year from now when we will be marking the fifth anniversary of not only this podcast, but a little technology called ChatGPT2 for this special anniversary episode of Pressing Matters from Big Valley Marketing, the podcast that brings you conversations with the top media and influencers in B2B Tech. I'm Dave Reddy, head of Big Valley Marketing's Media and Influencers Practice, and I'm your host. Through research and good old-fashioned relationship building, we've identified B2B Tech's top 200 media and influencers, including Amelia and John. Here's our chat with Amelia and John. Enjoy. John, Amelia, thanks so much for joining us for our fourth uh anniversary edition of Pressing Matters. Really happy to have you on. Thank you so much. You know, it's three years running now, but I guess, you know, if I if I founding a podcast in in uh October of 2022, I this is what I get. We are going to talk about AI because as has been the case for the most of the four years we've been doing this, AI is changing everything. And you two cover AI inside and out. And I'm curious what y'all think about it. So Amelia, we'll start with you. And it's gotten to the point, and maybe it's been like this for more than a year, but it's gotten to the point that there seem to be seismic shifts weekly, if not more often, in the AI landscape. Somebody's saying, doing, or saying and doing something that just blows everything out of the water. Aaron Ross Powell When you look back at the past 12 months, were there two, three times, two, three moments, or even one moment, the most impactful things that happened in the past year in AI?
SPEAKER_00Well, the most kind of important thing that's happened is that OpenAI agents reached hugging face. And that's I've been talking about agentix security and agent management for a very long time. But the fact that well, two things. The fact that like OpenAI's agents really took reward hacking to its like glorious end and actually escaped their sandbox is I'm not entirely sure if it just shows how powerful the agent is or just how completely how much we as an industry underestimated the kind of the the testing systems that we've put upon these models. And I only say open AI agents because the the thing that also is interesting about that is it's also created like a different kind of competition standpoint for AI companies, which is we can hack better than you. So that's why Anthropic and Meta suddenly were saying that, oh, our agents can also do it. But the reason they were able to do that was because they just had, they didn't double check their own testing environment. So I don't necessarily consider that as something that is indicative of how powerful a model is. But I uh the reason why it's very important is because it's really placed into very stark view that yeah, these agents can do something, whether it's by intention because it wants to fulfill a task or because humans did not double-check their own testing environment, it can happen and it's not theoretical anymore. And I guess the second thing that happened is now we have models like Mythos 5, GPT 5.6 SOL, that claim to be very, very good. And that's always been the case, sure. But this is the first time where because they've involved the government in saying that you have to, like, this is so dangerous. Like that marketing has really pushed the US government, especially this administration, in a to a place where I didn't think they would go, which was actually considering regulation.
Dave ReddyJohn, what's your take? I mean, you were at AI for we're taping this in August. You were at AI-4 just a couple of weeks ago in Vegas, which is your home away from home, uh, given all the conferences there.
SPEAKER_03Yeah, it reached it. It reached a point where he Yeah, it actually reached a point where they talked about eliminating all regulation, including statewide regulation, to the fact where they actually want to be quasi-responsible, although I take everything they say with the grain of salt because I always think there's an ulterior motive to everything they do. So what's interesting, and I'm glad, Emilia, you mentioned security because doesn't isn't this history repeating itself? Almost every technology, the last consideration, the last thing we think about are the consequences by a security point of view. So now we're trying to play catch-up or we're trying to be more responsible. And that's something I think that permeates and seeps through all the stories of the narratives of these companies. Before you rarely heard any of them mention security. What they would talk about was, oh, Jensen Wang's here on stage. Let's bring him on stage. Let's talk about our partnership. Let's bring out a company A. Company A is a customer that we're using who is going to use AI and change everything. It's going to increase productivity 30%. So that was the story we used to get. We used to used to talk about, well, we heard a lot about ROI about a year ago or even six, even up to six or nine months ago. Now the thing that we're starting to tear about whether these companies want to talk about it or not is CapEx. So these these kind of runaway costs, which is gonna manifest itself in things like, I don't know, the situation at Oracle, OpenAI. I mean, we had an exodus of open AI executives, including the chief revenue officer who left after eight months right before they go public. That's that's a good sign, isn't it? I I I mean when I think about AI agents, and I'll be quick about this, it's this was I guess it still is. I don't know, Amelia, if you agree with this or not, but this was the year of the AI agents, and in a sense it kind of moved fast, but it didn't. Like these these technologies are leaffrogging one another, but the adoption rate uh among companies or enterprises is not as fast for various reasons, maybe because of security or because they are too powerful. Oh, go ahead.
SPEAKER_00It's been the year of agents since 2024. It's yeah. It's not I like I don't honestly think that there has to be one year of doing this. I think it's just more because we keep thinking of agents as something larger than itself. It's a software. It is, yes, it can it can open up a lot of different paths and it can open up a lot of a lot of different like terrible doors, but it is still a software and like in the the cost calculation on that is something that I have always believed will not take one year or two years. Like it's it's a multi-year. It's something it's a multi-year one, yeah. But like you're right.
SPEAKER_03And it's it's but you know, going back to Silicon Valley though, they love to kind of market or gift wrapped presentations, whether it's at one of their own conferences. But as always, if this is the year of so-and-so, this is the year this happens, so it's a very convenient way for them to sell something, and they're trying to sell to the enterprise. And I think we've kind of reached a stage where we're we have these autonomous multi-step workflows that don't just output text but run code, APIs, navigate software, handle complex operational pipelines, and the speed of this shift caught a lot of these organizations off guard. Maybe it forced them to jump straight from how do we write good prompts to how do we govern autonomous agents acting on our behalf. I think maybe that's we're kind of at that stage. I don't know. It's all I can say is that this kind of wild west era that we live in, and this this rush to cash in, has does have consequences. And it has consequences not only from the security point of view, but a financial point of view, reputational point of view, and it's become a mainstream topic, and it's become, I think, one of the big talking points of the midterm elections with this whole debate over data centers, energy use, etc. And it's it's growing fast, but it's also growing with a lot of pain. And I think that in a sense kind of deters it or holds it back from from what these companies hoped we were where we would be at now.
Dave ReddySo let's let's let's it's hit on that, and I also want to go back to Amelia's point about hugging face. That's been everywhere, obviously. It's it's kind of a bit of a mild terminator story, if right to your point, John, uh about there will be consequences. To your point, Amelia, about the Trump administration is digesting regulation. Do you think this has woken up the big players to, oh, we might have a problem here. Let's be more responsible. They're just gonna keep moving down the road.
SPEAKER_00No, they've always known about this. Like, okay, I will I'll tell you when I first started writing about Agentix security. That was even before I started like working for a security publication. This was I was I went to AWS reInvent. And uh at reInvent, I spoke with the CISO of Anthropic and uh CISO of, or one of the security guys at AWS. And I just had this idea of like, hey, how do you secure an agent? Because agents touch literally everything within your enterprise. And they said, yeah, that is something that we have been thinking of so clearly. Like, and that I think that was like, I'm not, I believe it was 2024, so December of 2024, and I wrote a story about that. And that's kind of been the same thing over and over. It's like, oh, how do you protect agent, how do you protect your your enterprise from these agents who can touch everything? So that's when everyone started talking about permissioning and identity and OAuth. Sure. But what we've kind of moved into now is like, oh, agents can now tap other a other organizations, not just yourself. For like, honestly, I don't think they've changed their tune at all. They knew these were this was all going to happen. That's why they started talking about Mythos 5 as something that could really be damaging, not just because it can open up a lot of these vulnerabilities.
SPEAKER_03You know, you know, so after this breach, I'm thinking about the July 2026 breach with Hugging Face. Isn't it interesting, right before Black Hats, a lot of these companies started becoming a lot more talkative about security and security issues and what they were doing to address them. I almost think they they could treated it as a marketing opportunity. Right? They knew, I mean, I mean, to be cynical, you're a report.
Dave ReddyThat's okay. We allow we allow cynicism on this call.
SPEAKER_03Right, right, exactly, right? So they so they they they knew it was yeah, you're you're right. No, they knew exactly what was possible. And for them, they they pivoted. This is like the new story arc for them. The responsibility arc.
SPEAKER_00And it will play for the next couple of months at least until the next the next uh marketing phase comes uh Dario had a tweet, I think, yesterday, because there were a bunch of people who were saying the messaging and the marketing is a little off with AI. And yeah, this is all what it is. It is marketing and messaging. Like, I I'm not saying these things are not dangerous. They are dangerous. It's just that we've always known it was going to be dangerous, to the extent of which we didn't fully know and we still don't know. It's just that now we're leaning into that instead of what it what it used to be before. Like they were leaning towards other things. And now the choice is to make themselves scary and to push for whatever benefits or concessions from the government they can do. And they have been able to do that, especially coming from an administration who was, yes, not really very much on for regulation, but was really encroaching on a lot of what the companies wanted to be able to do. They wanted to they now they're making it so scary that, oh, Chinese models are also going to be like even more dangerous because unlike us, we guardrail, right? So now they're getting export potentially export control.
SPEAKER_03Yeah, the convenient busyman. There's always got to be somebody. Yeah. And you know, we now have without getting too political, but I will. The fear of the communism word is always brought up. It's like, okay, now we're gonna pivot to China because they are the enemy. They are much less responsible than we are. And and because we knew about security, but we're we're approaching it the way it should be. But I also think that they just were gonna hold up on talking about security unless until they absolutely had to, and now they have no choice but to talk about it.
SPEAKER_00Yeah. Yeah, but the the narrative arc is there, right? Like we're going to start by saying, we've made this thing, it's so dangerous, but you can con we can you can control us. And then the next step is there's a bogeyman, and that bogeyman is open weight models. And oh my the more convenient part of it is it's coming from China. And then you go through that narrative arc, you scaremonger everyone. Like, again, I'm not saying they're not dangerous, but I'm also saying that they also sell all of the tools to make it. Well, John, let's talk about fear. Yeah.
Dave ReddyAnd and uh sorry to trap a phobia, let's talk about that for a second. Speaking of fear, so a couple weeks ago, seems like forever ago, but I think it was only a couple of weeks ago, spoke about Jensen Wong. He comes out and he says, and in a in a statement that was clearly directed at the other most powerful person on the planet, open source models from China are safe.
SPEAKER_00I agree.
Dave ReddyYou agree they're safe?
SPEAKER_00Because open weight models, wherever they come from, are the same level of safe or not safe or unsafe, because they are open weight models. But when you also put them against a closed source model, they are also the same level of safe. Why? Because if the argument is that with open weight models, you don't know what they're putting in there, at least you can look at, well, not necessarily their source code, because it's open weight, not open source, you can see the weights that they use. And with more proprietary models, you also don't know what OpenAI or Anthropic or Google has put in. And the difference is that they very rarely share open research. Sometimes they will put out other technical details around the model, tell us about what their training data is, and that they actually stopped doing that. But it's the same level of like opacity. Yeah. Like you can't you still can't see it. It's still very opaque for both. So if that's the argument, then I don't know. The other tack is with open-weight models, there has always existed tools to guardrail it. The models themselves are not guardrailed, but enterprises can guardrail them. And enterprises already own a lot of these tools because even though proprietary models claim to have their own guardrails, enterprises tend to want to add their own. So if that's a debate, it's just it's the same thing. The difference maybe is that closed models tend to be more like, oh, open source models, yes, they are cheaper, but you kind of tend to spend the same amount of money anyway. But it's the same, it's the same thing. So for me, it's just a matter of do you want something that can be secured easily because the m it's in the model, or you have to add on to that. But to be honest, the whole open source, closed source debate for me is like it doesn't matter. You use both anyway.
SPEAKER_03So I so my interpretation of what Jensen was saying is I agree with this. So his defense centers on this thing here and architecture of open white. So once an open model is downloaded, it can be audited, fine-tuned, you can deploy air gapped on local infrastructure, eliminate remote covert data transmission or digital backdoors. But I also think NVIDIA's stance aligns directly with its business model, which is the proliferation of openweight models, regardless of origin, drives, compute to manage, hardware usage. So he's he's just noting, he's correctly noting that relying exclusively on a tiny oligopoly of proprietary models creates gradual single points of failure. So, and and again, treating open source models as neutral digital public infrastructure ignores geophysical reality. So in a sense, it is both strategic, but I also think it's logical what he's saying.
SPEAKER_00I mean, the the other part of there is that if we're really talking about concentrating the power on like certain companies, we've concentrated power on NVIDIA because everyone well, maybe not China as much, although they do maybe quote unquote steal NVIDIA chips, but like that's that's the most like clear um example of concentrating power into one company.
SPEAKER_03National security point of view that advanced foreign open models are built using smuggles hardware and distillation of Western Frontier systems. So there's that that scare tactic too, he's a law in play.
Dave ReddyQuestion though, and I and this is I I don't make political statements on this on this call. I'll leave that to the guests if you'd like to. Isn't there though a for lack of a better word, a hypocrisy in saying the folks who are saying this that open models are unsafe and then meantime the whole hugging face thing blows up?
SPEAKER_01Does that look screen?
SPEAKER_03Just ignore that, Dave. That's that's just that's just a coincidence.
SPEAKER_00Yeah, that's don't there's a reason that I made a sarcastic comment in a story and it stayed, which was but it wasn't a Chinese model that hacked that breached open it, that breached hogging phase.
SPEAKER_03It was a proprietary model because it's it's that's you know it's also the talk about hypocrisy, and I I have to I have to take a pause shot at it. So Zuckerberg is championing this open AI model because his company so far has failed to build a competitive closed one. I mean, that's well, it's just let's cut through his bullshit and get directly to where he is. I mean, these guys are all they're all talking uh conveniently what suits them best. And the one person, and I'm gonna say this, and I I'm not in the bag for NVIDIA, but in a sense, Jensen has done such a good job of trying to play above the frag and kind of be a voice of reason. And I'm not saying this is gonna last, by the way. But in a sense, he's be created or mythologized himself as this white knight while everybody else has been tarred, even Dari. And it it's interesting to me how this dynamic is playing out. They tend to they have by by far been the most successful company to date. And and when you juxtapose him, it's invariably it's against Sam Altman, who continues to be like the per the person of the year. And when I say person of the year, I don't mean that in a positive frame. You know, Time named Hitler, and I'm not comparing him to Hitler, but Time said Hitler was a man of the year. And I think with Altman, there's all this baggage and all this open AI, AI, anti-AI baggage rests solely or significantly at their doorstep. Well, I think Jensen, on the other hand, has played this beautifully well, can completely behind the scenes, being as cutthroat and as manipulative of Trump as anyone. I think he knows how to how to kind of control this guy.
SPEAKER_00Don't not don't get me started on the meta letter. It's it's because I use a lot of their technology. I like I have a problem of I'm always scrolling on Instagram and I can't focus on anything else. That is a extremely closed environment to the point that a lot of its other systems are just kind of going around each other. There's meta-ai on Instagram, right? I use Facebook, I use WhatsApp. Like I use them and I know they're not open. So don't like don't talk to me about you're a proponent of open source because everything you do is open source. Like I actually use your stuff.
Dave ReddyI'm gonna take a you guys could you guys could and can and should, and we uh I wish we had an hours to to to talk about these various things. Take a slight left turn because I do also like to speak to journalism a little bit on this podcast. And AI obviously I think is disrupting journalism. And uh, you know, hardly calling me Nostradamus when I say I think it's disrupting journalism more than even the internet did. We had the radio, we had television, Mayor Blips. When you look back on it, the internet blew up the model, and I still don't know what AI is gonna do the model. So, two questions. First of all, what's it gonna do the model? Let's start with that. Where where does journalism go from here in an AI world, John?
SPEAKER_03Okay, so you mentioned the internet, so I will try to make a comparison. It might not be good, but I will try nonetheless. Okay. So the internet kind of evolved from the standalone technology beat back in the mid-90s, and I can say this because when I was at again, I'm gonna date myself because I am old. In the mid-90s, I worked at the San Francisco Chronicle. I was hired to cover Apple. And they quickly got bored with that and they wanted to move towards the internet and social media. So that to me was a standalone technology beat, and I was the only one at the paper who was covering it. So consequently, if any type of story with an internet angle occurred, they threw it in my lap, whether it was sports, healthcare, education, etc., I was being I was being killed, I was being run into the ground covering this stuff. So it became part of this invisible background utility of every beat. And I think AI is kind of rapidly approaching that same tipping point. And what I'm thinking is within the next two to three years, treating AI as an isolated beat will feel redundant. Instead, every journalist is going to cover health, any journalist who covers healthcare will naturally cover the AI diagnostics. Every national security reporter will cover autonomous defense capabilities. Labor reporters will focus on workforce automation. I think that's how it's going to delineate, and I think there is kind of a pretty clean parallel to the internet. That's what I think. But whether by then I'll be out, I'll be out of this industry because I don't know if I I'll be gone by then because I'll be replaced by then. But Amelia, I you know, I don't know about you, but I've never been so overwhelmed by the amount of news. And I am drowning at 6 a.m., 7 a.m. every day. And I thought the internet was overwhelming. This is beyond that. And maybe we do need AI to cover some of this stuff. So we can cover things that are much that are much more that are smarter and that are maybe involved some type of institutional knowledge or a contrarian point of view, or we can cut through all the BS that's that's labeled out to us every day.
SPEAKER_00I was actually talking to a friend about this because I am I am extremely grateful that the AI beat became a beat. And like for my past like three jobs, that's been literally my job title was AI reporter. And now I'm associate editor for enterprise AI. But I wonder how long this would last. Like I I'd always wondered this like very specific, kind of very niche thing. Why isn't it just full like folded into the general tech reporter, tech editor job? And I've and I kind of like feel like it's a starting because I'd always believe that at some point everyone's going to be writing about AI. It's like we can track it in the same way how with startups. Like when I was covering startups, I was very specifically covering fintech, but all fintech startups had some sort of AI in it. And the healthcare person was also writing about startups with AI. So at a certain point, every single sector will also be writing about it. So I don't like I don't foresee this job title always remaining the same job title. But at the same time, it's it's just so interesting because everyone, like this is the the top thing. And, you know, but yeah. And in terms of like the amount of work, there are like I was so worried moving to ISMG, being the only well, everyone writes about AI here, but like the only really AI focused reporter that was gonna run out of stories. I have not run out of something to say in the morning, either.
SPEAKER_03It's gonna become so specialized. And I, you know, this actually happened. I so I started the trades in tech trade journalism, and I used to think, how am I gonna cover all this stuff? And I realized, wait a second, we're just gonna have an age of specialization where people, some people cover software, some people cover security, which I did for a while, some come some cover printers or I whatever, what have you. And I think the same thing is going is is going on now. And I it's interesting when you had uh Michael Nunez on. I remember being part of this panel with him and about seven other reporters. It's a couple years ago, maybe three years ago. And he correctly said, and you got a lot of shit for this, but he correctly said that a lot of the stuff that's gonna be produced from news organizations is gonna be AI driven. And if the reporters couldn't understand this or we couldn't accept this idea. And I'm starting to see it where I work, I'm starting to see it at other places. It's it's inevitable, but again, I th I look at Spotify, right? Uh look at these uh watermarks, like Spotify, the music, right, is gonna be watermarked, whether it's human or it's automated, the same thing's gonna have be happening with all types of content. I think the speech is gonna get bigger and bigger. You know the other thing too is I've never seen such a fascination in the among the general public towards technology. Usually, like if you wrote about TCIP, they would just they would run away. Or if you talked about Ethernet, you talk about AI, they all have skin in the game and they're all fascinated by it and they have this unquenchable thirst or or a b uh curiosity about it because they know it's gonna impact every facet of their life.
SPEAKER_00Ma something though that's been like really interesting is that I've seen a lot more people kind of be on like very polarizing sides of the technology. Like when I say I'm a reporter and I cover AI, and I also like I always say I also use AI because they do use it as a soundboard. I either get the oh, how do you how are you able to do that? Like that that sounds interesting. Tell me how that works. Like you must be very busy, you know, very curious, interested. And then I also get the how dare you. Like, how dare you like focus on this?
SPEAKER_03Oh my god I mean, so you know what's funny you you said it, Amelia. There's a friend of mine who I hike with every weekend. This person used to work in tech, is very hardcore in tech and got out. And so they're kind of from the outside looking in. And invariably the question is, so how do you use AI? Has it compromised you? It's like, no, it's my made my job a lot easier. And in a sense, I do better research than ever before, and I use it as a sounding board to uh get ideas or opinions or to push back and forth. I've asked it to edit. Sometimes I don't like what it how it edits, so I ignore it. But I'd be a fool not to use it.
SPEAKER_00Yeah, it's it's like it's it's honestly made me write much faster. And I like I constantly fight with my claw. I actually just because I was like asking it to do something and it wrote it writes terribly. And I was just fighting, I was fighting it. And it was like, you sound stupid. Like I think that's what I don't remember what I said. But it's it's been really helpful, but there's also a lot of people who are like, don't you feel like it it's it's like taking away your job? Like it's one, it writes terribly.
SPEAKER_03This is this is still it's like the internet. So the internet, how did that compromise us? It made us better at finding things or searching through things versus spending hours trying to write you know track down things, or or if we used a calculator, I mean that whole argument, you can't use technology to make your job better. Like engineers weren't allowed to use calculators.
SPEAKER_00Yeah. And I have a very romantic view of my relationship with my audience and like actually making AI write my entire story is like that's not something I am willing to do. But I also do not have the bandwidth to write an outline for the research paper that I'm writing a news story about. And that I would rather just like ask Claude or ChatGPT, just like, tell me what this is about. Give me a gist of what the outline is, and if I agree or not, I will follow. And if not, and you know, if I don't agree, I'll do whatever I want. I'm still writing the thing. You're just helping.
SPEAKER_03Yeah, and it's like I I write about lawsuits. I haven't write about them as often, but I used to write a lot about financial reports. What's wrong with asking you to sift through and find these things for you and summarize like a prospectus? I mean, I would spend hours doing this, grinding through stories, and by the end of the day, I had written maybe one, possibly two stories, and I was exhausted. And I a lot of it was spent spinning my wheels, and I thought this is totally unnecessary. So, in a sense, why would you turn your back on something that would make you more productive and also make you smarter? And you know what? I I'm not intimidated in the least by the abilities of a Claude or Gemini to edit or write something. I I'm writing this book and I wrote a chapter, and I said, Can you edit this chapter? You know what Claude did? Claude, I the the the intro to the chapter I thought actually was somewhat clever. I mean, I'll I'll pat myself on the back with just saying that. What it did is it stripped that out and put in the most generic lead as like, see, this is why I'm not afraid of of leaning on this stuff, because it it can't out sync us for now.
Dave ReddyIt's very good at associated press style writing, which is but not no offense to other other folks, warmer past paper reporters.
SPEAKER_00I should I should I should maybe let it make let it like copy edit some of my stuff because the the thing that my editor always tells me is you've forgotten AP style again.
SPEAKER_03It's like it's very good at AP. It is.
Dave ReddyBut let's let's talk a little bit about so the internet absolutely disrupted the the journalism business model. I I and and that's you know, you go to Andy Barnes and Noble and you'll see that because there's not that many magazines there. And I, you know, candidly, I'm in a guy with a journalism degree, and I've been in public relations for nearly 30 years, and I can't I can easily count the number of times I've picked up a British domain. What is AI gonna do to that? There's still four or five thousand jobs, uh according to uh research lost every year in the United States in journalism. Is AI gonna fix that? Is it does it have nothing to do with that? Are people using it as an excuse to clear out newsrooms? Go.
SPEAKER_00I have like a two-pronged link to that. One, I I don't think AI will be as disruptive as the internet has been to journalism. Because what the internet enabled was like at first a flourishing of the number of publications you can write for. And it created a whole new subset of journalism, digital media. That's just not just print, but it's also video, it's broadcast, it's audio, it's it's all of that. Sure. But it's also enabled the rise of too much of these things. And now that we're seeing, that's why we're seeing the collapse essentially of also digital media. I'm not entirely sure that AI would sort would change journalism in that way. It might become another avenue to distribute public articles from websites, sure. I think a lot of the disruption in the kind of the employment space of journalism that AI has wrought is yes, it's become kind of an excuse for a lot of companies, because it is, to be fair, cheaper to make an AI that can write the very quick breaking news stories. Like that is completely fair.
SPEAKER_03And I understand that by the way, it does happen.
SPEAKER_00It has happened.
SPEAKER_03Years and years ago, I'm gonna say 15 years ago, told me that they were in the early stages of using AI. They were experimenting with it on kind of small stories that they didn't have the bandwidth to do.
SPEAKER_00So that's that's they use it for sports and finance.
SPEAKER_03And yeah, I gotta tell you, so for years, all I did was bitch you complain about doing earnings stories, especially when I worked at a place like USA Today, and I said, Why am I doing this story when we could just use a wire story, which is what AP, which is what USA Today does now. But I want to say, going back to journalism, we were well on our way to losing jobs before AI. AI is just a convenient excuse because people do not read. I mean, things change. Yeah. People used to all listen to radio. They used to watch uh traditional television before cable came along, and they don't watch cable as much. Well, we have kids. We have people we know who are younger who depend on their phones. I mean, this was happening anyway. This is, I don't even think it's accelerating it. I think it's just creating another vehicle.
SPEAKER_00Yeah. The way I see it is there people will still continue to want to know what's going on with the world, and they will shift their questions from Google to an AI chatbot. But what these AI chatbots still need is the content. And whether or not that content comes from the content farm or a traditional kind of quote unquote legacy news media website, that content will still be there. The matter is that who is writing it? I I hope this doesn't happen, but I know it will. We're going to have newsrooms where there's only maybe two or three actual people, and the rest is maybe in AI. But I it's like I said, it's probably a new distribution layer. But there will always be people who will want to write. In whichever form it might it it comes out of, it'll it's much less the fault of AI, more the fault of the comp of the people who are own who own these companies. That's why we're seeing the rise of Substacks of all of these different places.
SPEAKER_03So the rise of Substack, the rise of all these highly specialized, extremely knowledgeable news sites that I think are tremendous, versus where I used to work at Kennet and Dow Jones, where we were all chasing the same story, presenting almost the same story in pursuit of the SEO. And it was just this downward spiral, this chase to the bottom. And you could see it coming for years. I mean, I remember going through like 15 rounds of layoffs the last eight years I was at USA Today. And I thought, you know, the reason why this is happening is we're not creating anything that's compelling, that's outside the box. We're all just pursuing the same stories, trying to protect our turf because we have a bigger platform than others. And that's it's not, it doesn't work. And you have to you have to go to where the reader goes. You have to think how the reader thinks. And most newspapers were never, especially newspapers, were never very good. They they had a, in a sense, a monopoly of sorts in their region or their areas, and they were never forced to think beyond this the the same audience. They didn't think about a younger audience, they didn't think about advertising and how that might change. And in a sense, they have themselves to blame, but also in a sense, things change in society. There are no payphones, there are no payphones anymore, there are no telephones that I know of. Right?
Dave ReddyRight. Yeah. So maybe it'll usher in an era of good old-fashioned well, not good old fashioned, but yeah, I guess good old fashioned enterprise recording or more analysis, which we're seeing, and maybe it'll maybe it'll actually it'll help in that regard.
SPEAKER_00Which, to be fair, has been happening long before Chat GPU came out. Like even back in 2016, my first job out of grad school was already asking me to not write press releases. It was like look for the deeper story. So that like that's been going on, and that was more market pressure because of dwindling advertising, because of the lack of clicks and that kind of focus on metrics rather than the quality of the story. Like that was always happening. AI is just another thing that is go that is a problem to contend with.
SPEAKER_03She was a great reporter, she's a great enterprise reporter. She was actually among the first to get laid off at the paper uh when we were at USA today together. And she would write maybe less than 10 stories a year, but they're tremendous stories. But the the appetite among the people back east was you need to do more. So what she told me was if you want to be an enterprise reporter and I want and I will help you do this, you have to learn to do you have to learn to do this. You've got to feed the beast and give them what they want in terms of the kind of low-hanging fruits, and then spend your extra time if you can find it at night or on the weekends doing the stories you really want to do. And unfortunately, I I followed that and that was the only way to get stories uh like projects or enterprise stories or longer-term things published. And they loved them, but they just didn't want you to spend time doing anything above and beyond the simplest types of stories.
SPEAKER_00I guess on the AI reporter front is that I majority of what I am supposed to be doing are enterprise stories, but because of the sheer number of things that are happening in the world, I often have to push back a lot of stories. Like I do a lot of do a lot of like case study stories where I talk to either a company that's made a model or an agent or talk to an end user and how they're using an agent. Because these are exclusive stories. I have to push them back because hugging face was a week, because OpenAI released something, because there's a new Kimmy model. So it's instead of that question of like, oh, you're only really provide like feeding the beast, it's like, yeah, because even yeah, I don't have a choice because even if the job description is to make is to write enterprise stories, I do not have time to I'll tell you something that's I find it very depressing, but I also kind of find it humorous.
SPEAKER_03So this guy that I work with, we did a project for a couple of years on cybersecurity and on personal information. This was back like in 2005-6 era. So the paper to its credit let us do it. And I remember we spent weeks and we overwrote and rewrote this story, I'm gonna say dozens of times. This is a story that was on the front page of the paper. It was 4,000 words. And I remember somebody back east saying to me, Congratulations, you wrote the longest story in the history of this paper. It'll never happen again, but I just want to congratulate you. So now go now, go write about Facebook's latest uh feature.
Dave ReddyAnd by the way, for those who are listening who did not go to journalism school, the enterprise story is not a story about enterprise technology, although it could be. It is more of a long lead, featury story, something where John or Amelia would go out and report. Now that I've been pedantic. Let's finish with the predictions as we always do. So it's a year from now. We are coming up, believe it or not, on the five year anniversary, a year from now of Chad GPT2. We've just walked out of AI for 2021. Which I presume will be in Vegas since everything's in Vegas now, or maybe maybe uh Davos or Dubai. What are we talking about?
SPEAKER_03It's gonna be the year of agentic AI.
Dave ReddyWell, we know that. What's the what God? I just said word on my own podcast. What is the take on it?
SPEAKER_03Okay, I'm just gonna say something quick and Amelia can end this. But I what I hope would happen, and I know it's not gonna happen. I hope a year from now the companies that tell us about all these jobs that AI is gonna create that are new jobs for the workforce, we start finding out what the some of those jobs are. Because I um I'm not having a lot of luck finding out through the companies what this whole new wave of AI created jobs are that we're gonna that's gonna help the workforce. That's what I hope, but I qu I honestly my predictions are terrible. So I'll leave it to them.
Dave ReddyThat's an optimistic one. We'll go with some optimism. Amelia, you get the last one.
SPEAKER_00By the way, by the way, that job's not gonna be prompt engineer because it's a job that came and went. I I think one, we're not gonna be talking about AGI because we don't know what AGI really is. I do think we're still going to be talking about securing powerful models. We we are definitely still going to be talking about the open, closed source debate, probably not specifically about China. I do think it's going we're going to start really talking about I was gonna say weaponizing, I guess that's a good word. Weaponizing either or both for enterprise usage. How best to use your hybrid portfolio of models. I definitely think we have already been talking about that, but we're definitely going to be expanding more around that, the actual use case, capital, all caps on use. Because yes, every enterprise does have some form of AI. They're not necessarily using it in the way they want to be using it. I do think we're gonna start seeing the first forays into this, and it's going to be very interesting to see how bambi-legged everything's going to be. Because I think we're going to start seeing a lot of companies suddenly shed their agentic programs. And then we're also going to see some companies really grow it. Because I think we're going to start seeing companies just realize that we've been spending way too much money on something we don't really need. And the hype is dying down because there's a lot more bigger concern. There's a lot more concerns around this. And I also think that because there's a lot of pushback on data centers, it's going to get more expensive. Those that's also going to be pushing a lot of companies to really be like, okay, so what is the actual use of this? How do we best kind of optimize this whole situation?
SPEAKER_03Wall Street's going to be so a year from now, ostensibly, open AI and Anthropic will both be public companies.
SPEAKER_00IPO, yeah.
SPEAKER_03So we're going to have Wall Street say, wait a second, this revenue is not what we it should be. So you're going to see a maybe a temporary boom in the stock and then kind of a reckoning of sorts. Boy, is that dark.
SPEAKER_00You know how Amazon wasn't making money for the first like 10 years of its life as a public company? Those are that's what open AI and Anthropics are both going to be because they're mud they're companies built on loans.
SPEAKER_03Yes, the circular financing and the debt financing. It's you're right. It's probably gonna be that, although I think anthropic will be given more rope. I think open AI, there's like this suspicion or this this this conspiracy belief, or maybe it's just this, maybe it's it's it's uh it's a knowledgeable belief that they're kind of in a sense like a modern-day Enron. I know that's a bit much, but there's a little bit, there's awariness around them given the turnover of executives, especially those in the financial side of things.
SPEAKER_00What I do think is that maybe by AI for next year is that we'll only have two big model providers as public companies. Uh when I say two, too new because Google and Meta have been public for a while, because I don't think there will be any other company. And to be fair, like other than Meta and Google, there's really no other kind of competitors. But I think we're going the IPOing of AI companies is going to stop at OpenAI and Anthropic, at least for the next year or so, simply because I don't know exactly how much market interest there really is on anything else. And I think application providers like being private.
Dave ReddyWell, the so well, there goes the optimism. I I'm a pessimist by nature, so I think everyone with a journalism degree is.
SPEAKER_01I can't. I haven't got the moment. We'll have to get back at 13 a.m.
Dave ReddySo thank you so so much for being part of this special show. I really appreciated all your thoughts, and we'll definitely check you both on your predictions next year. Promise. That'll be a holder one. Kidding. Thanks again for your time.
SPEAKER_03Congratulations for four years.
Dave ReddyThanks, John. I'd like to thank you all for listening today, and once again, a big thank you to our special guests, Amelia David of ISMG Group and John Swartz of TechStrong. Don't forget to join us next month when we kick off our fifth season with yet another member of the B2B Tech Top 200. In the meantime, if you've got feedback on today's podcast, or if you'd like to learn more about Big Valley marketing and how we identified the B2B Tech Top 200, be sure to drop me an email at d ready at bigvalley.co. That's d r e double d y at bigvalley, all one word.co. No M. You can also email the whole team at pressingmatters at bigvalley.co. Once again, thanks for listening, and as always, think big.
