This week on The Marketing AI Show, all three topics focus on AI and writing. This mega-episode brings to the surface many of the questions we’re hearing from our community. AI writing tools are rapidly transforming the art and science of storytelling, and in the process, they're disrupting writing as we know it. Everyone at every size company now has access to powerful and affordable AI technologies that are redefining writing and forcing companies to reimagine content teams and strategies to remain competitive.
We can already 10X our content output, like ads, articles, blog posts, emails, and more, with today’s AI—and it’s growing more powerful by the day. This has created massive uncertainty for writers, brands, agencies, and media companies.
And it’s made it essential that writers, editors, and content leaders take action to understand and apply AI.
Today’s podcast episode is a mega-episode on AI writing and the impact it will have on writers and writing as we know it—and it kicks off with a discussion on the future of AI and writing.
Funding is flooding the space, and we’re already seeing media businesses incorporate AI writing into their business models—with mixed results. But these businesses are moving forward anyway. It’s important for marketing and business leaders to remember that a tool like ChatGPT is just the tip of the AI iceberg for marketers and business leaders. It's the shiny object that's captured everyone's attention, and rightly so.
The second topic on AI and writing touches on the AI writing tools.
These tools can enable all sorts of highly valuable use cases across writing functions. Used intelligently, these tools can dramatically improve your productivity and performance as a writer.
On the other hand, they also have plenty of limitations that any writer using these tools must be able to identify if they want to succeed at AI-powered writing.
How should these tools be used by individuals and brands? We believe that understanding the very real issues raised by this technology is key to adopting it in a responsible way.
And finally, how will AI change writing careers? AI is going to have a major impact on the careers and skills of writers and content professionals. These more intelligent tools are already opening up new frontiers of creativity, productivity, and writing at scale for savvy professionals.
But they’re also changing the nature of writing work and business models, and writers and companies will need to adapt accordingly.
Mike and Paul invest these 45 minutes in helping listeners understand what these new developments mean, providing context on what marketers and business leaders can do, need to consider, and how to adapt.
Listen to this week’s episode on your favorite podcast player and be sure to explore the links below for more thoughts and perspective on these important topics.
Timestamps
00:04:24 — The Future of Writing in the Age of AI
00:20:26 — AI Writing Tools: Capabilities, Limitations, and Concerns
00:31:45 — How AI Will Change Writing Careers
Links referenced in the show
- Death By a Thousand Personality Quizzes
- Paul's LinkedIn post
- AI for Writers Summit | Marketing AI Event
- What Will Happen When A.I. Makes BuzzFeed Quizzes? - The New York Times
- BuzzFeed preps AI-written content while CNET fumbles | Ars Technica
- BuzzFeed's stock soars another 73% as analysts weigh in on AI shift and Meta deal - MarketWatch
- Artificial intelligence pioneers back $550mn fund for AI start-ups | Financial Times
- What Are Large Language Models Used For? | Nvidia
- ChatGPT Is Just the Beginning—Here's the Real Story
- Cohere tweet
- Generative AI with Cohere: Part 2 - Use Case Ideation
- A Marketer’s First Experience With ChatGPT From OpenAI
- 40+ Jaw-Dropping AI Writing Tools
- Free Ebook: Using Generative AI to Scale Your Content Operations
- Yann LeCun on Twitter: "LLMs are still making sh*t up. That's fine if you use them as writing assistants. Not good as question answerers, search engines, etc. RLHF merely mitigates the most frequent mistakes without actually fixing the problem." / Twitter
- Inside CNET’s AI-powered SEO money machine - The Verge
- CNET goes into damage control, pauses AI-written articles
- The Content Cyborg: How to Use AI Writing Tools in Content Marketing - Animalz
Watch the Video
Read the Interview Transcription
Disclaimer: This transcription was written by AI, thanks to Descript, and has not been edited for content.
[00:00:00] Paul Roetzer: So I wonder like as a content strategist, if you don't find the ways not to intelligently automate more content, just pump out more SEO content in the world. But how do you, like we've talked about before, how do you become more human? Yeah. As a brand. And maybe it is more stuff like this where you have podcasts and live events, you know, like that you people are gonna crave human content and interaction.
[00:00:26] Paul Roetzer: I think probably more than ever.
[00:00:28] Paul Roetzer: Welcome to the Marketing AI Show, the podcast that helps your business grow smarter by making artificial intelligence approachable and actionable. You'll hear from top authors, entrepreneurs, researchers, and executives as they share case studies, strategies, and technologies that have the power to transform your business and your career.
[00:00:49] Paul Roetzer: My name is Paul Roetzer. I'm the founder of Marketing AI Institute, and I'm your host. Welcome to episode 32 of the Marketing AI Show. I'm your host, Paul Roetzer, along with my co-host Mike Kaput, chief content Officer at Marketing AI Institute, and co-author of our book, marketing Artificial Intelligence. Today's episode is brought to us by the AI for Writers Summit, presented by Writer. This is a new event.
[00:01:16] Paul Roetzer: We actually just. A week ago, right? I think it was last Monday. Yep. Um, so recording this on January 20, or January 31st. So end of January. And the premise of this event is there's so many unknowns right now in writing the impact that these AI writing tools are going to have on writers, on editors, on content teams.
[00:01:38] Paul Roetzer: We're hearing it from, uh, people in the Marketing Institute community. We're feeling it ourselves as writers by trade. I'm, I came outta journalism school. Mike was a journalist by trade, um, before we, you know, started working in AI and marketing. . And so for us, we almost felt this like obligation to create something to bring everyone together and try and figure this out.
[00:01:59] Paul Roetzer: So, Mike's wife is a writer. Like, it's just, it's everywhere. We, uh, I talked with the journalism school where I graduated from college. We're trying to figure this out for the next generation, and we thought that if we could do a free virtual event, that we could reach the masses and try and kind of together.
[00:02:17] Paul Roetzer: Solve where this is going and ask a lot of the hard questions. So, uh, that's what we're gonna do. So it's the, the event is March 30th. It is gonna be from noon to four Eastern time on the 30th. Like I said, it is free. Um, there's a free pass courtesy of writer, our sponsor. You can do a, a private, uh, registration for $99 if you don't wanna share your contact info.
[00:02:38] Paul Roetzer: Um, and there's also a on-demand option, but overall it's a free event. We've had over 800 people, I think last I looked already registered for this event in the first week. So we were, we were thinking a thousand to 1500 was realistic. And, and now I'm, I'm thinking it might be a little bigger than we originally planned, which is fantastic.
[00:02:56] Paul Roetzer: So, uh, ai writer summit.com. We will put the link in the notes if you are interested in that. I highly encourage you to join us for that event. Um, I, I think we're going. One, A lot is going to happen between now and then . So I think there's gonna be more to talk about March 30th than, uh, we even know, but it's gonna be incredible.
[00:03:18] Paul Roetzer: Event we have, uh, Anne Hanley is gonna do a closing keynote. Um, wall Street Bestselling. Wall Street Journal, bestselling author of everybody writes, and a amazing friend, um, and marketer and speaker. Uh, may Habib from writer will be there. Mike's gonna be doing a talk on AI writing tools. I'll be doing a keynote on the state of AI and writing.
[00:03:37] Paul Roetzer: Should be a lot of fun. Um, again, uh, ai writer summit.com. Check that out. And then that leads us into today's episode, which is all about AI writing . So I will let, uh, Mike take it away with our three big topics for the week.
[00:03:54] Mike Kaput: Thanks Paul. Uh, like you said, we are talking everything and anything related to AI's impact on writing.
[00:04:01] Mike Kaput: We're going to still follow our kind of three topic format, but we're going to tackle in each topic a one big area of how AI is going to change writing as we know it, and we're going to get your thoughts and have a discussion around the very themes that we'll be diving deeper into. The writer Summit as well.
[00:04:23] Mike Kaput: So first up here, I want to get your thoughts and talk a little bit about the future of writing as a practice, as a career, as an industry. So we have all these AI writing tools that are rapidly transforming how we do storytelling, and in the process they're disrupting writing as we know it. Everyone today, at every different size company now has access to powerful and affordable AI tools that are redefining writing and forcing companies to reimagine their content teams and their strategies in order to remain competitive.
[00:05:02] Mike Kaput: You know, even in our own experiments we've seen that we can already. 10 X are content output. Things like ads, articles, posts, emails, and much more with today's ai, and it's all growing more powerful by the day. Now, like you alluded to talking about the writer's summit. This has created, created massive uncertainty for writers, brands, agencies, and media companies, and it's made it essential that writers, editors, and content leaders take action to start understanding and applying AI in writing.
[00:05:38] Mike Kaput: So today's mega episode of the podcast is really gonna kick this off by talking about. Kind of the bird's eye view of the changes that we're seeing coming to writing and the work of writers thanks to ai. So, Paul, give me, to start things off here, tell me how do you see AI impacting the future of writing and the future of writers?
[00:06:05] Paul Roetzer: Yeah. The way I look at this is there's just so much uncertainty, and I don't necessarily have the answers. I think that there's a lot of organizations, whether they're brands or media companies or agencies that are looking at the power of these AI writing tools and questioning, do we need as many writers as we previously had?
[00:06:24] Paul Roetzer: And I think that's a. It's a logical question, like I, I, I think that there's a possibility that maybe you don't. Um, but then when you start to understand the limitations of these things and the need for fact checking and the need for human layers of. Uh, you know, writing over the AI writing, you start to realize that there's way more complexity than just, oh, okay, I can write now we can get rid of our, you know, some of our writers.
[00:06:50] Paul Roetzer: The evolution of what an editor is and does, uh, evolves. The, um, you know, I think we saw, and we might talk a little bit more about like Fiver, the freelance platform just this week or last week announced the AI um, services, one of which was back checking, was. one of the services they offer because they know that the AI outputs things that it doesn't know facts, it doesn't know dates and places and times and people and all these things.
[00:07:15] Paul Roetzer: It can just make stuff up. So I think that if I had to pick a word to describe the future of writing and the writing profession, it's uncertain. Like we just don't know and we can't really. Very accurately project because we also don't know what the next version of these writing tools is going to be.
[00:07:36] Paul Roetzer: You know, we, this mythical G P T four that we keep talking about and hearing about, and you know, it probably will come at some point this year, maybe sooner than later. We don't really know what it's gonna be and what it's gonna be capable of. And so I just, I feel like it's a very uncertain time. And that's why I think, you know, the summit that we were bringing together, it's like, well, we, we have to talk about this.
[00:07:57] Paul Roetzer: We have to get different perspectives. We have to understand the legal implications. We have to understand the impact on workforce. We have to understand the impact on down to journalism schools and what do they even teaching there and how do you prepare students? So yeah, uncertainty is my biggest thing right now about where it's going, um, and what kind of we're facing today.
[00:08:17] Paul Roetzer: And what
[00:08:18] Mike Kaput: we're also seeing is that that kind of uncertainty without an industry conversation can really create some fear. We had a member of the Marketing AI Institute community say something to the effect of, to us recently, when I mention AI to any writer on my team, they literally s shutter. I think there's a lot of fear and angst around what this means for writing.
[00:08:42] Mike Kaput: So if I'm a writer or someone adjacent to writing who's hearing all of this news about chat j p t about ai, I'm probably feeling not only uncertain, but a little afraid. What would you say to writers who are afraid of this stuff and tempted to kind of, you know, either freak out or stick their head in the sand and say, nah, this can't be happening.
[00:09:04] Paul Roetzer: It's not going anywhere. Like you, you have to face it head on. Uh, the advice we've given previously is don't just test our writing tool. Test five of 'em. Like the, the only people that are gonna figure this out are. Are probably the writers and the editors who can figure out how these tools actually fit into their workflows and how it affects what they do and how they can maybe improve upon what they do.
[00:09:30] Paul Roetzer: So I would just encourage people, as scary as it may seem, or abstract, or even sci-fi in some ways, I, I get that if you're a writer that just doesn't seem possible that an AI can do what you do. I understand that feeling, that writing is uniquely human and creativity is. Uniquely human. Um, but if you haven't tried 'em, you can't really pass judgment or understand it.
[00:09:54] Paul Roetzer: So I would just say, uh, as hard as it may be, be proactive and go try these tools. They're affordable. If, if you can't get approval to, you know, budget within your company, just spend the hundred bucks yourself and go try a couple of 'em. Because once you do, I think you can start drawing your own conclusions about the impact it's gonna have and maybe the opportunities it presents to.
[00:10:17] Paul Roetzer: And I would try and go at it from a positive perspective. I think we have to accept that machines are here and they, that AI can write . Um, you can question how good it is or what formats it might be weak at, but it's, it is doable, um, today and it's accessible to everyone. And I think you should go into it saying, okay, what is the opportunity here?
[00:10:41] Paul Roetzer: Like, I, I just have to accept that they will be creators, we'll side by side, be creating content. Um, this is something Anne Hanley and I have talked about on, you know, recent webinar we did together, and I know she'll talk about it in her keynote. Um, different writers will use these tools differently. So like, as a, as a writer, I mean, I write, I don't know, a few thousand words of content a week to publish to the blog post or LinkedIn.
[00:11:06] Paul Roetzer: Rev and written books, and I honestly don't use them that much. So I, I consider myself a fairly good writer. Um, I don't use them in drafting my LinkedIn post. I don't use them in drafting my blog posts. I, I, I use them mainly for ideation and experimentation just to see what's possible if I'm creating an outline for something.
[00:11:27] Paul Roetzer: I'll still create my outline and then I'll go into the tool and say, create me an outline for this, uh, an ebook on this topic, and I'll see what it comes up with and maybe it has some other ideas I didn't have. So I mainly use the I writing tools for ideation, summarization, you know, transcription, things that I just don't have the time to do or desire to do.
[00:11:46] Paul Roetzer: But the actual process of writing I enjoy and I find is a very strategic thing. And I like to go through that process, so I don't need or want them to take that from me. Um, but there's a lot of people who aren't good writers, you know, that that's just not what they're trained to do. And for them, it may become a true assistant.
[00:12:10] Paul Roetzer: Like everything, they create emails, blog posts, articles, proposals, um, scripts, outlines everything. They may start with a first draft in ai and that's fine. It's for everyone to figure out how they want to use the tools, and I think that's what's maybe being missed in all of this is that we still have choices.
[00:12:32] Paul Roetzer: Like the tools are there to be assistive to us in however we choose to use them, and different people will choose to use them in different ways, and that's okay. That's
[00:12:43] Mike Kaput: a really good point, and I think it's worth reiterating to the audience that if you are a writer, there's no substitute for trying this out on your own because like we've talked about in a couple other podcasts, With how popular things like chat g p t are.
[00:12:59] Mike Kaput: There's a lot of what I would call bad takes out there from mm-hmm. . People that have just suddenly discovered chat, G p t don't really understand the wider AI opportunity and are kind of really just regurgitating the same takes, the same prompts, the same use cases for the tool and like, you know, there's nothing.
[00:13:21] Mike Kaput: Terrible about that, but you really can't read a couple of expert takes and get the full picture. Everyone's work is different. The use cases for tools like Chat J P T are different and deeper and more expansive than you might expect from just reading kind of the latest think piece on
[00:13:41] Paul Roetzer: the tool. Yeah, you have to experience it yourself.
[00:13:45] Paul Roetzer: It's like anything else. And I think that's the beauty of where we are today, is you can do that right now. Like you don't, when you're done listening to this podcast, just go get an account on Jasper or writer or hyper writer or wherever, or get an open AI playground account or a coherent playground account.
[00:14:02] Paul Roetzer: Like take your pick, do a few of them. Um, and I think we, we did an episode on that where I actually went through and tested. Five different AI writing platforms, and if you haven't listened to that episode, go back and listen to it. That's is exactly how we approached it. Like, well, let's figure this out. I don't know, like, I'm not sure.
[00:14:18] Paul Roetzer: As an institute we publish what, like 150 plus articles a year on the blog? Yeah. Like we write a lot and we do a podcast a week two, one to two webinars a month. Like we create a ton of content and we didn't have a, like a plan in place of how exactly we were going to infuse AI at. . It was me experimenting.
[00:14:38] Paul Roetzer: It was Mike experimenting. It was Cathy experimenting with descript and some other tools, and we were like individually figuring this out. And I stopped and was like, well wait a second. We have like six different tools we can be using right now. Which one should we use, which should be integrated into our daily workflow versus continue to experiment?
[00:14:56] Paul Roetzer: And so we started this project copy. 2.0 to try and figure this out. Like how are we gonna infuse these tools? And that's what I would just advise companies to do is like, figure this out for yourself. Don't listen to us, like use what we're saying as base knowledge, but you have to go experiment and figure out how it's gonna evolve in your own company.
[00:15:16] Paul Roetzer: It's the only way to do it. Don't listen to any of the talking heads in terms of like, oh, okay, that's the blueprint that I have to follow. It's, it's just take all the knowledge in and figure out what matters to you.
[00:15:30] Mike Kaput: So before we get into our next, Writing focus topic here. I want to end this discussion with one last question about how we see AI impacting the future of media companies.
[00:15:42] Mike Kaput: Now. Two quick stories in the last couple weeks. Jump out here. The first is a cautionary tale from, uh, Popular website called cnet and CNET is technology website. And they found out that the parent company that owns it, which has a very serious focus on ranking articles and Surge had um, used AI to publish content without everyone at the company knowing it was being published.
[00:16:11] Mike Kaput: It created a pretty big furor over. Who knew what and which writers knew which work was human generated, AI generated, what have you, and eventually CNET said, okay, we're gonna stop using AI for this at least temporarily. Now, that was kind of a PR nightmare for them. On the other hand, Buzzfeed just announced that they are leaning quite heavily into.
[00:16:39] Mike Kaput: And using it at least to start to create some of their very popular quizzes. And their stock went crazy. It basically was up like 300% based on the news. So whether that holds steady or not, the market started to reward them for adopting artificial intelligence. And my question is, how do you see this impacting business models and media companies moving forward?
[00:17:09] Paul Roetzer: The CNET thing is a perfect example of what not to do. Like it just, they, they tried to hide it, um, when there's a lack of transparency and authenticity and you do something for a quick dollar, which is exactly what they were doing, was just pumping crap content into the internet to put affiliate links into it and drive traffic.
[00:17:29] Paul Roetzer: Get what you deserve. Like it, it's just a shortcut way to do content and marketing and it's as old as any like content farms of the early teens. Um, you know, so I, I just have no interest in what they were doing and I think, I'm sure it's gonna be repeated and I'm sure it'll end in the same result eventually for whoever does that.
[00:17:49] Paul Roetzer: So hopefully you don't follow that path, like that's just a bad idea. Buzzfeed is interesting. When I saw it, I was like, okay, like, you know, fine. Cool. Um, I saw the stock market, the stock price jump and wish I would've been able to short it. Like I, like, it's just, there's no way to sustain that. And I even saw an employee yesterday, like there was an interview where she was like, yeah, we're not using it the way that people think we are.
[00:18:17] Paul Roetzer: Like , the way Wall Street believes we're using AI is not how it's actually gonna be. And, uh, if, if they knew that, like it probably wouldn't, the stock probably wouldn't have jumped. So I, that was a anomaly in my opinion. Um, if it was that easy, like a bunch of companies would just come out like, oh yeah, we're using AI too, and like stocks would get pumped.
[00:18:37] Paul Roetzer: And, uh, I think it's fine that Buzzfeed's doing what they're doing. It sounds like it's probably actually a more creative use of it than was maybe let on. , which is cool, but I mean, Buzzfeed certainly does their share of creating crap content. That gets a lot of clicks in traffic too. I, I don't know. So I, I'll be fascinated to see, I think there's gonna be a lot of really innovative uses though, that are gonna be worthy of studying.
[00:19:09] Paul Roetzer: I've always admired Buzzfeed's model. I mean, Buzzfeed's been doing some insane stuff with AI for years. Like Yeah, predictive modeling around what to publish and when. So whether you like buzzfeed's content or not, the reality is they were a very early mover in the application of AI to their content strategy and production and promotion.
[00:19:29] Paul Roetzer: Mm-hmm. and uh, I remember like there was a video of one of their, I think it was their head of AI or head of content strategy, something that we actually tried to get as a speaker at Macon the one year, because I watched his YouTube video. I was like, this is insane. Like this is brilliant stuff. So I would pay attention to what Budges feed's doing cuz they have a history of doing some really cool things.
[00:19:49] Paul Roetzer: But I think there's, yeah, just, I mean, media companies, which you could argue most brands are kind of view themselves as media companies in a way, publishing content, building an audience. You have to figure out how to infuse it into your strategy. Like you can't be a media company moving forward without AI at the center of everything you're doing.
[00:20:07] Paul Roetzer: So I think that's the real lesson here, is just like you can do it the right way or the wrong way, but it has to be at the center of everything.
[00:20:17] Mike Kaput: Absolutely. So I think that's a really good kind of high level view of what we're looking at and talking about when it comes to AI and writing. So I wanna actually switch gears and maybe bring it down to earth a little bit by talking about some AI writing tools, specifically their capabilities, limitations and some concerns around them.
[00:20:37] Mike Kaput: So I think we've kind of alluded to some of the main. Ways in which writing tools can be used. But could you just give us a really quick survey of just what are they capable of doing today? If I'm just hearing about this with like the chat g p T news and all that. Like, can AI write for me? Can it edit for me?
[00:20:56] Mike Kaput: What are we talking about exactly
[00:20:58] Paul Roetzer: here? . Yeah. The chat G p T thing. You know, the, what caught everybody by surprise was its ability to do a lot of general purpose fighting. Mm-hmm. , like any task you gave it, write me a blog post, write me an article, write me a press release, write me an email, write me a summary of this and that.
[00:21:13] Paul Roetzer: It could just do, it was I think what. Really caught people by surprise was how well it adapted without having to pick a model or a template like you do in some of these ad writing tools where you're, you know, selecting, I want an ad or I want an article, and then you're putting a bunch of inputs in. In this case, you're just saying, I would like an article on this and I want it to be this length and this tone in this style, and it just does it.
[00:21:38] Paul Roetzer: That's kind of where we are, is like you're now, whether you're working directly with the language model company, so OpenAI makes the language model. Same with Cohere. They like, they build their own language models and then third party vendors like Jasper, writer, hyper, right word, tune whoever they can patch into the APIs from those companies and then build user interfaces and solutions on top of.
[00:22:02] Paul Roetzer: Now some of these companies are also developing their own language models, or at least using their own proprietary data to tune and train these models further than what you can get by going direct to open ai. Um, but generally whether you're going to open AI through chat, G P T or going to one of these writing tools, you have the ability to generate language in any form you would today.
[00:22:22] Paul Roetzer: So whether, again, it's emails, proposals, articles, um, video scripts, whatever it is, They're getting really good at just developing drafts. The, I think the misconception at the moment is that, um, they're like, publish ready drafts. I, I wouldn't think about it that way. I would think, um, you know, at the most extent you're, you're creating a really solid first draft of whatever it is you're creating, but assume you're gonna need the human in the loop and you're gonna need an editor and you're probably gonna still want a writer.
[00:22:52] Paul Roetzer: Um, so outta the box, they're able to do. When you start getting into stuff like what writer is doing, which is just writer.com, uh, if, if you haven't been there, like they'll actually train the models on your brand guidelines, like they're specifically built for teams. And so it can actually be trained on your own data and your guidelines and tone and style so that when it is outputting something, it's actually fitting within your desired outcome or output based on your guidelines.
[00:23:21] Paul Roetzer: And so that's, that's possible today too. So I think. Different levels depending on what you're using it for, but that's generally what you're able to do, is it? It really can write, and you and I, we've told this story back in like 2015. We asked this question internally. We were a content agency at the time, like we were both working for my agency at the time, and it was like, can AI write, like we were doing press releases and articles and blog posts and eBooks and white papers and all this stuff for.
[00:23:48] Paul Roetzer: And I'd come to believe that AI was capable of doing it well, the answer at that time was no. It actually couldn't. We were sort of like misguided by some of the vendors in the market at the time, but that's not the case now. Like an, the answer is yes, it can write.
[00:24:02] Mike Kaput: Yeah, and without getting too deep in the technical weeds, it's really important for writers or people in this space where in the audience listening, if you tried out these tools several years ago and were unimpressed, you need to take another look because things several in several months ago, for sure, several months ago in some cases.
[00:24:21] Mike Kaput: Because since about 2017, with things like the development of transformers, which have made some large, these large language models really robust and possible, uh, the last several years have just been essentially a golden age of this technology really getting very, very, very good. And it's just the beginning.
[00:24:40] Mike Kaput: Yeah, for sure. So they can do these large language models and tools built on them, can do some pretty stunning things today. But what limitations. Should we be expecting as of today when we're using them?
[00:24:55] Paul Roetzer: The main ones that jump out to me, and you can add in if you can think of other, is, uh, lack of citation.
[00:25:01] Paul Roetzer: So you don't know where the data's coming from because it's not actually, as we've talked about before, copying and pasting anything. It's not just grabbing a line from Wikipedia or corporate site and then regurgitating it. It's writing original content. It's predicting the next token or word in a sequence, and it's generating it based on its training data.
[00:25:19] Paul Roetzer: So, It doesn't really have citations per se, uh, in that way, but that's a major deficit and it's a major thing that, that companies are working on. Like deep Google DeepMind, you know, has talked publicly about Sparrow, which is their chat, G p T style, um, agent, and that's supposed to have some kind of citations within it.
[00:25:40] Paul Roetzer: I, I've seen lately that the expectation with chat, G P T has allowed some sort of citations in it. Mm-hmm. , I don't know exactly how that'll work. . Um, I see people talking authoritatively about this, and I was like, how do you know that , I don't, I don't understand how, it's not public knowledge yet how they would exactly do it, but I think the way it would work is almost like the, well, there's two ways I could see it.
[00:26:02] Paul Roetzer: So one would be, it, it, the, the AI tries to sort of predict what sources influenced its output. Mm-hmm. . So it, it writes something basically, and then it tries to like, Where it learned data from and it's like, okay, these three links probably are the closest to site where, where it learned this information from.
[00:26:23] Paul Roetzer: The other thing I assume is going to happen is actually attaching a knowledge base or search function to the outputs. So as it's outputting something and writing this original content, it's simultaneously looking up to confirm that it's correct. So like it's looking up the facts that it's spitting out.
[00:26:41] Paul Roetzer: So it doesn't cite its sources and it doesn't know facts. Those are like the two biggest things. Now there's bias and there's all kinds of other concerns in it, but as a writer, if you're gonna go in today and say, okay, I'm gonna write a blog post about this topic today. Um, let's say you're a bank and about like 10 tips for people who are opening a retirement account, like Roth versus Simple IRA or something like that.
[00:27:05] Paul Roetzer: And it could write something like, that's really good. Now it may have completely made stuff up, like the facts in it could be completely wrong. Um, and, and it, it doesn't know that I'll a really practical example, I showed this on a webinar last week. I took our AI for Writer's Summit page, event page. I gave it to Open AI's playground, which is using G P T 3.5 to write, and I said, write me a press release about this event.
[00:27:34] Paul Roetzer: And I put the URL link. And it wrote like a 300 word press release that looked raped except that the city was San Francisco. It is a virtual event. It's not happening in San Francisco. It made up the dates the event was happening, , it made up who was speaking at the event and a few other facts. I was like, I have no idea where it would even come up with this stuff.
[00:27:55] Paul Roetzer: That's not even on our site anywhere. So it completely looked like a legit press release, but it. Completely wrong . And I think that's, those are the limitations. It's like, it's not, it doesn't know facts. It doesn't necessarily know where its source data came from. And so the more, um, like regulated your industry or the more important it is to be factual, you, you have to know that it's not good at those things right now to use 'em properly.
[00:28:25] Mike Kaput: Yeah. And essentially the limitations here, just naturally. Bring up the concerns, right, using the technology. Because if you are getting non-factual outputs, that sound really good. You need to know that you need to check the machine's work before you publish anything,
[00:28:42] Paul Roetzer: right? So you still need to be, and that this is where like could go back to the question, what happens to writers and editors?
[00:28:48] Paul Roetzer: I don't know because. You become the domain expert or you're like knowledgeable about the case. Like if you go back to our agency as an example, we created a lot of content for a lot of complicated industries, robotics, engineering, um, chemicals, like stuff that we knew nothing about. And to be able to vet that work and turn it over to the client for review, you had to go learn the topic.
[00:29:10] Paul Roetzer: Mm-hmm. . So you do all kinds of research to understand the topic. So that like, even for me, like let's say. I handed off a white paper on some chemical engineering process to an associate at the agency, and that associate wrote a piece back and I'm like, sounds really good. If I knew nothing about chemical engineering, I could completely miss things in that article that were wrong.
[00:29:33] Paul Roetzer: Yeah. Then we turned over the client and the client's like, where did you get this information? This is like, you don't even understand how our business works. So part of being a writer is understanding your topic. You do research, you do interviews. You learn the topic to write about it. And so even if you rely on an AI writing tool to write the draft, you still have to have the knowledge to determine if that draft is actually any good or not.
[00:29:58] Paul Roetzer: It can look good and it can sound good, but it might be completely wrong. So I don't, I don't know actually what happens in that case because we become experts on the topic by researching and writing about the topic. Hmm. If the AI just does the writing, how do we ever know if it's. I don't, and I don't know the answer to that.
[00:30:17] Paul Roetzer: I'm just kind of thinking out loud here.
[00:30:20] Mike Kaput: This is just one more, one of the many reasons that a writer's summit is needed. Right. Seriously. I mean it's, it's, there's so many different perspectives. No one, including us has anything resembling total objective truth or answers. Here we, you know, everyone is figuring this out on their own.
[00:30:36] Mike Kaput: Some poorly,
[00:30:37] Paul Roetzer: some. Well, yeah, and we're learning from our community in the process. That's why I love like so much. Like if you follow me on LinkedIn, like, you know, with a lot of our community has really started engaging there. And you'll see sometimes I'll put things out that aren't, aren't like half baked.
[00:30:52] Paul Roetzer: Like I didn't think about it at all. But it's things that maybe historically I would've sat on for a few months before I put something out that I was like, perfect. . But now I'll just put stuff and be like, I don't actually know for sure, but here's what I think we should be thinking about. Mm-hmm. And then I'll get like immediate feedback from the community of like, yeah, here's what I think, or here's this.
[00:31:11] Paul Roetzer: And it's like, okay, that's perfect. Like, so a lot of what we're trying to do with the institute, with this podcast is to stimulate discussion, to like put things out there and when we aren't confident and where it's gonna go. We'll say that cuz like, I think the more people are just talking about these topics and challenging each other, the, the faster we're gonna figure this stuff.
[00:31:29] Mike Kaput: Yeah, that's so, that's so, so true. I mean, it really is, we're we're at the phase where we're figuring out still the right questions to ask. Yeah. In a lot of ways. , seriously, I think . Yeah. So I wanted to end our discussion today on this third topic related to how is AI going to change writing careers?
[00:31:51] Mike Kaput: Obviously this technology is not going to have a small impact on our jobs as writers, regardless. We're not saying, you know, writers are all getting replaced by ai, but no doubt the job is going to evolve and change and grow with the tools that are now coming out. So do you have any thoughts about initially how writing careers and jobs could change as a result of ai?
[00:32:17] Paul Roetzer: I just don't see. Writing career paths moving forward where AI isn't infused. So again, I, you know, if I, if I'm a seasoned writer working for an enterprise, like on the content team, or I'm a journalist, or I'm, uh, an editor, or I'm leading a content team, like regardless of what your role is, your use of the tools will vary.
[00:32:43] Paul Roetzer: Your dependence on the tools will vary. But they're going to be there, like they're going to be a part of it and they're going to get increasingly intelligent. Um, so the example to kind of take it to an extreme is, um, and I think we've talked about this example before, you know, I have on the institute's site maybe 200 articles that I've written.
[00:33:05] Paul Roetzer: You've probably got 600 articles you've written. So in a perfect world, rather than us using an out-of-the-box chat, g p t or writing tool, We should have that foundation and then we should fine tune it day one on our history of content. Hmm. So that it learns our style, it learns the way we prefer to phrase things.
[00:33:26] Paul Roetzer: It learns, you know, the guidelines with which we write within. Um, and then when we use it for the first time and it generates an output for us, a draft, it's doing it in our voice and style. Mm-hmm. , like it's not some general. And so that to me is very doable now, and I think it's gonna become a standard feature of AI writing tools that you can like learn on this data set and you can just tune it based on that right now, you'd need some help to do that, but I think we're gonna have not just AI writing assistance, but personal AI writing assistance.
[00:34:00] Paul Roetzer: Mm-hmm. That learns the voice of the individual. And so the giraffes are gonna get infinitely better because they're gonna be tailored to you and your. And so assuming that that is the case, I don't know how it doesn't transform careers completely. Like it's just gonna have to become an integral part of everything you do as a writer.
[00:34:24] Paul Roetzer: How do you
[00:34:24] Mike Kaput: see businesses? Changing their content strategies and needs. I mean, obviously the buzzfeeds and the C Nets of the world are pretty extreme examples and kind of in their own little corner of the media world. But you know, if I'm a brand that's creating content on the regular, What do I need to be thinking about
[00:34:45] Paul Roetzer: with these tools?
[00:34:46] Paul Roetzer: I think this is gonna be a little slower to move because the, there's so much education that's needed about what they're capable of doing and how this is gonna work. Mm-hmm. , that you can actually infuse it into your strategy. Um, you know, I, I think about how we're doing it at the institute as an example.
[00:35:03] Paul Roetzer: Like we have Market Muse as the, the platform we've historically used for content intelligence and strategy. Um, we continue to use. But I also look at what's now possible with these, um, tools and just like, you know, we've talked about before our podcast process. So we take this podcast transcript, we create a transcript using ai, then we break that down, then we use AI to do summaries of the transcript.
[00:35:28] Paul Roetzer: Then we turn those summaries into blog posts that Mike, you know, drafts and, but in with AI as an assistant. Um, so we have AI like touching like five different use cases just on the podcast. And so when you think about our content strategy of his pre previously having to write three original blog posts every week, versus doing a podcast for 45 minutes and turning that into six or 10 pieces of content, multimedia on YouTube and podcast networks and like, that's a completely different content strategy.
[00:36:01] Paul Roetzer: So because we understand what AI is capable of us CR helping us create, we look at our content strategy through an entirely different. And so until the people building the strategies that these companies and, and agencies understand what AI is capable of. Summarization, transcription, text to speech, speech to text, drafting of original content outlines.
[00:36:23] Paul Roetzer: Like once you know all the applications, then you can sit down and start thinking about, well, what does our content strategy look like? But right now, I, I just don't, I don't know that there's many people capable of actually doing. Hmm.
[00:36:38] Mike Kaput: Yeah. I think one thing I've been thinking about related to this as well is I think a lot of companies are going to have to start getting much more comfortable with having strong perspectives and points of view on their industry.
[00:36:52] Mike Kaput: Yeah. Their markets, their businesses. Because in a world where. You know, somewhat in the near future, you can write relatively generic expert content. You're going to have to differentiate in some way, and that's gonna be unique ideas, unique thinking. Um, and you know, we're like you said, quite a ways off from that, but I could very well see it becoming much more the norm to say, here is our kind of, for lack of a better word, man, manifesto or pov Yeah.
[00:37:23] Mike Kaput: On, on the.
[00:37:25] Paul Roetzer: It's a really interesting take because I actually was having this conversation a couple times this morning where people were asking me related questions, and at the time I said that I, I think there's gonna be re a revolt against AI generated content in the not too distant future. Like I think it's gonna be so.
[00:37:42] Paul Roetzer: Easy to create language and videos and images with ai and almost like how I've, I've said in the past, I feel about Twitter at the moment. Like I feel like Twitter's just one big AI generated thread from like the same 10 people , at least my homepages, all of this stuff. And it's like, I don't even wanna see it.
[00:37:58] Paul Roetzer: Like I just, I have zero interest. I don't even care if it's valuable content. I don't want to see it if I feel like it's another chat. G P T generated top 10 list. So I'm already feeling this like, no, I want to, I want Pure, I wanna read Anne Hanley's newsletter. Like I want, I wanna read Pure Writing from a great writer that like, I feel something when I read it.
[00:38:19] Paul Roetzer: Now, I'm not saying I can't simulate that feeling, but I think there's going to be a desire in the very near future to know something was human created, that it's gonna, it's almost like there's gonna be a premium. on content and ideas that you know are theirs. Mm-hmm. And so it almost does make the argument to your point, like, does podcasting actually take on a greater influence in content strategy?
[00:38:45] Paul Roetzer: Because it is like you're listening to us talking, this is us. Like AI's not writing any of this, it's not prompting me what to say. Like it's actually how we feel and think about the topic. And that becomes very valuable from a brand perspective. I think that you can, you can have those strong points of view and you.
[00:39:02] Paul Roetzer: Make human connections with people cuz they feel like they're actually hearing how you think. Like, um, one of my favorite things in learning AI has been the access we have because of the way society works today to people like Jan Lacoon and Lex Friedman, and even Elon Musk and Nolan Brown, and like all these Demis of Saabas and Jeff Dean and like all these people.
[00:39:25] Paul Roetzer: Who we read about in the books, who like we learned AI through. And I can, I can see on LinkedIn or Twitter, their thoughts in real time. Sam Altman, like Greg Brockman, they're, they're there sharing their insights and that's them like and I, that to me is so valuable to just be able to have access to people and hear what they're thinking and feeling and how they see the future.
[00:39:48] Paul Roetzer: So I. Wonder like as a content strategist, if you don't find the ways not to intelligently automate more content, just pump out more SEO content in the world. But how do you, like we've talked about before, how do you become more human? Yeah. As a brand. And maybe it is more stuff like this where you have podcasts and live events, you know, like that you people are gonna crave human content and interaction.
[00:40:14] Paul Roetzer: I think probably more than,
[00:40:15] Paul Roetzer: ever
[00:40:17] Mike Kaput: That's a really good point and that definitely kind of preemptively answered my question about kind of what, what stuff will be the purview of humans in this age of ai. So I want to actually end our chat today on a different question. We've talked about the need for writers to get started experimenting with these tools, so like, Table stakes.
[00:40:40] Mike Kaput: Mm-hmm. , what else do writers need to be doing starting today? To evolve their skills, their processes, their workloads, their careers.
[00:40:51] Paul Roetzer: To me, it's very helpful to have a little bit deeper knowledge of how this stuff actually works. You know, what a language model is, how it makes its predictions about the next words in a sequence, like the more you understand.
[00:41:06] Paul Roetzer: The fundamentals of AI beyond just, I can put a prompt in and it gives me an output the deeper. Comprehension you'll have about where this is all going. And the quicker when new innovations arise, like G P T four or whatever comes next Lambda from Google or Bloom or what, whatever it is, whatever the next major innovation is, you'll understand that innovation way better.
[00:41:32] Paul Roetzer: If you understand the foundation of how this stuff is built and how it works and what it's intended for. and so hopefully this podcast helps people along that journey. You know, we do try and go beyond the surface of fancy tools and magic little AI capabilities to try and give some depth to what's going on in the space and how the things work.
[00:41:55] Paul Roetzer: But I think that that's very valuable for people to take an interest in it, to be curious about ai, cuz it's gonna have such an impact on your life and your career that I. Taking that interest is, will be very valuable to people down the road.
[00:42:10] Mike Kaput: Absolutely. Um, and you know, just as another quick reminder, honestly, I know we're biased, but our ai ai writers summit.com, the, the AI for Writers Summit is really, It goes beyond the headlines.
[00:42:26] Mike Kaput: We're going to dive deep into exactly how these topics affect writers and content creators, like the ones listening. So if you can make the time, you know, you got nothing to lose, it's a free ticket. Highly recommend signing up for
[00:42:39] Paul Roetzer: that. Yeah, and the, just a kind of closing thought, when I posted something, I think it was last week on LinkedIn, I had said that, you know, I long believed that writing is the most important skillset of.
[00:42:51] Paul Roetzer: In every industry, in every business, in every function, because it's everywhere. Whether you're in sales or service or marketing or executive leadership, you, you have to influence people with words. And so the ability to understand how these tools can help you, uh, influence and affect change and, um, you know, drive impact.
[00:43:14] Paul Roetzer: Drive performance in your company. I just really think it's critical that people take the time to understand how they can work these tools into their own processes and own areas of the business. I know we have a lot of people listening to the podcast who probably aren't marketers. Hmm. Um, maybe not even, don't even think of themselves as writers.
[00:43:31] Paul Roetzer: But when you step back and think about how much you need to affect people with your words through again, just simple emails, um, social media posts, things like that, we, we all write all the time. And so yeah, I think it's just a critical topic. I'm glad we, you know, dedicated an episode just to this topic.
[00:43:49] Paul Roetzer: Like there's so much going on. It's hard to just focus on a single thing for a whole episode today. But you know, I think it's such a wide reaching topic. It's an important thing to cover. So we appreciate everybody indulging us on this topic, and hopefully it's been really valuable for you.
[00:44:04] Mike Kaput: Awesome, Paul.
[00:44:05] Mike Kaput: Well, as always, thank you for your time and your thoughts. I think everyone gets a lot of value out of it, and until next week, um, we will be back with, uh, three more hot topics in ai.
[00:44:16] Paul Roetzer: All right. Thanks Mike. Thanks everyone for listening.
[00:44:19] Paul Roetzer: Thanks for listening to the Marketing AI Show. If you like what you heard, you can subscribe on your favorite podcast app, and if you're ready to continue your learning, head over to marketing ai institute.com. Be sure to subscribe to our weekly newsletter, check out our free monthly webinars, and explore dozens of online courses and professional certifications.
[00:44:40] Paul Roetzer: Until next time, stay curious and explore ai.