Deborah Carver - Artificial Intelligence Through the Eyes of a Content Technologist

The conversation this week is with Deborah Carver. Deborah is the writer and publisher of The Content Technologist. When she's not compiling the weekly newsletter, she's a digital strategy consultant at the intersection of technology and content. She truly believes in helping content professionals navigate to technical ins and outs of algorithms, automation, and analytics so they can free up more brain space for creativity. I love this mission. And we all need time for more creativity. Since 2011. She's developed award-winning content strategies, guided website redesigns, and ensured algorithmic visibility for Fortune 500 companies, nonprofits, b2b SaaS, startups, and everything in between. 

If you are interested in learning about how AI is being applied across multiple industries, be sure to join us at a future AppliedAI Monthly meetup and help support us so we can make future Emerging Technologies North non-profit events!

• Emerging Technologies North
• AppliedAI Meetup

Resources and Topics Mentioned in this Episode

• The Content Technologist
• DALL·E 2
• Stable Diffusion
• Blade Runner
• Amelia Winger-Bearskin on Conversations on Applied AI
• AI-Powered Wireframing
• Jasper - AI Copywriter
• Copy.ai

Enjoy!

Your host,
Justin Grammens

Transcript

Deborah Carver  0:00  

I think that if your job is doing mediocre content really fast your jobs at risk, but if you're actually a creative person and know how to sound not like a computer, I think you'll be just fine. That said, I do think that you need to learn how to use the tools and use them as part of your toolset and integrate them into your toolset and learn how to craft proverbs and learn how to edit for AI. I actually think that editing and the role and craft of editing is definitely going to be become more important than just straight up being a writer.

AI Announcer  0:33  

Welcome to the conversations on Applied AI podcast where Justin Grammens and the team at emerging technologies North talk with experts in the fields of artificial intelligence and deep learning. In each episode, we cut through the hype and dive into how these technologies are being applied to real world problems today. We hope that you find this episode educational and applicable to your industry and connect with us to learn more about our organization at applied ai.mn. Enjoy.

Justin Grammens  1:04  

Welcome everyone to the conversations on applied AI Podcast. Today on the program we have Deborah Carver, Deborah is the writer and publisher of the content technologist. When she's not compiling the weekly newsletter, she's a digital strategy consultant at the intersection of technology and content. She truly believes helping content professionals navigate to technical ins and outs of algorithms, automations and analytics so they can free up more brain space for creativity. I love this mission. And we all need time for more creativity. Since 2011. She's developed award winning content strategies, guided website redesigns, and ensured algorithmic visibility for Fortune 500 companies, nonprofits, b2b SaaS, startups and everything in between. And this whole new world of AI generating content is one of the hottest spaces these days. So I'm thrilled to have someone who works deeply in content creation and the program to give her perspective, it's going to be a fun conversation. So thank you, Deborah, for being on the program today.

Deborah Carver  1:54  

Well, thanks very much for having me. I'm really excited about this.

Justin Grammens  1:58  

Excellent. Well, looking at your website, I know you have done a number of different jobs that kind of led to where you are today, maybe you could give our listeners a quick overview of some of the things that you've done in your career and how you got to where you are today.

Deborah Carver  2:09  

So I actually started my career in book publishing and the early 2000s, I had a couple of internships at New York publishing companies. And I worked in actually like a small agency that basically functions as a stock image agency, like kind of like they were a competitor of Geddys, but much, much smaller. And I help them with their transition to a digital search based database versus mailing out little slides of different images and filing them back, that's my job started, as mailing out these different little slides based on historical images. And then by the end of that job, I was keywording in and putting these images in the database. So then the researchers could just self serve send those digitally. So that's kind of how things started. For me. I also had a number of editorial book publishing jobs again, and a bunch of small jobs in between, because I don't know how much you know about the editorial field is that jobs can be hard to come by, and you're doing a lot of freelancing, especially in your early years. So yeah, I worked at a restaurant I worked at some restaurants, I worked at a place that, you know, sold gas lamps and dog art. I did a bunch of other small odd jobs for years. But mostly went back to graduate school in 2007, at the University of Minnesota, which is how I got here. And I studied mass communication. Now, my masters at the J school at the University, and yeah, have been, you know, graduated into a recession, where it says restaurants again, and since then I have taken some a lot of communications jobs and got more into kind of corporate communications and b2b communications. 2011 is kind of when I think of this chunk of my career as starting. And I took a job as an editor and content manager at a b2b company that publish content for caterers. And I was in charge of the print magazine, the website, the burgeoning social media accounts. And from that I was making all this content I was making so many content, pieces of content for so many formats. And I wanted to know who was looking at it what was working, I was doing so much I had no idea who was reading or like I knew some people were looking at some things but not others. And so I actually kind of switched career tracks to something that's more to performance marketing, what eventually became performance marketing, and SEO. I worked at the company they're called not called collective measures. They're called me and a while back in the day, you When I was there, I learned the ins and outs of digital marketing, and especially search optimization. Also content analytics, I had some really great coworkers who were fantastic teachers and taught me how to use Google Analytics. And I still use that to this day. And from there, I moved to back to a publishing company, although on the digital strategy side and then have been out on my own as a freelancer freelance consultant for the past four years, almost, it'll be four years in May. So have been up and down. But definitely a very different background than I think most of the guests on this podcast, I was an English major, you know, yeah, my master's in communications, I definitely come at it from the language side versus the engineering side. But also, I've been making websites since I was 15. Again, I wanted to publish my poetry on the internet in the late 90s. And so I made websites that would appeal to other teenage girls, there is a big community of teenagers on the internet at that time. And that kind of pre social media also got me very interested. So now, my job kind of combines all of that like the get the performance marketing, finding out what's working, what's not the content part writing, I write and edit a weekly newsletter, and know also how to leverage that to get consulting clients. I typically help by typical consulting clients or like big content focused websites who are either undergoing a redesign, but I do a lot kind of at the intersection of web content, email content, I do less on social media now, and more about processes operations and kind of higher level content strategy, and how that plays into what engineering teams are doing what what IT teams are doing. And we're at this really interesting Nexus now where that's that line is becoming way less separated. Now, especially with the popularization of kind of the AI Generated Content software, there's, it's definitely more in focus, when I was talking about AI generated content on my blog, like on my website, newsletter three years ago, and no one wanted to look at that content. Like, I stopped talking about it because it was the least popular content that I was publishing. And now it's, it's like, oh, yeah, he's like this. Oh, well, I can talk about that, too. So there's a lot going on.

Justin Grammens  7:35  

funny all the time. She's, ya know, we'll we'll definitely get into all of that stuff with regards to Asia, AI and generative tools, some of the things that you're seeing going on. But first of all, congrats on being out on your own for that for the past four years or so, being another sort of fellow small business owner, I know that it's not an easy task to do. It's a scary thing to leave. Some people might call a little bit more of a comfortable place, I guess, working for somebody else to kind of go out on your own. But congratulations to sort of follow your passion and do what you wanted to do. I'm sure it's been an up and down for sure. For every everyone's got a story. It's interesting times. Sure, sure. And my wife graduated from University Minnesota with a journalism degree and she was worked at the Minnesota daily. So I'm not sure if that was even a thing when you were when you were there when you're going through your program at all. But

Deborah Carver  8:17  

oh, yeah, that's yeah, the daily was definitely a thing. I was in the grad program. So I actually taught I taught editing for a semester and ITA, the editing class while I was there, and then yeah, when I'm teaching it. And so I actually, when I was teaching editing at the U, it was very print based. And this was like 2008 2009. And I kind of insisted that we do a blog like that, that the students write a blog entry, just so they were, you know, learn how to use a content management system, because I was like, You're gonna have to know how to do this, it will help you in your career. So yeah, it was definitely was definitely slower going what was being taught in journalism school, you know, 15 years ago, then now, I think there's much more of that digital skill set taught now. So

Justin Grammens  9:11  

yeah, yeah, well, good. It's interesting, just to sort of see how AI is sort of changing the landscape of what people are talking about these days. And you mentioned about, you know, not being a technologist or I like to use the word geek, I guess, in some ways, coming at it more from you know, the content, the creative side. And that's really what I love about this program, this podcast that I'm running, it's really conversations on the applications of artificial intelligence. And so whether or not we talk about you know, all the algorithms and all the all the other stuff that gets that goes down to that level, I don't really care I'm really actually interested more in in how people and how it's being applied in the world today. And you said, you wrote about some of this content, you know, years ago and people people weren't really into it. Now. They are what what do you think has shifted?

Deborah Carver  9:54  

I mean, the biggest thing has been the announcement around at open AI and all the related announcements and tools and products that they've launched in the past three years, like developing GPT. Three, until that point, it was, or like, the thing with open AI is, like they're open, they're not open source, but they, you know, distribute their tech, so that different companies can use it. Before that it was definitely more of a, you know, Google and IBM are working in their own little, you know, corporate and corporate silos on these projects. And as a digital strategist at an agency, I'd get to see those previews. And I thought they were really cool. But when I talk about them to other people, they'd be like, what, I don't know why I'd want to use that, or repeal, like, that's really expensive, because it was. So the barrier to entry has lowered significantly, and especially this year with, with Dolly to stable diffusion and the image AI tools, and then also chat GPT with text generation. And it's interesting to me, especially on the tech side, text side, because the technology underneath is not particularly new, or it's just being trained on more and more data. And it's faster, it's so much faster. Also, people can see actual use cases for it now, versus just saying, Well, I have to spend as much time typing in my prompt as I do getting the results. So why would I use? Why would I use this for any practical application? So yeah, I think we can all see the practical applications more clearly now than we could a few years ago.

Justin Grammens  11:39  

Totally well said. Yeah, I mean, I remember back in 2015, you know, listening to the startup company, you know, they were startup at the time, but they had been funded, and what their whole deal was, was they would listen to, and this was specifically and this is one of the things that I think has changed, it was it was so narrow before, so they would listen to basically earnings calls that were going on, and then it would write a summarization of it. And they had the one that was summarized by the human and then the one that was summarized by the AI and the AI was pretty dang good. In fact, good enough that you probably wouldn't hire an editor to sit there and listen to earnings calls and basically put their stuff in because the AI could do it. And at the end of the art at the end, this was the thing on NPR or something like that, or where they basically said, oh, yeah, and by the way, we can have 10,000 of these things running at the same time. You know what I mean? It was like, oh, yeah, you're right, we scale. You know, in the human world, we don't scale. So people were talking about that. And I was like, this is gonna change. And this is one of these things where sometimes you know, you're working on a technology for so long, you're in the space for so long. It's kind of like, no one else is listening to you. And are you to kind of an echo chamber, you're yelling about this thing. But I think AI I agree with you. I think there's two things that happen. Number one is it became more generalized right now you can prompt on anything, you can literally, like last Ascott anything, whereas the solutions before were very much like, Oh, you want to get a you know, basically summarization of a stock call. Okay, here you go. And then I think the other thing is, is it's just open to the masses. Oh, it's a it's more generalized and be anyone can go to the website, if, if it's up, you know, that's the other problem that we're having right now. But they can just start playing with it. I mean, it's one of those things where it's like, once anybody from you know, your daughter to your grandma can start typing things into this, you're going to see a lot more interest in whatever the technology is.

Deborah Carver  13:20  

Yes, yeah. It's not this obscure concept anymore. It's actually something and that's, I think, would chat GPT, at least compared to the other tools that have been around for a few years, like chat. GPT was like, Oh, well, we're not a business writing tool. It's just, it's a chat that like, talk to him. And that really intrigued a lot of people. And it could handle that many users and put out, spit out sentences that fast, like, extremely impressive the speed at which it goes. Yeah, I also remember seeing a demo from some folks at the Washington Post, it would have been around 2016 2017, about how they are using AI for local elections. And just to shape basic stories around local elections and to sentences. And also like local high school sports, or like low level sports that people like to read scores about, but are like they like to read the scores, or they'd like to see what happened in the game. But you don't actually want to, you know, spend a journalist going to cover that. Although the all that content was always then edited and fact checked and everything by actual people after it was done. But yeah, that was that was years ago. And so it's really interesting to see all the stories now about like how different news organizations are using this technology. And I'm like, Yeah, I saw that from the post. Like a while ago. This isn't particularly new. It's just definitely more widespread.

Justin Grammens  14:49  

Well, one thing that I like to ask is, how do you define artificial intelligence? Thought about that? 

Deborah Carver  14:55  

So I'm one of those people who tends to because I had writing this for a while, I tend to go back to naturally or machine generated content or natural language generation, because those are how I was kind of taught to talk about it so that we weren't overselling its intelligence, I guess. Yeah. And I still think about it that way, you can get really Blade Runner really fast, like the nature of what human intelligence is. And I'm a very practical person and don't actually love going down that route. But I just think, as we think about AI right now, it's in these, I would say, like, I think about it as any rapid processing beyond human brain power of, you know, big data sets, and coming out with understandable outputs, I guess. That's kind of how I think about it. I don't really sit down and think about philosophy of intelligence or like humanity at all. I'm just thinking about how we're using it right now. And to be honest, how we'll be using it pretty much at least through my lifetime. I don't really see, I watch a lot of sci fi, I don't see a lot of things happening in sci fi happening, you know, Skynet, it's not actually really happening right now, as far as, as far as I know, maybe, you know, some other people know some other things. But yeah,

Justin Grammens  16:24  

sure, sure. No, no, I'm with you. 100%. I mean, you can go down a rabbit hole talking about if things are sentient or not, if they express feelings, and all that sort of stuff. But the end of the day, I think what we've shown here is is you know, you can feed a machine, a lot of data, and it can create some pretty amazing things. And when it comes to the current state of AI generated content, we've talked a little bit about where you think it at. But if you wanted to add a little bit more to that you certainly could, and then maybe like, what do you see in the next, you know, six to 12? months to couple of years ahead?

Deborah Carver  16:54  

Yes. So the technology that we're using, basically identifies and duplicates patterns in language. So from, you know, from your text, autocomplete, or Google Autocomplete to what's happening, and Chachi Beatty, it's just looking at a huge data set and predicting Well, this is the word that's most likely to come next, or this is the sentence structure that's most likely to come next, you know, putting a lot of that data together as an into coherent sentences. what it can't do right now is actually, you know, come up with original ideas on its own or do something that is radically new. There's this woman, Amelia Winger-Bearskin, who was saying, the podcast. Oh, we did? Oh, I should, yes, she was talking about the generative image API. And basically, if you ask it to do something that it hasn't seen in bulk, or like, hasn't seen examples of in bulk, it just doesn't know what to do. So it's not coming. You know, it's imaginative in that if you can put two things that already exist together, then you can create a prompt around that. But you can't really come up with a whole new idea. That said, you know, there's so like, what are new ideas? Again, like that's more in the nature of intelligence. But I think right now, where we're at, is that the generative, whether it's text or whether it's images, it can replicate popular patterns and common speech, and common mountain absurdly speech, common writing because it was trained on the internet. So it can replicate the text of the internet really well, in the style of writing on the internet really well. What can't necessarily do is kind of do that from a multi dimensional perspective, it really is just predicting based on the super wide data set, but predicting what comes next. And so what are the examples of that is where you kind of look at chat GPT and you is a lot of people have been like, eight wrote a poem in the style of Shakespeare wrote a sonnet and the style of Shakespeare and it has rhyming couplets at the end. And me English major says, Well, yes, technically, there are some words that maybe were in Shakespearean sonnets, and the Brian scream is there, but it's not in iambic pentameter, which is also one of the requirements to be a Shakespearean sonnet. And actually, I've had several arguments with chat GPT about trying to teach it I am big pentameter, because it can identify. But when it generates the writing, it isn't able to generate. So like in the poetry it's producing. It does not recognize that it's not doing iambic pentameter when it's actually generating the content, because actually, like meter and scansion is something you could easily teach a computer it's another you know, it's code. What is next like might be those finer dimensions, those finer points, so things don't seem quite so generic when we ask it to do so. Something in the style of something else, and how that like, so it becomes a little less replicative. And actually, you know, you know, understanding, you know, what a song in the style of Snoop Dogg actually means versus what some, like, kind of, you know, internet dad would write quickly. So I think that, yeah, in the next six months to a few years, we're gonna see some finer tuning of it. And the, at least the language will be clear. From the image side, I think, you know, at some point, it will learn to draw hands better than it does and understand those finer details, and those finer details that make up those different dimensions of style. However, I think that also requires a lot more, it requires a lot more human training and human interaction on the training data to say, Okay, this, like, do this not that for this very specific thing. And at that point, it's like, I mean, it's all collaborative. All tech is collaborative. So and, you know, I think it becomes, yeah, it's less artificial, and definitely more editorial.

Justin Grammens  21:13  

Gotcha. Yeah. And so when you're talking about getting the finer points, and because yeah, you're right, there's definitely text, I feel like that, as you can tell, it's just sort of spitting out what it read. And it's kind of like, in my world, a lot of it is like technical blog posts and stuff like that, I see people just writing out the same stuff. And it's just the same garbage over and over again, because it doesn't really know any better, right? There isn't a human feel to it. And you can tell that it's not based on much experience of somebody actually like living it. But if it starts getting to that point, when you basically are generating content, that is that you can't really distinguish, in some ways. A couple of things. I mean, do you think your job is at risk? You know, how do you think that this is going to affect our jobs? And also, you said, you taught the university and I teach at the University of St. Thomas, myself, too. I teach coding, I don't teach writing, but it is there's all this blow up going on right now. Like how, you know, students could just be able to sort of, like, fake their way through all of these things. So they are they'll, they'll be able to not really do the assignments. But, you know, the sort of two parts of that question one is, is, you know, kind of faking your way through, like, what does it look like to be an educator or to be educated? In some ways? And then also like, yeah, or do you fear for your job at all,

Deborah Carver  22:21  

for my job, though, because of my job is to position and train things in a way that actually stand out to algorithms rather than reflect them, like in search part of it is that you are using original words and the same data set that everyone else is using, and AI generated content can augment that, especially when do you want to see what generic looks like? Like, that's what AI generated content is really good for is like, Oh, this is what generic looks like. Okay, so I have to write something better than this. So the algorithm will see it. And I would say, as far as I think the types of jobs that will probably become less common would be more like low level internal communications, things that can be easily automated with, with a text generator, I would say like, my thing with the images. And so my partner's an artist, and he draws everything by hand. And his style can't be picked up by a computer super easily, because of the finer details of it. But also, like, I don't know, the people who are using image generators, instead of Illustrator, human illustrators, they weren't buying illustration anyway, those jobs aren't really at risk, because they weren't ever really they were, they haven't been there for quite a while, at least, I think with the proliferation of kind of cheap, easily generated human content through places like Upwork and Fiverr. I think that if your job is doing, doing mediocre content really fast your jobs at risk, but if you're actually a creative person, and know how to sound not like a computer, I think you will be just fine. That said, I do think that you need to learn how to use the tools and use them as part of your tool set and integrate them into your tool set, and learn how to craft prompts and learn how to edit for AI. I actually think that editing and the role and craft of editing is definitely going to become more important than just straight up being a writer. And kind of the differences between that and how that works. Like knowing how to edit AI generated content. Like I think that's the skill set of the next five to 10 years.

Justin Grammens  24:39  

I see. So use it to generate some of this some maybe an outline or some sort of initial draft and stuff like that, and then kind of pick away at it and edit around.

Deborah Carver  24:47  

So the really good AI generators content generators is where you'll feed it your data points, you'll say, Okay, I know that we get outcomes, X percent of the time. These are the three reasons we think this is is happening now put that into sentences and put that into paragraphs. So we can deliver it to our executive team, something like that, where you're putting in the original inputs, I and the like the AI teams that I work with, really discourage, trying to use a tool to just talk about something that you know nothing the an AI generated tool to do something that you know nothing about already, like, you really should have some expertise in what you're asking the tool to do, either by feeding it your data points or feeding it your proof points, or by knowing how to edit the things that are not relevant out, because it will just kind of spit out things that you'll read it and know that that's not that's not the case, that's not true. And you really have to have that kind of eagle eye out for that versus when you're editing a human, you kind of assume that they're at least trying to be truthful, even when they're not always. But I mean, that's just a like a larger problem of the internet, though, is I think we need to be a little more aware of like, Oh, hey, that's a fact. I should check it. And what might sound like BS, like probably is so sure.

Justin Grammens  26:08  

Sure. Yeah, you know, I think people can see the direct correlation, I think with writing at least, I feel like they could see that in a couple different ways. And I've used chat GPT, for a lot of different things. Maybe with regards to initial drafts, I've thrown at stuff that I've written and I said, Could you make this better? Right? Yes. And I've been impressed with the word choices that it's done, you know, some vocabulary stuff that, you know, that I'm like, well, that's, I like that. So let me add that in and not add the other stuff. And so it's been able to massage what I've given it, we talked a little bit about image generation, you know, what are some applications that you're seeing there, because everything I'm seeing is just kind of fun stuff, you know, put a monkey on a roller coaster and have it flying through space, say, Okay, fine, is generating this stuff. But I feel like I feel like in those cases, it's a little bit more out there. I don't I don't know if there's a lot of business use cases today. But again, I'm not sure what what you're seeing, or if you're seeing things that are sort of like truly monumental or truly game changing in that space. In communications,

Deborah Carver  27:07  

in an especially in web communications, I do a lot of wireframes and wireframes, being abstract looks at how a page design or an app design will be laid out. Wireframes get so much better feedback when they actually have something roughed in. Yeah, so one of the big, like, ways I've been using AI generated both text and images, is to create, you know, here's what we're thinking for this space, it will not be to create basically the replacement for lorem ipsum. Like, instead of just reimage. It's like, I was like, Oh, well, really thinking an image that looks like this bike is actually customized for your brand. That's the biggest use case. For me. It's like, oh, I can rough in something for this high level concept and not have to, you know, as long as I've pre warned the clients that like don't get too bogged down in the details, this isn't Peter generated, we're gonna make this really good. But like, high level concept Does this meet your needs, that's kind of how I've been using it. And most often, I would say, like, if you're a lightly storyboarding something, or putting together visuals for like a PowerPoint, I can see using some of those image generators. The problem with the most popular image generators is that there are copyright issues that are very much still being worked out. And I think at a professional level, it's a Use with caution, versus, you know, something like it's fine if it's, you know, an internal use PowerPoint or something like that. But when you're actually going ahead and pressing publish, like, you want to make sure you have the rights to what you're publishing. And so stable diffusion isn't, isn't necessarily the best source for that. So there's definitely some intellectual property issues on both chat and an edge that we should all be aware of, as well. It's an exciting new space, but there's, you know, there's benefits and drawbacks, and there's a lot of there's a lot of conversation to be had about some definitely about at least from the content space, like, Oh, these are things that I've been thinking about for years. Yes, intellectual property. Yeah, let's talk about that. And like, how we can safely use this content in a way that is supports our business that isn't ripping anyone off or that we're not going to get a lawsuit for later.

Justin Grammens  29:30  

Yeah, when I started this podcast, I went out to some AI generators and the beginning voice that people don't don't know or is actually an AI. You know, I just I fed it if it had some text, and it says what I told it, you know about the podcast and who I am in the background group, that sort of stuff. So that was all totally cool. But the music that I wanted to underlay it, I also tried to find some music generators and I use a bunch of different AI music generators. And I ran into the exact problem. They basically said don't do not go public with this without paying huge royalty. And the royalty is like literally, you know, per minute, you know, of audio that's played and I'm like, I'm just starting to sing. I'm not making it. I'm still not making any money on this thing. So I actually I was in a band after college we played around. So the the music for this is actually music that we generated in our band. But cool. But no, I ran into that exact same problem. And I was really worried about it. You know, at first I was like, Yeah, let's, let's use this. And then I kind of read into the details of it. And I'm like, yeah, it's a generated, but they still own it. No, it's not really Yeah, I did.

Deborah Carver  30:31  

Yeah. Checking the terms of service of the tools that you're using in understanding where their copyright is coming from. And if they're looking at it, there's companies that are more aboveboard than others, I would say as far as like, scraping lots of content and then you know, republishing those patterns and what that means and at what point it you know, stops being an artist style, and just become something new. These are intellectual property issues that have existed for a long time like before AI. So like, what exactly you know, what exactly is plagiarism? Is it plagiarism? If you just take the same sentence and replace every word with a synonym? Yes, it technically is. But sometimes in larger business settings, you don't need to, like you can just have a generic form letter. And like, that's okay. Like, it's okay to kind of self plagiarize a generic form letter versus, you know, your most recent piece of thought leadership? I would say, yeah, just generally, it's understanding what you're using the tool for? Why you're going to AI, and like what your purpose is. And then also, yeah, whether that's a internally publishable thing or something that you can share externally.

Justin Grammens  31:49  

What are you seeing with regards to you mentioned why you are going to AI? I liked that. I liked that question. I'd like to think a little bit about that. Are most people going to it because they just want to generate more content faster? Or do they? Are they trying to augment an existing process that they have? Where are you? How are you seeing organizations? Or how are you using it,

Deborah Carver  32:09  

I would say there are definitely a lot of kind of SEO focused companies that are using it to generate more content faster. In my opinion, based on how those algorithms work, that is not going to work long term, Google's going to come down hard on that. And basically, the the trend has been that you shouldn't just generate content for content sake, because it doesn't mean anything if you don't actually lead or think about the meaning of what you're putting out there. And also, you know, again, fact checking, and just kind of general writing quality. There's a lot out there, I think, some of the more promising uses, I would say, in addition to kind of the you know, roughing in something, or using it as a base, or using it to clean up something that you are looking at different ways of writing it. Those are really great uses. There's also kind of looking at, the more the parts of the technology that are just kind of duplicating patterns and duplicating, you know, essentially like clone stamping, what already exists creatively. There's a lot around that, too, with, there's some podcast tools, I think I talked to you about descript, which will basically if you give it enough audio input, we'll make a clone of your own voice. So you don't actually have to, if you fumble a word and you want to recite it, you can just give it to descript. And it will do that. I have not tried this yet, but I'm I am intrigued. And then I actually just saw something where you know, generative AI is being used to dub in movies. It's this movie called Thol, where it's about teenagers. And one of the teenagers in the original film said the F word and they needed to change it to be PG 13 appropriate so they actually AI it in a dub of her saying freaking, as you can tell on her mouth too, because you know, we used to watch dubbed, you know, dubbed curses on TV. And yeah, it does not look like they are saying what they are saying. Now they're using AI more to generate that and kind of depending on the you know, if it's just one word you don't really notice when it's a few sentences, you start to notice, it looks a little uncanny. But again, using AI for editing and for correction and for finessing. I think that there's a lot. There's a lot of potential there for sure.

Justin Grammens  34:28  

Well, we've talked about Chat GPT and I talked about the script here. Are there any other tools that you have looked at abuse that can be can be used in a space? So yeah,

Deborah Carver  34:37  

I have looked at actually a lot of the AI generative tools and they're definitely kind of varying quality. And I am a partner of and work closely with the tool writer. They actually started as a style guide tool. So for writing in an organization like this is how we talk about our brand. This is how we talk. These are the grammatical conventions we use like and it was based uglier a brand specific grammar checker. But they have evolved to have a generative AI tool. And it's, it's trained on b2b content. So it's much more applicable than something like chat GPG for the actual workplace, can understands the difference between a workplace speak and not. And I just really like it, because it gets that professional tone. It's not trained on the same data, like it's only trained on, on b2b. And then individual companies can train it only on their own content, and it doesn't use that elsewhere. I use Jasper a little bit. I liked that one a lot, I thought that one was pretty original. And there are some others, like they all come out with, you know, different levels of of results, depending on what you're looking to do what the prompts are, I find that the ones that are a little more focused on the technical language versus like, what's considered style, do a little bit better. Because sometimes, like style is a very natural human thing. It's not just adding in words that, you know, make you sound, make you sound smart, or make you sound excited. Like that's something that's very hard to mimic. And Eva, humans don't do a great job of mimicking it. I find the better tools are definitely ones that are a little more specialized, that they're looking at data and that they're refining all the time to that you see an actual improvement over time. And yeah, but it's also it depends on what you're looking for. The cool thing with chat GPT, is it like you have a conversation with it, and it remembers what happened before. And some of the tools do that. And some of them don't. And also, chat GPG just launched its pay model. But you know, there's the difference between free and paid. Yeah, a free always looks free. And you know, you get what you pay for? Yeah,

Justin Grammens  36:53  

I've used a thing called Copy. Copy data. Yes. Which is interesting as well.

Deborah Carver  36:58  

I liked that one. That one was also like, there are some of them that sound pretty similar, no matter which, you know, what you put in copyright AI, kind of what I did a review of it for the content technologist newsletter, I just was like, so out of left field for so different from the other ones, like when I gave it the same exact prob, which is like, what would you be concerned about for editing AI, I was just like, I can't even like kind of fit this into my narratives. And it also depends on what the tech they're using, and you know, how they're training and refining it. And there's a lot out there, I think we're gonna start seeing more, but there's also like, the previous versions of GPT. Like, we've been able to use those to generate content for a while. And there are definitely still some of those out there. And you're like, oh, this just isn't as good. Like, you can kind of try and it's a GPT too. So

Justin Grammens  37:51  

yeah, you know, now GPT. Four is the next thing that people are talking about. Now, Microsoft is And Microsoft has invested $10 billion, or something like that, if we get the amount in open AI. So yeah, they're only going to get better. And when you talk about, you know, continually retraining, I think that is going to be the differentiator, right, everyone forgets that Chat GPT is only up to 2021. So it doesn't really know what much of the more current things. And I think the leaders are going to be the ones that are continually improve and get better. You're only as good as the data, you have to continually spend the time and the effort to feed it more and more content. And I

Deborah Carver  38:23  

think also another consideration for the different tools is are are they going to use your data as more training data. And depending on your business, that might be something that you want to if you're if your business is very concerned about its intellectual property and its brand voice, you probably don't want your AI tool using what you've input to make more input. There's definitely like there's legal and IP implications. But again, if you're just you know, if you're trying to write a cold email and you're just having writer's block, like go to chat GBT and it'll help you out. And it's it's great.

Justin Grammens  38:59  

For sure. Grammarly is not to that point, but I love Grammarly. I mean, Grammarly has completely changed. I mean, I was the guy kind of sitting over in the corner working the math problems, not really the writer. So I'm not I'm not really a writer at heart, but I have to write a lot of content, write a lot of things, and especially when the emails, it's, you know, it's really, really helped me refine my my writing skills. So it's been a huge benefit to teach me the right way. In some ways.

Deborah Carver  39:23  

That's a really cool thing about these tools is that, you know, if writing isn't your strong suit, they're great assists, and they are equalizing in that way. I do like that there's all this focus on kind of the output of these tools and what what it actually means because that also means we're all reading a little more closely. And I think that that's a little that's definitely helpful as well is that we're actually paying attention to the content in the content. And not just the fact that there's something there. And you know, it's, again, I'm like, it's just weird to me that it's just like all of a sudden at the Late last year, it was just the new cool thing to talk about this. And I'm like, Oh, I've been kind of prepared for this for a while. And to the point where I'm like, Well, hey, I already said all these things years ago, but it's also it's cool that everyone's focusing on content and writing. And as a writer, there's definitely things that are frustrating. I'm like, I mean, yeah, the output of this writing isn't great. But also the speed at which it comes out is really amazing. And I couldn't do that. Or even just, you know, this is pretty good. And it just needs a little bit of an edit and a little bit of a polish. And I think that if everyone focuses a little more on editing, and not just, you know, straight up creation, like, that's a good thing, too.

Justin Grammens  40:40  

Sure. Well, cool. Cool. Well, Deborah, how do people get a hold of you? We will have liner notes and stuff like that. So I'll put links to everything. But you know, talk about your newsletter real quick, and how people can find you. For sure. Yeah.

Deborah Carver  40:52  

So it's called the content technologist. And you can just Google that and it shows up first, I'm also have a link, it's content dash technologist.com. And I'm Deborah DBO RH at content, Dash technologist.com. And yeah, shoot me an email, just visit the website, sign up for the newsletter, it's really designed for, for content professionals, whether you're an internal content strategist at like, an enterprise organization, or you're a journalist, or you're an independent creator, like there's entire worlds of people working in content for whom digital tools are a huge part of their lives. And that's kind of the goal of the newsletter is to help people understand and help content professionals understand, you know, which, number one which tools are worth your time, but also which tech like how not to be afraid of the technology and how to kind of embrace that this is how the content is distributed now. And like we didn't even talk about like, tick tock in those AI algorithms are nuts, just embracing the different ways that computers process content, and that humans process content, and that it's not one or the other, it's always both now. So that's kind of what content technologist is about. I'm also on LinkedIn, very easy to find me there. And like, marginally on Twitter. Now, I'm figuring that one out.

Justin Grammens  42:18  

thing, I think everyone is trying to figure out if it's gonna stick around for a while or not. So

Deborah Carver  42:22  

yeah, yeah. Well, great.

Justin Grammens  42:24  

Is there. Is there anything you wanted to share that you want to talk about? Maybe that we didn't, we didn't we didn't touch on today?

Deborah Carver  42:29  

Just the only thing that I'm really working on right now is that, you know, and there's not at all we touched on today, I have a course about understanding Google Analytics for so if your organization is in the process of kind of freaking out about the forest transition to the new Google Analytics, I really into educating and helping people who are not analytics, people who are writers and English majors, like me understand what they need to know about content, and then also, or about Google Analytics, and just analyze web analytics in general. But also just, you know, transferring the setup from one to the other. That's kind of the other the other part of my work that is, that is not as fun as generating new content, but analyzing old content and seeing how it's performing and seeing what you need for your website. So

Justin Grammens  43:23  

is that all on line? And is it one on one instruction? Do videos? Like how do people this video, and

Deborah Carver  43:29  

we're actually making the transition to deliver it via email, just so people don't have to login to an extra platform? I think that is a big barrier on the user experience perspective. But it's text and email. And it's designed to be completed within a month. So it's as useful to you if you've never used Google Analytics before, or if you're used to using the old analytics, and you want to find your favorite reports and the new one. So it's equally designed for both.

Justin Grammens  43:59  

Cool, cool, well, that's great. Well, we'll be sure to promote that as well. And, yeah, I want to pick your brain maybe a little bit on which platform you use to publish this out, because I've got some courses that I've been teaching in person at the university that I actually would like to put online as well. And there's so many different ways to do it out there. But pick your brain a little bit more offline on that.

Deborah Carver  44:17  

Yeah, no, I'm really I have a lot of opinions, because the platform I picked I thought was going to be a little more successful than it is. So I, we're going to email and I'm especially also the thing is content people like Arch. Exactly the first to jump on new things like they like the old familiar formats. So that's kind of where we're, where we're looking. But it's also like, yes, it's video, there is an interactive component, but if I can do this in email, YouTube and slack, then those are three tools that I don't have to teach other people that they've already learned. So

Justin Grammens  44:53  

yeah, yeah, sure. Well, great. Deborah, I appreciate the time today. This was awesome. And you know, you Being a person that sort of, like you said, been in this space for a number of years to get your perspective on where we're at today, and where we're headed, especially around generative, you know, content, we haven't really done much with regards to that topic at applied AI. So, you're the first of I think many people that will have on around this this space, this isn't going anywhere, you know, at all. I think it's a, it's a fascinating area. And I think how we use this is going to be obviously changing and evolving over time. So it's going to be exciting, exciting road ahead. But thank you again, so much for your time.

Deborah Carver  45:31  

Yes, thank you very much, Justin, have a great day. Thanks.

AI Announcer  45:35  

You've listened to another episode of the conversations on applied AI podcast. We hope you are eager to learn more about applying artificial intelligence and deep learning within your organization. You can visit us at applied ai.mn To keep up to date on our events and connect with our amazing community. Please don't hesitate to reach out to Justin at applied ai.mn If you are interested in participating in a future episode. Thank you for listening