Stories and Strategies Podcast
Guest: Robert Lund, Chief Creative Officer and Co-Founder Studio TBD
Published April 30, 2023
Doug Downs (00:05):
Recently I was invited to speak at a nonprofit event about public relations. During one of the breaks, I got caught up in conversation with Judy. After exchanging pleasantries, we began talking about what really motivates us. She shared with me what had always been her first love. Even before high school,
My passion actually started long before high school. My grandmother was a pastry chef and I was making cakes and bread when I was old enough to push the stool up to the counter. So what did you become? I became a baker and cake decorator and loved it, absolutely loved it. It was 100% my passion. Yes,
Doug Downs (00:51):
Judy was one of the lucky people who knew at an early age what she wanted to do and was able to do it and make a decent living from it. She made that living for 10 years and was pretty happy
Professionally from about 1986 to 1996 was professional. Prior to that, I mean, like I said, since I was like five years old and what happened, a large company came in and basically took over all of the big box stores by making everything either or frozen. And you no longer needed to have the passion or the baking science to be a good journeyman baker because anyone could take something out of a box frozen and put it on a pan so they could bring people in green that had no journeyman knowledge to just bake off at whatever they needed. Unfortunately, these products are exactly what you would expect them to be with and frozen. They were completely flavorless and standardized and cardboardy
Doug Downs (02:17):
This new approach. However, flavourless proved a business success. Journeyman Bakers like Judy were pushed out of business. Success wasn’t about quality, it was about price and process.
I’ve always referred to the baking as the second oldest profession was something that I thought, well, I would have job security in for the rest of my life. I mean that there was always the saying that you’ll always need bread, right? But it basically destroyed the industry.
Doug Downs (02:51):
Judy eventually realized she had to give up her dream career and find something else. She now works in the nonprofit sector tapping into another part of her personality involving being a humanitarian. And she loves the work she does today, having learned to adapt, but there’s still a part of her that misses what it feels she was almost meant to do.
That’s always been a bit of a sorrow, been in my soul, just not being able to live in my passion, finding other things that I can do well and enjoy, but not being able to live in what I believe is my purpose and my passion has always been a sorrow in my soul.
Doug Downs (03:35):
In the public relations and marketing fields, there’s a different pre-mix or frozen solution that’s becoming more daunting, artificial intelligence. And while AI is obviously already here, the speed at which new tools are being developed and implemented is remarkable. A new study available through the Chartered Institute of Public Relations examines what’s really happening, where our industry might be going. Whatever your current role, it’s going to change and it won’t be a piece of cake. My name is Doug Downs my guest this week is Stephen Waddington joining us today from the London area of the UK. Hello, Stephen.
Stephen Waddington (04:26):
Hey Doug. How are you doing? Thanks for having me. Although many of my friends and colleagues have been interviewed you by you, this is my debut.
Doug Downs (04:36):
Excellent. I know Anne Gregory for sure. She’s been on twice. Who else that you can think of? Jean Valin, maybe?
Stephen Waddington (04:43):
Jean Valin. Have you had Jim McNamara on here?
Doug Downs (04:46):
Not yet, but that’s a good tip. That’s a good tip.
Stephen Waddington (04:48):
You need him.
Doug Downs (04:49):
So London is big and wide. What part of the London area are you checking in with us today from?
Stephen Waddington (04:57):
So I’m actually on a houseboat on the river.
Doug Downs (05:02):
On the Thames?
Stephen Waddington (05:02):
About 10 years ago, my wife and I bought a houseboat that we converted and it became our office and the place where we stay in London when we’re working in London, and actually our family house is about 250 miles north up in the north of England place called Newcastle.
Doug Downs (05:21):
Of course, yes. Make some great brown ale up there in Newcastle, but I’m envious houseboat on the Thames. That’s fantastic. Stephen. The name of the report is Artificial Intelligence Tools and the Impact on Public Relations Practice generally. Yeah, there’s the idea that AI is not new. It’s been creeping into our work day by day, but as our friend and Gregory puts it, maybe as PR practitioners, we’ve kind of been sleepwalking a little bit. Do you agree?
Stephen Waddington (05:53):
Yeah, absolutely agree, Doug. So our only awareness of it really has come I think in the last 10 years where it’s become an algorithm, it’s become a reputational issue, and there’s maybe destroyed value within an organization. And then as PR, the consciousness of the public relations practitioner is an area of risk. I think the notion that we would apply it to our own area of practice is something that hasn’t really been considered up until very, very recently.
Doug Downs (06:28):
You examined the dataset provided by Chiefmartec. I hope I’m saying that correctly. Looking at how many AI tools there currently are for public relations and marketing. You found a lot of them. This is not new. Yeah,
Stephen Waddington (06:44):
We did. So Scott Brinkler, he’s actually a VP of HubSpot now. He’s been running this database for 15 years, chiefmartec database, and it’s an open source database. You can any vendor, anyone, any practitioner working in the marketing PR Comms, broadly media space, can submit tools. And over that time he’s gathered 10,000 a database of 10,000 tools from around the world. He very kindly opened that up to us and gave us access to it. And one of the things he’s done, he’s done really good job tagging and creating a sort of topology of an index for it. And so Andrew Smith and myself in writing this report the middle of last year, went through this database and looked at the tools that could a sensibly be applied to public relations and many already are media databases, monitoring, social media, content management, and that’s that sort of thing. Yeah, we found 6,000 now we dug into those 6,000 and discovered this is probably about 2% of them actually applied any degree of machine learning and AI. And we did that where we got until two, we were ready to publish that in October, and then G P T happened and G P T chat was launched and we pulled it, pulled the report, and then followed what’s happened over the last six months.
Doug Downs (08:17):
So some of these other tools that we would be familiar with, anything like a, a CRM tool, I guess HubSpot, all of these different tools that we do play with. I use Basecamp. Yeah, tools like this. All examples of the existing,
Stephen Waddington (08:34):
Yeah, there’s sort of two classes there. There’s tools that have been built within the public relations industry, so media, databases, press release, distribution and so forth. I guess the oldest examples of technology applied to public relations. And then you got whole class of monitoring and evaluation tools as well. That is a very well established market, quite mature market. And then you’ve got, I guess the, what’s developed over the last 10 years is this subset outside the market of public relations that we then brought into practice. And this is relates to everything really from tools that we might apply to our own workflow, like time sheet management and project management through to more sophisticated monitoring technologies
Doug Downs (09:28):
Chat G P T three, I guess I must have missed chat. G P T one and two. I’m were there were, sorry, I’m I’m naive. Were there renditions of chat G P T?
Stephen Waddington (09:40):
Yeah. What happened with chat G P T three was open AI made it available via a command prompt in a web browser and suddenly anybody with access to the internet, it was one of those moments, anyone with access into the internet could access this tool that hadn’t been possible before. And so it was completely open and you saw this incredible growth within a week, a million users within by January billion users. And so it was the accessibility that really made the big, big, big difference.
Doug Downs (10:23):
So G PT three comes out and the date is stamped in PR and marketing lore now November 30th, 2022. Yeah. Really steals all the headlines. Lots more tools existing I sleepwalked if that’s the right term, through G P T one and two. I honestly did, I had not heard of it. What are these other tools and why was G P T three this turning point for AI and how is that going to impact PR in marketing?
Stephen Waddington (10:57):
So unpack that if we unpack that question. So the first point, why was it a turning point? Because of the simplicity of being able to access and use the tool also open ai, the company behind the GPT three data set made that available publicly via the web and as an api. So the moment that happened, two things happened. You and I were able to go onto the open AI website and use chat G P T three and also other tools like playground that it made available at very low cost. And then also they made the API available, which are the vendors were then able to incorporate in their existing tools. So you’ve got tools that we’ve been using for a while, like Grammarly that suddenly have AI functionality brought in because they brought the G P T three data set into, we had the announcement last week from Microsoft that it’s incorporating this GPT three API into its office suite of products. Microsoft’s big investor in an open ai. So that, that’s the change.
There’s been sort of this first wave, as you said, that date stamped in our mind when GPT three initially looked was launched by OpenAI, but there’s been a second wave since then created by G P T four. Yes. Which was launched the week before last. So where are we? Early March, it was launched GPT three based on a data set of 17 gigabytes roughly. So that’s books, websites, Wikipedia, GPT four, much larger dataset 400, sorry, 45 gigabytes. So there’s, there’s already been that tripling of the data set and G P T five, you won’t be surprised as in development and being trained in as open as it will be released in the middle of 2024. We talk specifically about, okay, where are these applications going in public relations? There’s sort of four areas that I’ve seen. There’s the open, the browser point at chat G P T and use it to create texts and images and idea generation.
Okay, for creating the first draft of something, but frequently makes some mistakes and will hallucinate. Jean Valin used that word And I think it’s a very good word that it will start making stuff up and you’ve got to be really very careful of that. Second category, I think slightly more useful is editing and summarization. So if you’ve got a large chunk of text and you think about research papers, if you think about you’ve written a 5,000 word white paper, maybe you can use editing and summarization tools, then to cut that down to size to make a 400 word version, to make a PowerPoint, to make a short report or transcript, you know, could use it as the basis of a press relation, then you could use things like chatbots, create that into a database itself, which then can be interrogated using a chatbot. That’s kind of novel and interesting there because the original data set is your own data set.
It’s a piece of content that you created. It’s not going to make mistakes. And the other categories that we’re only really beginning to explore the use for evaluation, there’s a lot of interest in the measurement and evaluation community of using Chat GPT to query earned content, well any form of content and ask questions of it. And then finally, planning and decision making. So you’ve got a data set you use, not specifically Chat GPT, but other artificial intelligence tools to spot trends and patterns in that data set that then can help you with planning decision making. And ultimately I think that’s potentially the really interesting area because you’ve got the opportunity then to actually create new knowledge.
Doug Downs (15:16):
Is there to you, do you think there is a new niche discipline within the practice that’s emerging, whereas there are some who have social media expertise? Right now we’re seeing experts that have PR and marketing skills that niche to Web three point, which is simply attached to blockchain. Is there a niche skill where PR practitioners have, maybe it’s a computer science skill or maybe it’s simply experience with the interfaces that will allow us to use these tools?
Stephen Waddington (15:54):
Absolutely. So I’ve liken this to, well, I started my career in public relations in the nineties. I think we’re of a similar vintage, right?
Doug Downs (16:07):
Stephen Waddington (16:07):
Without being rude, but there was that great explosion in the nineties where we discovered the internet and email and web and that created this whole opportunity for people who understood how to start to build websites and build communities around the web. And then that happened again in the early part of the 21st century. So 2005 to 2010 when there was that huge explosion around social forms of media and this new emergent area of practice developed and course social media and it’s still very point now exactly the same. I think it’s going to happen here in there are already practitioners who are leaning into these tools and exploring them with anger in how they might use be used in practice. And I think that creates, for those practitioners that are prepared to lean into that same level of opportunity, it’s why I’m so excited by this because want to future proof the rest of my career to make sure that are stay relevant and current.
Yeah, it’s the other point related to computer say interesting. You say computer science there we were very quick to abstract or assign human character to machines. Actually, it’s really important to remember that these are machines and we are instructing them and programming and coding them. And so yes, you have seen already this conversation about the skills required to engage the machine as a stakeholder and code IT programming, get the best performance out of it through the command line prompts. So I think he, New York Times World Economic Forum have said the job of 2023 is a prompt engineer.
Doug Downs (18:15):
I see a whole slew of email invitations to masterclass on how to harness these tools in the future, but
Stephen Waddington (18:24):
Probably sent by AI machine learning. That’s right.
Doug Downs (18:26):
Yeah. Catch my eye. Yeah, there was a phrase in your report, and I don’t think AI could write this. The phrase was, A word is known by the company it keeps,
Stephen Waddington (18:39):
Doug Downs (18:40):
I literally had to put my finger on the page and put the report down for 10 minutes while I wrap my head around that “a word is known by the company it keeps.” What does that mean?
Stephen Waddington (18:55):
So the think about how artificial intelligence works and how machines work. They work by understanding patterns, understanding words. And so we’ve talked about the huge volumes of data. 17 gigabytes for chat, G P T 3 45 gigabytes for G P T, four, huge data sets of books, websites and Wikipedia that have been crawled. And so what the machines have done in crawling those, all that content is understand how one word fits within the context of another word fits within a phrase and build up on ontologies of words related to topics, subject and so forth. And then when you make a query of the tools, we are using chat GPT three as an example. But of course we are, because it’s so accessible, all you are doing is asking it to all it is doing, sorry, is providing what it thinks is the most likely output based on the words phrases that you’ve used.
So very good example. I used very early on testing the veracity and accuracy of these tools. And one great theory is that they can create huge volumes of information, but is it accurate, actually accurate? So I asked it to write a biography of myself for a speaking event and it got the first paragraph absolutely spot on. But no surprise, it almost certainly pulled from Wikipedia, the second paragraph and the third paragraph, it went off completely poor Sean Valin did it and it wrote his obituary so they very quickly can go off and get it completely wrong because they’re based on understanding the structure of words within a topic.
Doug Downs (21:10):
Stephen, thank you. It took a bit, it took a while to get you on the podcast, but I really appreciate this and I’d love to have you back. Thanks for your time today.
Stephen Waddington (21:19):
I do, you know what, we should do this in another three months to see what we’ve got for
Doug Downs (21:23):
GPT five or, well, I guess that’s
Stephen Waddington (21:25):
Too further out. Well, five will be the middle of next year. Yeah. But honestly, the tool project we did with the C I P R was an education in itself because we got thought we delivered a report and the innovation in the last six months has just been incredible.
Doug Downs (21:46):
If you’d like to send a message to my guest, Stephen Waddington, you can email him. The address is in the show notes and do click on the report. There’s a link to that. It is a fascinating, not an overly long read, but absolutely fascinating stories and strategies as a co-production of JG Communications and Stories and Strategies podcasts. If you like this episode, do us a favor, share it with one friend. Thanks for listening.