Larissa Padden 0:05
Hello everyone, welcome to another episode of Cogcast, Cognito’s podcast where we talk to journalists and media pros on the latest happenings in journalism and PR. I’m Larissa Padden, your co-host today, here with my colleague Vanja. Vanja. Do you want to give us a little overview of what we’re speaking about today?Vanja Lakic 0:25
Thank you, Larissa. At Cognito we’re closely watching the latest AI developments and how they’re impacting our work and the work of our clients. We’re collaborating across our global offices to test out, and shortlist, new AI tools, offer AI enabled solutions to our clients – like through our media training programs. We’ve held client workshops, experimented with various CI tools, written numerous blogs on the topic, and talk to journalists on this very podcast to discover how AI is changing media.

We’ll hear today from our CEO, Tom Coombes, and our special guest, Andrew Grill on how AI is impacting the future of PR and communications, and how business leaders at agencies and inside businesses are adapting and thinking forward to what’s next.

Larissa Padden 01:22
It would be great if you could give us a little idea of what you’re focused on what you do. And we saw online that you are a technology futurist, correct me if I got that wrong, but explain a little bit about what that means.

Andrew Gril 01:35
I’m Andrew Grill, I’m the Actual Futurist and for last 30 years, I’ve been looking at the impact of technology on business and society. I run a podcast, I’ve done 500 talks, I’ve worked closely with Cognito. I love understanding technology, and then explaining how that works to senior audiences. And so it’s great to be here today on the podcast to lend my lens, my experience to how AI will disrupt, augment and accelerate the business of communication.

Larissa Padden 02:01
Great. Yeah. And Tom, we’re obviously very familiar with you. But do you want to give a little bit about yourself and your view on this space, and I would say what you do at Cognito, but you know who you are.

Tom Coombes 02:13
My name is Tom Coombes. I’m the Chief Executive and founder of Cognito. I’m ultimately responsible for making sure that our customers, all over the world, benefit from the best advice, the best technology and the best communication strategy available. I therefore have a very close interest in how technology is going to benefit, not only our business, but our clients businesses. That’s a huge task for us and one that we are dedicating more and more resources and investment to, and in an increasingly complicated and crowded world.

Larissa Padden 02:42
Tom, we’ll start with you. Can you take us through the last few months at Cognito? And how you’ve been thinking about AI as relates to the future of the company and our clients?

Tom Coombes 02:53
Yeah, thanks, Vanya. So, it’s really interesting to put what’s happened in the last few months in the broader context of technology developments, you know, the industry as a whole – and I think it’s important to look at the backdrop to that. So, we know that there are some five or 6000 tools out there. And we probably use something like 50 to 100 of them during our day-to-day work.

So of course, a lot of our work is, and always has been, about technology and how we can use that to the to the best use for our clients and give them as much impact as possible. But you know, like everyone else, the advent of chat GPT, and other tools has really brought things to the fore and highlighted the special sort of abilities of Generative AI tools.

So, we invest a lot in technology, we have done always, that’s only going to increase. I think we also need to remember that the AI has been around for a long time. But I think the advent of things like chat GP has brought expectations and abilities right to the forefront of everyone’s mind so that we have a longer-term technology strategy within the business, we have many partners that were really close to, but I think for many of our customers, the expectation now is that the technology can and should do more, and will do more, and their prioritization of technology within comms and marketing has increased.

So, it’s been very interesting and we are working in a number of ways to make sure that we are bringing our clients, frankly, as much value as possible from the tools. I would say that there is a lot of information around, there is a huge amount of noise, and in some cases, there’s a huge amount of fluff. But what is also true is that the abilities of these tools is rapidly increasing and keeping on top of what is possible and what is out there – as well as trying to make sure that that is played into our work with our clients, in a safe and responsible and effective way.

It’s not an easy task. We’re fortunate to have computer scientists in the group, people from technical backgrounds, and we’ve even got people who studied AI and communications at university. We have a working group across the organization, in all our seven cities, who are experimenting, working with clients, looking at the technologies, sharing information, and making sure that is all done in a careful, sensible and safe way. So, it’s a minefield, we’re excited about what it can do for our clients. And we’re very conscious to do things in a systematic and valuable way. That was a long answer. But there’s a lot in there.

Vanja Lakic 05:54
And on the topic of safety for clients, of course, the introduction of new and shiny tools always brings risks for clients, how are we accounting for those risks?

Tom Coombes 06:05
Well, the first and most immediate pitfall is that not everybody is aware of the risks. And by putting information into a system or into the internet, you’re releasing it into the wild, and that can be dangerous. And I think that is where we’ve seen companies ban some of the Generative AI tools in their offices. So, we’re very conscious to make sure that we have guidelines and policies that educate not only our own teams, but also our clients.

Equally, you’ve got to be able to use these tools to learn what they can do. So, we are setting up safe environments with our technology partners, where our experiments, if you like, can be run.

Vanja Lakic 06:49
What are you doing now, that’s going set the stage for the next six months at Cognito?

Tom Coombes 06:54
We’re investing in the tools that we have, the tools that we’re interested in, we are dedicating more time from our people to learn to use the tools, we are hiring people specifically to work in our API practice, and we are learning and sharing as much as we can about the tools that are available and the practices that are emerging.

We’re trying to make sure we identify the most valuable sources of information, because you sometimes you have to wade through a lot to get to the real valuable content. And I think crucially, we’re working with our partners across all the tech companies to understand what they’re doing and how they see AI improving their offerings, and making sure we can play that back to our clients.

Larissa Padden 7:42
And Andrew, I have a question for you. You know Tom, you’ve already mentioned and touched on how rapidly this technology is developing, and you called it a minefield, which certainly has been said in the press. But what, Andrew, is your impression on how AI is disrupting, marketing, PR, communications, etc?

Andrew Grill 8:00
Look, it’s fair to say that everyone’s experimenting, ChatGPT was a real watershed back in November 2022. When it came on the scene, it took some people by surprise about what it could do, it really removed the friction.

What I say to all of my clients is the reason that AI is now a hot topic is, [why] it’s in the media, [why] my 80 and 87 year old parents in Australia are talking about it, is because it’s the first time you can actually play with an AI model, without needing to have a computer science degree and a Python script.

So, it allows everyone to engage and play with and see what it can do. But now the tools been out for more than six months, I think it’s about experimentation. Often, I’ll ask for show of hands, in my audience, who’s used chat GPT – and most hands go up. I then say who uses it on a daily basis in your workflow – and fewer hands go up.

So, I think people have played with it, but they haven’t got their head around: How do we actually use it and do something? But the answer isn’t always it’s AI. I think a lot of people are jumping on the bandwagon, saying that we can do AI capabilities, but sometimes people need to take a step back and say what do we actually want to achieve? And is it that we need to do it with AI, or [with] something else?

Areas where people are experimenting in marketing and PR: Copywriting and research that’s certainly very helpful. Product photos: If you have a lot of photographs, you would normally go and have a photographer go and take them, but now you can augment them with a Generative AI. SEO copy: You can now do A/B testing to see what copy is going to resonate. Imagery: There’s a beautiful new tool within Adobe Photoshop called Generative AI fill, that allows you, in real-time, to actually change sections of your imagery. With AI-generated stock photography and video, you can now have an AI presenter. So for training, if you want to scale training videos, you could actually have an AI agenda presenter.

So what we’re seeing is you’re now able to do things at a scale and a speed that wasn’t possible before. So, I think the watchword is: it’s a lot of disruption. But what we need to be aware of, and we might talk about in a moment, is what I call the magnet of mediocrity. If we’re using Generative AI systems to generate copy, it’s going to be boring copy, because it’s the same as everyone else.

So, we still need that creative spark, we still need creative thinkers to say, “I might use the tool to actually research something, but don’t actually use the tool to generate the final copy, because that still needs that human touch to it all.”

Tom Coombes 10:16
If I can just come in on that, I think one of the challenges in our industry generally, is how do you cut through, and the reality is there is a vast amount of content produced by many, many, many – thousands of companies – just in our sector, that is hardly read or watched or digested. And that is typically up till today, generated by people.

So, if you then add on the ability of every organization and every communicator, to generate vast amounts of information based on all the tools that they will all have, you potentially run the risk of increasingly flooding the market with lots of content and imagery and videos that no one actually reads or digest. So, in a way, the human layer to that becomes even more valuable, rather than us just exposing ourselves to an increasing amount of content that doesn’t actually work.

Andrew Grill 11:11
Think about it this way, someone said the other day that AI can be used to create a new Rembrandt or a picture of the style of Rembrandt. But the whole nature of Generative AI, is it’s generating on something else. But what are the new sparked ideas? I look back to music in the 60s with The Beatles: The Beatles were so popular then and their music cut through today, because it was so different. They looked at a style and a quality of the music and the way they tinged it together. That was just revolutionary. And so that’s why they were the ultimate sort of Britpop back. Then you can do a song in the style of the Beatles, but it’s still based on something else.

So, to your point, Tom, I think what we need the creative spark to do, is to create something new. So, by all means, use this AI technology to do research. In fact, what I get my clients to do, is to do a first draft of something using chat GPT, and what it spits out will be average. So how do you rise above the noise? How do you cut through? You need people to do that.

Tom Coombes 12:05
Exactly. And I think one of the things you said that really resonated with me is, if you look at the what, the why, and the how: AI is really [about] the how. There’s been a lot of focus on the tool for the sake of the tool, but actually, it’s about what are these organizations actually trying to achieve? And how can AI enable that?

Larissa Padden 12:24
Interesting, I was talking to a client of ours about AI a couple of weeks ago, and they said that this is just the beginning, that this is actually the slowest, the technology is going to evolve, and it’s going to go at a breakneck pace from here on out. So, I wanted to throw it out to both of you and get your perspective on: Where are we headed? What is this going to look like three months from now? What is it going to look like six months from now?

Tom Coombes 12:45
What we know is that almost every tech company either uses, or is about to use AI and huge amounts of money are being invested in all of these tools. And AI is going to be an embedded component of pretty much all of them.

So, I think it’s a given that we will see an increase in ability, of the many thousands of tools in our sector and in our markets that are out there already. Certainly, we see that in the things that Andrew mentioned, but also in targeting, data analytics, predictive analytics, attribution – even things like crisis communication – the AI turbochargers our abilities there.

So, I think we’re going to see many more offerings and enhancements from the established vendors. I think we’re going to see a lot more solutions pop up in the market. But I think that the fundamental question that needs to be in our customers’ minds, is what are they trying to achieve with it and how can it actually help them?

You can spend all day or week, or a year probably, reading, researching, looking at these tools and seeing what is possible. But I think one of the most interesting elements of AI has always been its ability to learn to achieve the user or the client’s objectives. And I think we are seeing modest enhancements to ability, I think we’re seeing a lot of generation of new content, a lot of noise about that. But I think what’s really interesting is how tools and systems can learn about your preferences, and how they can end up delivering on and fulfilling the objectives you set them. And I think that is probably a way off for us, in terms of stitching all the tools together. But I think fundamentally, that is what we want the AI to achieve for us, rather than showing its creativity or ability to generate new content.

Andrew Grill 14:39
AI, as Tom said, has been around for a while. In fact, the whole ChatGPT stands for generative pre-trained transformer. The transformer bit really came on its own, back in about 2017 when Google researchers stumbled on the way to start to train AI systems in a very massively parallel way. So rather than to tag every piece of content, it can now be done in parallel, that is really transformed. And when ChatGPT came on the scene in November 2022, we saw ChatGPT 3.5. So, there’s been a 1, 2, 3.5. and now, ChatGPT4 is out. And we’ll see by the end of the year, ChatGPT5 – and that’s just one company.

What I think also is interesting, is that OpenAI, the owners and creators of chat GPT, have basically put out a Betta system: They know it’s flawed, they know there are issues with it. And Google and Meta and others have held back because they have more to lose. Google and Meta have huge advertising businesses.

One thing that people don’t realize is that the computing power required for Generative AI query, versus a normal query, is about 10 times. So, if you do a Google search query, and it costs one cent, it’s going to cost Google 10 cents for a general AI query. Who pays for that? Because the content isn’t going to be 10 times more brilliant. So what we’re going to see is a lot of competitors. If I single one out at the moment, Claude.ai are already in the frame, they’re doing things that mimic what ChatGPT does: it allows you to put your own content in there. We’re going to start to see models that are easy to train, that you can drop in a financial services model, you can drop in a FMCG model and these large language models will be prepared, where you can tap into them.

What I think we’ll see also is the ability for young and upcoming companies and agencies to plug in via an API, an application programming interface into a large language model, pay a transaction fee and actually leverage off someone else’s work to then modify that into something that is very, very industry specific. And I think what we’ll actually see, and what I’m excited about is enterprise GPT: Where you take internal data, ethically, legally, and feed it into the system that’s firewalled. You can then ask the question, “What’s the most appropriate channel for this content on a Thursday,” and rather than you having to go off and do all the research, it comes back in 10 seconds, and says, “Based on everything I know, here’s the best channel.” That for me will be revolutionary; it won’t change jobs, it’ll just make jobs more targeted, and leave us more time to do more strategic critical thinking.

Larissa Padden 17:03
Well, you kind of answered my next question already. But, AI has been framed, especially in the media, as this big scary thing that’s going to come in and take everyone’s job. But I wanted to get both of your perspectives on where do you see areas for improvement, maybe inefficiency? I know, you said that there’s still the need for creativity, but what are some of the good things that AI can improve?

Tom Coombes 17:23
The general consensus is that the repeatable tasks will be automated, which I think anyone who works in our industry will probably be super excited about. Because we know that we do a lot of repeatable tasks. So, there’s the potential for efficiency and also making jobs more interesting – i.e. focusing on the strategy or the consultative part of our job – becomes more and more viable as the computers, the machines, do more of the repeatable tasks.

So, I think that’s hopefully where things are headed. It’s really clear that everyone needs to get a better handle on all the technologies and tools that are out there. And they will become even more embedded in our daily workflow. And I think generally people say, “Will we lose our jobs to AI,” I think the answer for me is, not at the moment. But you may well be under threat from someone who’s as good as you, who knows how to use the tools. And I think that is exactly where we need to be, we need to know what’s available, and how to use it best, but man that is no mean feat.

Andrew Grill 18:28
I like ChatGPT to an always on, enthusiastic intern: He or she will be given a task to go off and research it, which may take two, three hours to three days to do that, and they come back with the answer. Whereas if ChatGPT, or these Generative AI systems are programmed properly, it’ll do it in seconds. It doesn’t mean the intern loses their job, it means that, again, what you said, Tom, those repetitive tasks can be done more quickly, can be done at scale and at speed. I think it’s going to be a very important weapon in a communication expert or communication executive toolkit, that you can actually get answers [from] a lot more quickly. And in crisis management, those sort of things, do more quickly. But to Tom’s point, it’s about training and education.

One thing I’ll encourage everyone listening to this podcast is, have you actually set aside some time to train people on ethical use of AI? Do they use a Generative AI platform in their day-to-day work? Is there someone in the company that really understands that can train other people?

I’ve spoken to people on my own podcast, they’ve actually set up multiple training sessions to get people up to speed. I think people need to be comfortable playing with the technology and augment their current job with this new technology. That comes down to training and Tom leading from the top, you’re using this, you should encourage everyone in your organization, your clients to be playing with it. So, you’re going to have to understand what the limitations are and use it for best effect.

Tom Coombes 19:48
The idea of putting together these practice groups and encouraging people to use them, as I said, in a safe way, but within the organization, developing a center of expertise where – to your point –  people know who to go to, in order to find out, not only what tools are available and how to use them, but where we’re seeing real value from them in client work.

Vanja Lakic 20:08
Moving on a little bit to how the media is covering, Generative AI. Is there anything that’s missing for you in the conversation? Is there anything that they’re getting wrong?

Andrew Grill 20:18 
When I talk to clients about AI and Generative AI, I highlight some of the negative aspects about the regulatory issues, the risks associated with it: When you set these models to work, I think we’re going to see a lot more regulation in all industries around explainability. Can you explain why an input AI model, produced the output? And once you’ve set it to work, and it’s running, I think there will be regulators who are saying “Can you prove observability,” So is the model actually running the way it should?”

I liken this to a car, if I’m driving my car along the road, I don’t know how it all works. But I know if the brake light comes on, or the oil light comes on, that’s bad. It’s saying that some tolerance has been exceeded, and you need to take some action, I think we’re going to start to see those sorts of things with AI systems, that in real-time, it’s seeing whether there are any conscious bias that’s crept into the system, whether it keeps producing the right answers, or it’s going off the rails?

So, I encourage clients that they need to involve the legal teams, the risks teams, and when I say that they say “We haven’t even thought about that.” And by the way, GDPR still applies data that’s been sourced from the open internet, the GDPR, in this part of the world and other parts of the world still applies – and people haven’t thought about that. And the media does focus on the loss of jobs, they focus on the negative aspects; that AI will eat itself. [But] there are regulations around that, there are ways to program that. I think what they’re missing is the fact that existing laws apply, and we haven’t looked at how AI would bump up against those existing regulations.

Tom Coombes 21:45 
I think it’s early, early stages in the whole discussion, and the huge amounts being written about what AI can do, obviously, on one hand, you’ve got the huge existential threat that AI presents to society. You’ve got lots of stories and doom-mongering about job losses, lots of comparisons to the Industrial Revolution, and lots of uncertainty, frankly, about where it’s all heading. I think it’s clear that AI is going to generate a huge number of new jobs that it hasn’t generated in the past. In terms of if the media has got it wrong, I think that you can understand why doomsday stories are easier to get readers for than the sensible middle ground –

Andrew Grill 22:45
They cut through, their stories cut through because they’re bad news, so they get through.

Tom Coombes 22:38 
So bad news, so they got through. It’s easy to understand the sort of Doom-mongering about it. The reality is, it’s happening, it’s going to develop really, really fast and people have got to be on top of it.

Vanja Lakicb 22:42 
Andrew, you had mentioned that one of the things that excites you is AI’s potential to transform the data game – enterprise data, specifically. What else excites you, as you both look ahead to what’s possible for the future of Cognito and for the business world at large?

Andrew Grill 23:01
Well, I think because AI works at speed and scale, you’re going to get faster, more accurate answers for clients. And that means that things will move a lot faster. I think about crisis comms when every second counts, how do we get information on that? If I had trained the whole genitive AI system on the company and I could say, “Who is the best person to ask about this?” Rather than running around, it would say, “It’s Tom Coombes, for this reason: This is where we got the data from, and this is what you should say.” And based on that, that is going to actually allow you to have a better hold on what’s going on. So, in the world where everything is moving in AI speed, you need to fight AI with AI.

Tom Coombes 23:38 
Yeah, I agree. And I think, you know, one of the massive problems in our industry, in our sector, is, as we said earlier, it’s extremely crowded, it’s extremely noisy. There’s a huge amount of messaging and content being put out by many, many companies. There are more PRs than there are journalists, there are more marketing tools, and there are recipients of it.

So, it’s extremely crowded and busy. And the biggest challenge in a way is, how, where and when do you get traction and cut through in that engagement with your end audiences? And how do you manage that and do that on a repeatable sustainable basis? And I think a lot of money is spent, and probably there’s a huge amount of opportunity for efficiency. And if AI can help us with the targeting, with the predictive analytics, with the attribution, with the data and analysis, and can make campaigns much more effective or our ability to react to imminent crises, much more effective, I think that will be huge for the industry.

But the converse of that is a lot more companies producing a lot more content and adding to the noise and the clutter, and the end audiences really getting fatigued about all the noise, and all the information, and all the messaging they’re getting, which is all often quite generic.

Andrew Grill 24:56
And you know what will be important is; was that bit about content produced by a human or by AI? And I think the humans will say I trust a human more than the AI. But not everyone’s going to be authentic and ethical to say, this content has been automatically generated. I’m now seeing copy. And I’m seeing books – in fact, I’m writing the opening to my book at the moment called Digitally Curious and the focus sentence says “This book has been written by human,” [because] I want to overtly say that a human being has actually – it’s been assisted by AI, and I’ll talk about where – but I want people to know that they’re reading copy that’s come from a human being,

Tom Coombes 25:28
It’s funny, you said that because w