AI, Creativity, and Crisis Readiness — with Inês Costa
Artificial intelligence is reshaping how we train for emergencies — from the way we design crisis exercises to how we think about communication, collaboration, and creativity under pressure.
In this episode, OSRL’s Inês Costa joins host Emma Smillie to explore how AI is helping transform preparedness. Inês shares how her background in marketing and digital media led her to experiment with AI tools in training and crisis management — and how those early tests evolved into fully immersive exercises now used across OSRL.
Together, they discuss what AI can (and can’t) do, the ethics of using deepfakes in drills, the power of storytelling in technical training, and why experimentation is key to innovation.
Whether you’re in emergency response, communications, or tech, this conversation offers a grounded look at what happens when human creativity meets machine intelligence.
PODCAST
Episode Eleven - Podcast Links
Podcast Transcript
Speakers
Emma Smillie and Inês Costa.
AI generated: 00:01
The fire's spreading and the leaks getting worse. What do we do?
AI generated: 00:06
The Coast Guard has launched a level two response, deploying firefighting vessels and oil boom barriers around the site. Investigators are now examining the cause of the collision.
AI generated: 00:18
Boss, we're trying to contain the leak.
AI generated: 00:20
There's backup equipment on the port side. Try that one.
AI generated: 00:26
What you just heard was AI-generated simulation. There was no fire, no spill, just a digital scenario helping OSLR teams prepare for the real thing. This episode, we look at how AI is transforming emergency training and what happens when innovation meets readiness.
Emma Smillie: 00:57
On the Response Force Multiplier, we bring together compelling experts and thought leaders to provide a fresh take on key issues and cutting-edge techniques in this field. In each episode, we'll dive into one aspect, and we'll use OSRL's unique pool of experts and collaborators to distill that down into actual tools and techniques for better preparedness and response to incidents and emergencies. My name is Emma Smiley. We are OSRL. And this is the Response Force Multiplier.
AI generated: 01:29
Today on the Response Force Multiplier, we're exploring how artificial intelligence is reshaping the way we prepare for and respond to crisis. My guest is Inez Costa, an executive here at OSLR. She brings a fascinating perspective. She started in marketing and communications, but her curiosity about AI led her into the heart of crisis management. Together we talk about how AI is helping teams design more realistic exercises, bridge skill gaps, cross departments, and make emergency training feel more like real life. We also dig into the challenges, the skepticism, the ethics, and the environmental cost, and what it means to experiment responsibly with new technology. This is a conversation about innovation, creativity, and what happens when someone from outside the technical world brings fresh energy to the field of preparedness. Let's get to it.
Emma Smillie: 02:21
Hi Inesh. Welcome to the Responsibly Multiplier. Thanks for joining me today. Uh let's start by just you telling me a little bit about yourself, who you are, and what you do at AerosRel.
Ines Costa: 02:34
Hello, thank you so much for having me. This is really exciting. So I'm Inesh Costa. I work at OSRL as marketing executive. So actually, I've helped promote some of the previous episodes of this podcast. So it feels like an honor to now be the guest on it. It's so cool. But yeah, so I help with OSRLs like social media and email marketing and just promoting who we are like as a brand and what we do. But in the last year and a half, I would say, I've also been involved with supporting our incident and crisis management exercises and kind of like bringing a little bit of digital transformation into that world.
Emma Smillie: 03:16
And that's what you're here today to talk about. So we're specifically gonna talk about the work you've been doing with AI and AI and exercises and also in marketing and commons. But let's take a step back first of all. So what sparked your interest in AI to begin with?
Ines Costa: 03:32
So I've actually was using AI even before I started with OSRL almost like two years ago. So when Chat GPT was first starting to become a bit more popular and people were starting to use it for all sorts of random things. So yeah, I mean, I guess like the cover letter that you read and led to you giving me this job. ChatGPT made it, sorry to tell you.
Emma Smillie: 03:57
Oh no, I even tricked that, to be honest, probably.
Ines Costa: 04:02
But uh yeah, sorry about this revelation time. Yeah, no, I definitely I think I started using it initially with stuff that I was struggling with. And in this case, I think sometimes translating, you know, those skills and all these things that you want to say, especially for someone who's like not a native English speaker, many times like that extra support in bringing things that I would easily be able to communicate when I'm speaking, but sometimes when I'm writing, it doesn't come across as fluid. So it kind of supported me initially with that sort of stuff. But then once I started uh working for SRL, and even in some of my freelance work that I sometimes will do outside, many times you find yourself working in really small teams where you know you have to wear many hats. Um, and you know this, like at OSRL, you're in our marketing team, you know, one day you're trying to make a strategy for social media, and then you're trying to improve our email marketing, and then you're looking at SEO. These are all like highly specialized things that I think AI was really like a good way to kind of adapt to that and support with getting insights to be able to perform on all those like different specialities, really. I think coming from a background where I was working more in the creative industries where the style and the language is very different. And then I came into be, but also like you know, a very serious kind of industry where the language was very different, the way that I communicate, that I had to communicate now, especially in our marketing, you know, like is very client-facing. And, you know, it's not like I can turn around to anyone and just say something like, you know, Slay Queen or something like that. No, you actually have to like use a different type of language, have to communicate very differently. And again, AI was something that was there to support me in bridging that gap a bit. I feel like I used it in so many things, and being able to, through OSRL, actually get access to external training where I got to learn a lot about the different ways that you can use it is really opened up possibilities for how much I can use it for.
Emma Smillie: 06:04
So, what's your secret then? Because I would never have said your cover letter was AI from looking back at it. How do you approach using AI but then making sure that it's still, I guess, I don't know, your voice or just not say like you read some things and you're like, oh, that's definitely AI.
Ines Costa: 06:25
Oh, 100%. I think my approach to it is like I literally speak speak to it like it's a human being. So if you see like the conversations that I have with the app, the chat GPT on my phone, like I always use, for example, the um speaking button, like that option at that feature using your voice. And I'm just literally chatting to it like it's a person. And if it's giving me like a response, for example, that feels like classic ChatGPT sort of reply, then I'll actually turn around into it and say, This is trash. This is like you're talking to me like you're a robot. It doesn't sound authentic at all. Like, do you even know me? We've been talking for so long. How can how do you not know my style? So really just giving it continuous feedback to teach it what you like, what you don't like, what it should sound like, has been really helpful in building those results, that output that actually reflects what you want it to achieve. And it almost helped me build a bit of trust in the tool as well, that I don't think you have when you first start and you get those kind of like sketchy outcomes, and then you're a bit like, this is not very good. What uh what is all the what is all the hype about AI? Like, I can do better than that. I think once you train it and you actually start building that weird relationship with a digital being, uh, suddenly it feels different. Yeah.
Emma Smillie: 07:48
Yeah, yeah. Okay, but I haven't used the voice bit so much, but yeah, I do give my feedback. Yeah, and it it it learns and it remembers. I did actually change a setting today that said it for it to sound like a robot on its reply, like really straight and blunt. Turn that off straight away. I hated that. I didn't as a tone, it was not one I engaged with.
Ines Costa: 08:08
Um but you have to, otherwise it's gonna start doing its own thing, and you're just like, this is not well, you know, we already have to manage like people in our day-to-day life. I don't want to also have to manage like a robot personality. This is too much, too much managing for me. It's not good. Yeah.
Emma Smillie: 08:26
Yeah. Even though I know that they've coded out the pleas and thank yous in um TPT, I still always say please and thank you, because you never know. Come the revolution of AI taking over the world.
Ines Costa: 08:38
You have to be prepared. Yeah. Because they actually remember they will have like a count of all the times that you said please and thank you. So I'm like, yeah, you have to use that.
Emma Smillie: 08:47
So you came in, AI and marketing, you you made a huge impact at that time. And then exercises. How did it then trans start translating into exercises?
Ines Costa: 08:57
Well, it felt very random, to be honest. I feel like I definitely was just dropped into it because you know, I came into the marketing team and I was mostly using these capabilities for content creation, for video and image editing, for design, like all the sort of bits that is quite useful for. And, you know, at the time we also encountered a bit of a skill gap in the preparedness team, more in the sense that, you know, you have these teams that have very technical people that are incredible at their job when it comes to the preparedness side of things, but they don't really have the skill set to introduce design elements, video, photos, all these other things that can add value to it. So that's kind of how I fell into it. That was this feeling that we could make our exercises a little bit more engaging by adding that element of videos, of photos, simulating modern media.
AI generated: 09:53
Good evening. We begin with breaking news off Taiwan's western coast. A service vessel has collided with an offshore substation at a wind farm, resulting in a fire and raising concerns of a million oils.
Ines Costa: 10:04
Didn't really have the capabilities in-house to do it at a time. And that's when I thought, well, if I'm using AI for all this other stuff in marketing, surely we can do it for preparedness as well. You know, if I'm using it to bridge my own skill set gap in marketing, the people in preparedness can use it to now be able to edit a video themselves too, or create a cool image that reflects the scenario they're playing out with the exercise participants. I think in the beginning it felt a little bit weird because, you know, not coming from that technical background, not I've never done like an exercise, an emergency response or anything before. I wasn't really sure how I was gonna add any value to it. So it was very casual in the sense where I was like, okay, here's like what we can create if you're interested. And we've come to discover that I think there are so many skills outside of the technical response or training delivery that are actually so valuable to create that really holistic experience, I think.
Emma Smillie: 11:13
Yeah, I mean, I liken a good exercise to a good story. You have to be able to tell a story with it, don't you? In the same way you do marketing communications, it was all about how you tell that story.
AI generated: 11:24
The collision happened this evening at one of Taiwan's main offshore wind farms. All crew members have been accounted for, though once sustained minor injuries.
Ines Costa: 11:32
Yeah, I think what we discovered is that it actually kind of feeds then into every element of it because in that way you start from the beginning, from the very beginning, even when you're starting to plan anything, suddenly you start finding, oh, I could use AI to maybe analyze all of this data from the client, all this data from their operations, their history, things that happened in the past, current issues they might have, or even their long-term goals. Now we can also analyze the local legislation, the new legal requirements, all this like big sets of data that if you are a single person, you're suddenly like, oh, how am I gonna analyze all this? It's so much. And suddenly you can quickly analyze all of that, and all of that feeds into that story, that narrative that you're creating. And it feels so real, it feels really authentic because you are truly reflecting like the real life scenario instead of me just being, you know, I'm a random person in the UK building a scenario for the Caribbean, and I have no context whatsoever of what's going on there. So whatever it is that I'm gonna come up with is always gonna feel a bit disconnected in a way. So I feel like using AI at that stage, especially, really feeds into that building of a narrative. And then the visual elements that come afterwards, like the cool videos, the images, the social media posts, anything that you create, it just adds to it. You know, it the visual almost like brings it to life in a way that feels a bit more immersive for sure, because you know, we tend to be very visual beings, whether we like it or not. So seeing it, it feels like, oh, this is the real thing. I can see what it would look like.
AI generated: 13:19
Emergency crews are battling flames while responders race to contain a potential marine fuel leak. Authorities say the vessel may have been carrying marine gas oil, a serious pollution threat if containment fails.
Emma Smillie: 13:33
Did you experience um any challenges or skepticism when you started getting involved with exercises in AI?
Ines Costa: 13:39
Yeah, absolutely. There was a bit of resistance on something that feels like two sides of a coin in the sense that on one hand, you know, you come into these projects, and I'm like quite new to the OSRL, and also like I come from the marketing department. So everyone's just kind of asking, what exactly are you doing here? Why are you involved with like an incident response exercise or whatever it is that we're doing? So it felt a bit like having to prove your value, to having to prove that there is a reason why you're there, that you're bringing something to the table that so far the technical team hadn't had the chance to do just yet. And then the other side of the coin is you get thrown into it because now you're in it, and suddenly every it's not like anyone stops to teach you what's going on. They're just kind of like, well, you're in the team now, so they just keep everyone just keeps acting like as if you had the exact same level of like knowledge to start with when it comes to the technical knowledge of an exercise and a response operation. So it's like you're proving yourself, and at the same time you're treated the same, you know? So it was a bit of having to learn about all of it by being thrown into the deep end, just kind of understanding through practice and through experience how exercises work. Having the knowledge of how these AI systems work is almost like you're learning and identifying at the same time. Oh, we can improve this bit by adding this element. We could fix that issue or that lack of communication or anything that's going on by using these different skills.
AI generated: 15:12
Boss, the backup equipment is a really old system. Only Travis was trained on it and he's on leave. No one knows how to use it.
Ines Costa: 15:19
So that's then suddenly when you can prove your value, when you can actually show this is how we can make it better. I saw that pain point that you have. This is how I can contribute to making it better. So I think on a personal level, that's sort of my experience. But I think the resistance in terms of a accepting AI, there's just a resistant to resistance to it as a concept. I think it's very fair that people who have done been doing this for like decades, who've been out on the field getting that getting that like real life experience. Suddenly you come and you show them, you know, a bit of software that, okay, this this can do the same job as you can, which obviously obviously is not true, but you can understand how someone would be like very skeptical and you know a bit apprehensive as well. So that's why it's the process, you know. You start with little things and you show people like the benefits that it can bring, and slowly, slowly you can see people turning around and actually thinking, okay, I can see how this can help me.
Emma Smillie: 16:24
Okay, so let's touch on the process then, because um you talk about little by little, that sounds like your approach is more about experimenting and kind of rather than sort of starting out with right, this is what uh kind of the solution is to a problem.
Ines Costa: 16:40
Is that right? Yeah, there's like bits to it, I guess. I think my approach to it is very much like I don't know. I don't know if it's if it is from coming from a more creative background, but it's very much like the heart of it is experimentation. It really is you you just mess around with things and see what happens, really. There's no point in adding like elements of like technology to it just for the sake of it, right? So I think a way a process that's been working really well for us is been you know, you identify an issue or something that isn't really quite working, and then that's when you start experimenting, you start looking at what's out there, what tools are out there, what solutions could exist, and you just start like trying it out, just really like without many expectations, without trying to force it in a specific direction. Um, I think that's really important, is like going into it with an open mind and seeing what comes back. I think one thing that we found, like trying it out sometimes with very technical, like technical focused people, is that things have to be done a certain way very strictly. And that goes, I think, a bit naturally against what these tools are offering. Or it just goes against like innovation, really. You know, if you're trying to move forward, you need to be open to other ways of doing it. Like that process that you're used to doing it for 10 years, maybe it's gonna look a little bit different now. So I think experimentation is at the heart of it where to come up with a creative solution, yes, you have to identify the issue and you have to be open-minded to what it can look like now. So yeah, at the end of the day, it's not just for the sake of it, is with the intent of fixing an issue, with the intent of making, you know, our lives a little bit easier, really. That's what technology should be there for. So the human element is always there, but it's always like about yeah, just trying it out and see what happens, really.
Emma Smillie: 18:48
Yeah, and I think that's quite hard for a lot of people to get their heads around, actually. I think in marketing, we've been doing it for quite a while. Yeah, yeah. You try something out, it doesn't work, you you try it again in a different way. It's just the way we're wired, but it can can translate quite differently in other areas.
Ines Costa: 19:04
Well, which I understand, because if you think about, I'm just thinking like when I do work with people from uh Swiss, from Subsee, for example, sometimes you're talking about like really serious stuff, you know. We're talking about like if you're not quite detail-oriented and very technical and very precise with things, this could have like catastrophic consequences, right? So you can understand how people would get into that mindset, but it's also about knowing when it's not that deep with other things, you know. When you were talking about changing up a little bit the process that you follow to create an exercise, it's different. There is more freedom to experiment, there is more freedom to try it out a little bit different, and being open to errors. I think being open to failure and actually realizing, you know what, this doesn't actually work. We tried it, let's find a new way. I think that's really important. I don't think you're you have less value for trying something new and it doesn't really work out. That's okay.
Emma Smillie: 20:06
What other skills do you think do you think would people need to I guess learn AI, start with AI?
Ines Costa: 20:14
Is it anyone can do it? Uh I I mean, I definitely, definitely think so, because I mean, from my own background, you know, I don't come from a scientific or technical background, nothing IT related, nothing like that at all. You know, I come from a very much more like creative background. I even like, you know, was traveling for a few years and I was like totally off the grid and like barely using a laptop or you know, anything like that. So now being in a position where I'm like leading certain conversations internally at OSRL that are related to like artificial intelligence, feels a bit like I'm a bit out of place. But I think that's the beauty of it is that anyone can do it is, you know, as long as you have that curiosity, that openness for learning something new. Um, I think for me there was a bit of like maybe a little bit of like almost going into ADHD hyper focused mode, and suddenly you lose yourself in a rabbit hole of learning new things about AI. So maybe there's a dose of that or being open-minded to learn something completely new outside of your skill set and the knowledge that you already have and already been building for so many years throughout your career. I think that's a bit scary for a lot of people, which I get. But I think, yeah, I think as long as you're curious, curious and open-minded, and as long as working with computers isn't too difficult for you. As long as you can do the basics, you can do AI for sure.
Emma Smillie: 21:46
Yeah, I actually to be fair, I firmly believe in that. I think you just have to give it a go sometimes, don't you, and not be scared of it.
Ines Costa: 21:52
I think up until now, the world of like IT and computers like just felt very that's very restrictive because you had to know coding and all the sort of stuff that is not accessible to most people, but that's not the case anymore. So that's why I think that it really is like we're used to seeing this new, like this innovations as something that only a select few can do with, and you have to study and you have to do all this sort of stuff. And it's not like that anymore, I don't think. Okay.
Emma Smillie: 22:23
Um, you've touched on the ability to of AI to bridge skill gap. Well, you want to talk about your own skill gap and how it bridges skill gaps. Um, maybe could you explain a little bit more how that actually works in practice and how you can use it to connect information across teams?
Ines Costa: 22:39
It's more of a reflection of like how traditional organizations are set up. You know, we all exist most of us will exist in like different departments, you know, you have your finance department and then your HR and your comms, like we're all sitting in different departments, and everyone within their team would will have like similar skill sets to a certain extent, right? So it's almost like you have these pockets working individually, almost like a machine, you know, they're all doing their own little thing and contributing to something bigger, but most times you don't have access to that bigger picture. You don't really know how that other department is actually contributing to the bigger picture. So I think what AI can support with is connecting us all in that bigger picture, having us all working in this like common sort of like operating system, but also allow us to kind of because yes, fine, you can have a SharePoint and I can have access to all these documents from finance, but I don't understand what they mean. I have no idea. There's no insights that I'm taking from it. But having AI kind of bridging that skill set where it can potentially help me analyze all these documents and all this data sources from across the company to bring me insights that can inform my own job, my own role within that big machine. I think that's a massive thing. I think it can allow us to make better use of the data that we have. It can allow us to make more informed decisions, more connected decisions, and more collaborative as a way in a certain way, because I'm also like, now I actually know what the other teams are doing. Now I actually understand what's going on. And I can think, oh, maybe we can collaborate on this. Maybe what I'm doing actually complements what that other person is doing too. And we can still work together, even though we're not even speaking the same language, both like literal like languages across the world, but like skill set languages, you know, even though we're doing very different things, we can still understand each other and still collaborate in that sense. It really will allow us to go from that traditional, we're all separated but working towards the same thing kind of organization to a much more like almost mission-focused sort of organization, I think, where we see the bigger picture and we're working in a more proactive and more collaborative way towards the same goal, I think.
Emma Smillie: 25:10
The question in mind then is will AI drive better cross-functional collaboration, or do you need that cross-functional collaboration first in order to get the most out of AI?
Ines Costa: 25:22
Yeah, I think that collaboration is not about a tool. A tool just like enables it. I think collaboration comes from the culture that you exist within. I think we have to recognize that if you don't set up the correct tools for it and you actually have just blockers to collaboration, then yeah, people will have much less incentive to uh to do it. But when you put out the tools out there to do it and you make it so much easier for people to collaborate with each other, then it's all down to culture 100%. I think that's why it's important to before we before you start like spending on all sorts of tools and all sorts of like new gadgets and processes and whatever it is, it's all about actually creating a structure that promotes that culture and invites people to collaborate with each other. I think that's what we're trying to do with OSRL right now. So at this point, we kind of created a sort of working group focusing on AI, as you know, because you're in it too. Basically, we created this group, which the whole point of it is that creating that culture of collaboration around AI, we're bringing in people from the whole organization, like from all this different department. And the idea is to have this sort of collaborative approach to AI. So instead of reinforcing that idea of having different pockets, each one of them trying to figure out how AI fits into their own team, we're actually bringing them all together so we can have organizational-wide conversations about it, because the solutions that I find for my team might actually be relevant to the pain points you have in your team. And you might have really cool ideas that I would benefit from. And we can actually support each other in implementing them, specifically when we're talking about quite complex issues. It's all about promoting that support, that talking, the sharing ideas. It's not about gatekeeping, it's actually all about opening up the conversation so the more people can contribute, the better outcomes we will have. I think when we think about AI nowadays, we most of us without even realizing we're talking about like content generation, right? You know, creating images, creating social media posts, creating emails, whatever it is. But I think the real power of it lies like within like data and the way that it can, like the management of data, the access to it, the insights it can bring, the way it can connect people who are, you know, on the ground, for example, to the people who are planning, to the people who are communicating, to the people who are strategizing. It's all about connecting all of those because they're all working towards the same thing.
Emma Smillie: 28:03
Yeah, yeah. And that's where the power comes from, isn't it?
Ines Costa: 28:10
Similar to like, you know, when you think about the internet and giving us all this, like, you know, when it was first becoming more accessible to everyone, the level of power that gave people and having so much access to this level of information. I think AI is just another natural step in that same progress, to be honest.
Emma Smillie: 28:31
Yeah, it's the power, but then there's also the challenge, I guess, with data. I know that's something you've come across again when we've been talking about China, how we harness it better. Maybe you could explain a little bit about the challenges that you've faced in exploring AI and the the s concerns, the legitimate concerns of people.
Ines Costa: 28:53
Yeah, 100%. I think there's, and that's something that I've included in, you know, a couple of the presentations I've done on it, is there are many layers of issues with it. I think obviously data security is immediately the first one that you come across with, especially for us operating in the oil and gas industry. We have major corporations that are, you know, trusting us with their data, with their information, and we're just thinking, are we putting all of that at risk for the sake of writing a quicker email? You know what I mean? It really, I think we've had uh we've been very lucky because we do have a very supportive team with an OSRL who's you know really willing to work with our teams to find those solutions, to find what is doable and what is like a real risk that we need to like be aware of. So that I think would be the biggest risk that we encountered. And I think meeting halfway has been the way to go where we have those teams whose job is to be the Debbie Downer who's Says no, you absolutely can't do that for like legal issues or IT security, data security issues, but then this is what you can do instead. And then being the person on the other side who's trying to push for these new tools, also having that capability of saying, no, actually, I shouldn't be pushing forward just for the sake of it. I need to do it in a way that's safe, in a way that's like as ethical as possible, in a way that actually doesn't put us all at risk in a way that's completely unnecessary. So, yes, maybe I won't get that tool or that future, that feature that I wanted exactly. I will get this one instead that allows me to do however much it can do. And I think that's fine. I think it's about meeting in the middle to actually find something that works. But there's so many other considerations to have, you know, for me, like something that's been really highlighted in the last year has been the environmental impact of it, which is something that actually was, I got a real reality check speaking to people at Interspill who kind of showed me like actual images, like photos of the environmental impact it was having in the regions around where the data centers are operating from. And it's really made me think more like, right, this is why we actually need to invest in educating people and we need to actually be clear about what we are using it for? Are we using it for things that are like worthwhile? Should I be, you know, using a technology that might potentially harm the environment to write little emails and you know, create little messages, or should I use that power to actually for big stuff, for stuff that actually makes a real difference that brings good out of it, almost in a way that like, do the two offset each other? I don't know, but I'm hoping that using it for the right things rather than just for anything I can think of actually makes it better.
Emma Smillie: 32:02
Carbon footprint of it is quite significant, the more data I'm reading on it.
Ines Costa: 32:07
Yeah, a hundred percent. And I think like more and more, I don't know, more and more legislation might come from it and to try to like stop the overuse of it and stop the corporations that are leading it from just doing whatever to get more profit. We don't know what's gonna happen. But I think if we get ourselves in that role of using it in a conscious way, in a way that actually like brings benefits, if we're like teaching each other how to not be lazy with it and just using it for any single thing, yeah. Who knows? Maybe we might use it for something good. Yeah, absolutely.
Emma Smillie: 32:49
You mentioned we've done a few presentations on AI and AI and exercises. You did a paper with Dave Rouse on Interstill, I think. Um is that the one that if you're not using AI in your exercises, what are you doing? Is that the one?
Ines Costa: 33:04
Yeah, that's the one. Strong title. Strong title.
Emma Smillie: 33:10
What was the inspiration behind that one?
Ines Costa: 33:12
Once again, it was something that I was pushed into it. I did not think, maybe a bit with like right at the beginning when I was talking about getting involved into exercises, not really seeing the value that I could add to it. That's exactly how I felt about writing a paper for conference in the oil spill response industry. I could not see how I could bring any value to any of this because, you know, I haven't been doing it for that long. And there's people with decades of technical knowledge there that I'm sure have so much more to offer. So what am I doing here? So I just got pushed into it, which I think is okay because I I'm happy that I did it. Um, because as we're we were writing it and as we were discussing all the things that we've been doing and all the value we've been creating, all the cool stuff that can come from it in the future if we decide to explore it more. I think I actually realized somewhere along the way that wow, what we're doing is actually can bring some change. It could actually allow us to grow and it has applications outside of exercises too. It can really be almost like an entry point to a world of change within oil spill response. So yeah, it really was a revelation moment of okay, maybe we do need to talk about it. Maybe we do need to showcase all the work that's been done in the background, even if like 80% of it was just pure experimentation and seeing what's gonna happen with it. And yeah, we had really like such good feedback from it. It was really like overwhelmingly positive. People were so happy with the presentation, they were so interested in it. We had so much follow-up. And every day we have more clients like and members reaching out to us saying that they're interested in having AI input into their exercises. So I think that's that really tells me that we needed to do it. We needed to share that knowledge and get everyone involved and get everyone curious about what all of this is.
Emma Smillie: 35:17
We've talked kind of top level, but if you could maybe dive more into the detail around that. So how does AI help make exercises more engaging, creative, tailored, kind of from a storytelling perspective, or even with the data elements that you talked about?
Ines Costa: 35:34
I think we touched a little bit on it earlier. What you know, I think the very beginning of it is that access to like all this different type of information, what it really gives us is easy access to in-depth kind of like insights. I think in a traditional exercise, you'd have this team, which many times you know is very, very small, kind of might be just one person designing the whole thing. Being realistic, we are limited in terms of budget, in terms of resources. And, you know, you might be working halfway across the world in a completely different region and trying to build an exercise for a company that's completely different from where you're used to, with very specific operations, with very specific regulations that have very specific goals that they're trying to achieve with their exercise. So in an ideal world, you'd have a team with a specialist in each one of these topics that you have to bear in mind when you're building an exercise. But that's just not realistic, right? Um, you'd spend so much time going through all this documentation, all these papers, all these plans. It's just not realistic, really. So I think that's what it brings is that capability of like quickly analyzing all of these sources of information and then bring it all into one narrative, into a format that fits the exercise you're trying to deliver. And just allows us to kind of have insights on our who our audience is as well. I think similar to like how in marketing we have to know our audience, like, you know, understand their needs, understand their history, their motivation. The reason why we do that is because we're trying to build something that resonates quite deeply with them. So that's the same that we're doing. That's the same that we do with our injects. So we will use AI to create, you know, scripts from our phone calls, our emails. We can use it to create images of the environment we're playing the exercise in, suddenly affected by a massive disaster, and we make it quite real and it really resonates with the participants. We can create this sort of like media, news reports, social media interactions, social feeds, all this sort of stuff that actually mimics what it looks like on that side of the world. We're just doing something recently, for example, for El Salvador, and you know, having access to the information of how the media works there, then you know, being able to mimic that in our exercise too. This is all things that if you didn't have access to AI to do that quick search and quick, quick kind of like reasoning and bringing it into your narrative. One person who's a specialist in a very specific part of oil spill response wouldn't have the capacity to do all that. So I think that's all that we're adding to it when it comes to the design and delivery side of things.
Emma Smillie: 38:29
Have you got any other examples of exercises that um you've been involved with, which have used AI?
Ines Costa: 38:36
Yeah, I think we've we've used it in quite a few things at this point. I think the first time that we had a really interesting reaction to it was the first time where we used video news report in the crisis management exercise for a client. And it was like a very high-level executive team, right? Everyone was very confident in what they're doing. Oh, I I know how to handle myself in the crisis. This is easy, this is fine, we have it under control. And then you show a news report video of the biggest national news channel showing the face of like we try to mimic that look of the news report, like really down to every detail.
AI generated: 39:14
Environmental officials confirm nearby wetlands and habitats for the Chinese white dolphin are at risk. Satellite and drone surveillance are being used to monitor any surface sheen spreading towards shore.
Ines Costa: 39:25
And then you had a photo of the CEO and criticizing what the company is doing, really doubting their capabilities, their competency, and just really making them look really bad. And suddenly the feeling in the room, the energy changed completely. Suddenly, the CEO was very serious, and everyone was like, right, let's get down to work. We need to get this under control. This is serious business now. And just getting that feeling on how an inject like that can really change the energy in the room, because I think we have to be honest here. Technical people don't necessarily always create the most exciting documents or the most exciting injects. They love a blank page PowerPoint. So I think when you bring in this element, those visuals that feel so real and like quite high production, but it was pretty good. Uh, when um, you know, when you bring those elements to to play, it really sticks, you know, it really makes a difference. Yeah, we've used it in so many other things. And, you know, Emma, you were there when we did our interactive workshop at the members' forum uh last December, uh, where we gave our members a bit of a taster of all these different things we've been experimenting with. And we ran a full-day workshop where we introduced incident management and crisis management with like interactive elements, with a reactive uh scenario exercise. And I mean, we couldn't have done any of that without the help with AI. You know, I think we managed to deliver all sorts of like realistic media injects. We had personalized video news reports that perfectly reflected all the actions they took throughout the day. There was branching scenarios as well, like anything we could think of. We threw it in there.
AI generated: 41:31
Where is it? Come on, where's the contingency plan? Boss, Alvin is badly injured. He needs urgent medical care. What do we do? There's a phone number on the plan. Let me find it.
Ines Costa: 41:45
And none of those things would have been maybe possible, but they definitely wouldn't have been as like streamlined and kind of slick in the delivery as they were if it wasn't for the support of those like digital tools and AI. I don't know if you agree. I I think so.
Emma Smillie: 42:01
Yeah, no, I agree. Definitely. I think just the ability to dynamically change what we're doing and um yeah, the real-time stuff, the branching, yeah. We wouldn't have to do any of that without AI. Yeah. I mean, I I haven't I've been involved in a few exercises without AI, but now primarily it's always got some sort of AI element to it, I guess. It just feels a bit more, I don't know, flexible, a little bit less rigid. Um than when we've had it very much kind of scripted out, etc. Exactly what's gonna happen and when it's gonna happen. What are you what are your thoughts on terms of the participants' experience? How does it make a difference?
Ines Costa: 42:45
I definitely think that is exactly what you were saying, is like, you know, you have this team, the prep team, the planning team, who puts together the script, right? And especially when we're talking in um imagine a larger scenario or like a longer scenario that can last a couple days, you know, you have this team who prepped so much for it when you have the inject timeline. It can be massive, like 30 pages of injects, you know what I mean? And imagine on hour four of a two or three-day exercise, the participants decide to make something crazy, like they have they make a crazy decision that completely alters the way that you plan the rest of your exercise. You know, changing all of those injects that are going to come after it would be a really crazy amount of work. And it would take the delivery team and the sim cell team a bit away from that focus on the participants and the focus on what's happening in the room because you'd be so busy in the background trying to edit your scenario and your inject. So I think the fact that now you can literally just feed it in whatever tool you're using and say, this happened, change the next inject to reflect that, and they can do it in like in a couple of minutes, that changes everything, you know. It allows it to be more flexible. So now, as a participant, I won't have someone in the room tell me, oh yeah, I know that that happened, but this is how we're playing it. Never mind that. We're playing it like that because this is a script that doesn't feel real. That's not what would happen in real life. If there was an actual incident, oh, this is what so this is what we prepared for. Well, too bad. You need to adapt. So I think our exercise is being able to adapt to that too. It just feels so much more real. And you know, it almost makes you feel as a participant. So, for example, what we experienced in that workshop in the members' forum where we had branching scenarios and the media injects were like reflecting what they decided to do. The feeling that people were getting was, oh, every action matters. Everything that I decide to do actually actually reflects in the real world of this exercise. And that feels more real and that feels more engaging. And I think for the participants, that makes like a world of difference. And I feel confident in saying that because the feedback, like you said, we've been using it quite a lot in the exercises we've been delivering this year. And the feedback from it is always like so positive. We always get comments on like how great it was and how much like it added to the experience for them.
Emma Smillie: 45:26
What about things like deep fake videos, things like that? Are we have you been using them in AI and exercises?
Ines Costa: 45:34
So funny enough, I think those are like our best sellers. We're just so good at it. It's good. That those are exactly like you know, these videos that we've been creating reflecting like the media or reflecting people from those countries that we're delivering the exercises in. Those are definitely the best sellers because that's when suddenly it feels less like an hypothetical scenario and more like something that's actually happening. And it feels wrong doing it almost, you know, because you know, when we talk about deep face, we're talking about like using people's likeness and creating fake news and all this sort of stuff. It feels evil, it feels wrong, but we're using it for good because we're, you know, preparing them for an emergency. So is it okay? I don't know, but it's definitely like our best seller, I would think.
Emma Smillie: 46:29
Yeah, there are obviously there are ethical implications in AI, but that's for a whole other podcast.
Ines Costa: 46:35
I mean, to be fair, now we're getting to a stage where obviously through during the experimentation stage, we really were just messing around with it and see what we could create. And now that it's become almost part of our service, we have done as much as we can to kind of keep it, you know, to keep it safe for the people involved. So for example, we're not creating avatars or deepfakes of like any of our clients or any of our employees or anything like that. We're making sure to protect everyone's like likeness and uh, you know, not creating videos that could be dangerous. But also like all of our videos, for example, have watermarks across the entire screen saying exercise, exercise. So no one could really use those videos in a way that could look like fake news. The more we learn about it, the more we're very aware of the fact that you need to do all these little things to keep people as safe as possible.
Emma Smillie: 47:27
You touched on um gamification, which I know we kind of we did a little bit at the AGM. Um, have you done any more work in that area in terms of how we can use gamification exercises?
Ines Costa: 47:39
It's still very early stages for that, I would say, for us. I mean, it's been really cool to see how other people in the industry are kind of using like low-tech solutions for this. So obviously, we've seen people delivering incident management exercises using Lego, which was really cool. Or was it response, something like that? But they use like Lego for it. And we've had like CLARM come to USRL and deliver their wildlife board game and stuff like that. So I think now we're trying to figure out how to bring those experiences into the digital world. So far, we only tried it with like little workshops, small training sessions, nothing too crazy, but we found that introducing stuff like digital quizzes that have like a leader score kind of board to add that element of competition has really worked really well, gets people really excited and really invested in things that normally we really struggle with engagement when we're talking about like ICS and IMS kind of stuff. Suddenly people are very invested in it, they're loving all that content. But translating also like board games into digital board games, and um we've done something with the cone of response where we use like a digital sort of like board map where people teams can interact together in the same map, and they can draw, for example, zones of response, they can identify environmental sensitivities, they can identify according to that what's the best method of response, but all in a map where the whole team can play together in the same place. But the future, I think, will be really interesting. We are we already discussed and we already started looking into a little bigger projects. So we're talking about like sort of gaming platforms, like a gamified experience of exercises where you have this one platform where you have access to like all your resources and all your methods of responding to an incident, and then you have this board with an environment where you know your incident is happening and you can drop resources into that. So it's almost like oil spill response meets Sims or something like that, which would be quite cool. But this is all like ideas and little tests that we've been running in the background, and hopefully in the near future they can come true.
Emma Smillie: 50:03
Yeah, I guess again, another podcast potentially about how you'd combine like AI with VR and AR to create a whole visceral learning experience could be something to explore at some point.
Ines Costa: 50:14
100%. I mean, there's always like limitations in that because obviously, like none of this is cheap. And yeah, it's about picking again, picking what's right and picking what works and not adding tech for the sake of it, but actually creating an experience that adds value. Yeah, I think we're getting there.
Emma Smillie: 50:35
So let's let's let's look to the future a little bit. In general, what are the trends you're seeing in the broader kind of AI space?
Ines Costa: 50:44
I think we've seen, most of us probably have seen that it seems like every single tool, like software or anything that you have has an AI integration now. So I think that's the biggest thing is just been this expansion into every almost element of your life. Some work better than what others, you know. I think you can really see the companies who are very strongly invested in their AI integrated capabilities. So I think that's one of the reasons why I feel like it feels a bit redundant to be, oh, I'm totally against AI and I just refuse to use it. No, is basically the main trend happening at the moment is full integration into all the systems that you use nowadays. But I would say the big trends coming up are definitely the agents. I think that's what everyone is really excited about. Everyone's loving that idea of being able to use text and natural language to code an agent that will systematically do this task for you. I think that's very exciting for your day-to-day life, but also particularly for your job. I think we're going now, and you can see it with like open AI releasing sort of their like lifestyle agent mode, where we're kind of using, we're going from that you open your chat and you ask it, you ask for a specific question or a specific task, and it does sort of you go step by step in doing it. We're going now towards that more you go in there and you do a more broader, more like general request, and it can actually integrate into this like different apps that you have, and it can the AI itself can break down the problem, the request, into its many little steps and perform each one of them almost like a real life assistant, you know. So I think that's like the the real big big trend that we're seeing at the moment uh with the agent. Gemini as well, they're releasing like the deep thinking kind of setting, right? Allows it to do like also deep search online for really specialized type of content, which I think in our field is like super important because it will allow AI to kind of be able to give us access to insights that are like less generalized online and make it like really specific to the technical knowledge that we need to have to perform some of the tasks we need to do for the operational side of our business. Uh, I think it will allow us to gain more trust in AI for to support that side of a business, which I don't think has been happening so far. But yeah, just having that capability of performing deep search, retrieve specialized content, accurate information, and really focus on like a deep thinking mode that almost feels like more human, you know, because it it is capable of processing a complex issue into you know a set of tasks and a final outcome, which is how our brain works, really. Yeah, absolutely. And then specifically for kind of the oils for response industry? I think the main interest that we've seen in the like so when we've been out talking to our members and just going to this like industry engagement, the questions that we always get are okay, how are we going to bring this into operations? You know, it's all well and good that you use it for marketing or you use it for exercises. How do we bring it into operations though? And I think that's how I see it going next, is as we're building these capabilities for like deep thinking, for cross-functional agents, we will now be able to use AI almost like system integrators. So we can integrate the systems from ops directly into the systems of support planning, and to comms, and to all the other departments that support each other. I think we'll be able to learn more from the existing data to kind of support our learnings and our strategies in the industry, and particularly for operational terms, we can learn more from like past incidents, from current operations, from current plans. We'll just be able to like quality data will be so important, you know what I mean? Like from an operational side of things, instead of looking at keeping data as just something that you check the box that you've done it, when you actually focus on creating that quality data, then that is going to be what's going to directly inform your strategy. That is like so powerful. You know, you're basically bringing the data from the ground directly into the heart of how you strategize for the future. And sometimes that's been a disconnect because they feel so far apart in the hierarchy of how corporations work. So I think that's gonna be it's gonna connect the two, it's gonna bring them closer. One thing I'm really excited about actually, and I've been talking to some people internally about this for like months now. But when we talk about agents, you know, when we're talking about like this agent that, you know, use you train it for a specific task, and then, you know, it can do it all the time for you. I can definitely see how we can train our own agents to create a little oil spill response advisor in your pocket kind of situation, you know, you train an agent, an AI agent with like all the oil spill response techniques and all the field guides and all the regulations. You can even train it with like learnings from past incidents or from preparedness work. You can, you know, if you're lucky enough, you might even have people in your organization with decades of experience who are willing to speak with the agent to train it in certain things. And then suddenly, when you're out on the field and you're like not sure what to do in that specific situation, you just don't have, you know, those field guys with you at that moment to really advise you the best, you can just ask your little pocket advisor, hey friend, what should I do in this situation? What do you recommend? And obviously you shouldn't blindly follow what the little AI tool is telling you, but it can actually like inform you in certain things that you wouldn't be able to access that information in another way, or if you would, it would take you quite a long time to dig it out. Yeah, could turn the TPR wheel into a little agent. Exactly. You know, I I can see how that like for those more practical terms, how we could definitely already come up with some stuff that can be used right now. But yeah, I don't know. I think more long term, I think having that access to information in a completely different way, I just think it will open people's worlds a little bit. Okay, so if we're working in organizations and in teams or people come from the same background and are have similar skill sets, or maybe are living inside the same bubble, it can be really difficult to think outside the box. So I'm just thinking how, like, maybe AI is a tool that is capable to retrieve information and insights from anything you can imagine all over the world, from any perspective that you can imagine. I'm just imagining we're like some good brainstorming sessions that you can have that maybe can give you ideas of things that you didn't know, you didn't think about. And will that inform the future strategy of like, you know, oil and gas industry, of how we adapt to the energy transition? I don't know. So many things that can come from that open-mindedness and using it to uh broaden our horizons, I think.
Emma Smillie: 58:46
So let's move on to some closing reflections. So, what's one misconception about AI and exercises or marketing or in general, actually, in any sort of context that you'd love to clear up?
Ines Costa: 58:59
I think the main one is like AI will not do your job for you. I think the the misconception being that people think AI is gonna come and take away your job and do everything for you, or that you, you know, you're gonna start to use Chat GPT and suddenly you don't have to do your job anymore, and that's it. You can just get paid to sit without doing anything. I think that's a big misconception. It cannot do your job for you. There is no magical button that you click and suddenly you say, do this for me, and it just does it perfectly, and that's it. That doesn't exist. I think what AI is not gonna take your job away, it's just gonna change aspects of it, right? So maybe, you know, for myself before, I'm just thinking when I first started, I would spend hours and hours actually writing the content of the social media posts or the emails that we were sending. Right now, I have a tool that does those very quickly for me, but then I spend the rest of the time like researching and strategizing and getting insights to. Know what to ask the tool to do for me, right? I think at the end of the day is just a tool, you're still the human, you're still in charge of it, right? So the creativity, the knowledge, the value that you put into it, like the rule with AI is garbage in, garbage out, right? So if the prompt that you're writing, the information that you're putting into it, the things that you're asking are not, they're not good, they're not, they don't have good quality, what you're gonna get out of it is also bad. So your job's not gonna be done. I think it's about that change of mindset of being like, I need to adapt to instead of being someone who spends hours writing this, now I'm someone who has deep insights of the market that I'm operating in, someone who has a clear vision of what I want to achieve, and someone who knows how to communicate with this tool to then get the best outcome. I think it's just a change in mindset is at the end of the day, you're the human, you're in charge, the tool will do what you make it do, basically. So you decide what is worth using it for. Yeah.
Emma Smillie: 01:01:20
I think creativity, problem solving, critical thinking are gonna become the skills that people need going forward. 100% more than anything, and then you can use AI to help what max out your productivity, for example, or um just save time.
Ines Costa: 01:01:38
100%. I think we talked about this even the other day. The two of us were saying, like, about the AI sandwich, right? And how like, you know, you are the beginning point where you have the clear vision where you're going, and then you use AI in the middle to kind of do all those like boring bits in the middle that really do have the capacity to like that's where you can work your productivity in. And then you are again at the end of the process where you will review what's been done, you will like make sure that it lives up to the standards of quality of what you want, and you understand how to integrate it into the wider picture. It's all gonna come down like to your creativity and your problem solving. That I think our jobs are gonna start paying us more, not more. I don't think they're gonna pay us more. I think our jobs are gonna start paying us to be more high-level thinkers and more problem solvers than you know, machines who do repetitive tasks, I think.
Emma Smillie: 01:02:41
So, what's the one thing? If somebody's listened to this podcast, what's the one thing you would like them to take away from this conversation?
Ines Costa: 01:02:49
I think don't be scared of it. Uh, I think I can confidently say, just from the way things are going right now, AI really isn't going anywhere. It's kind of like it's a development that it's been, you know, we've been seeing it like theorized and imagined in sci-fi for so long. It was just bound to happen and it was just bound to become a thing. So now it's here and it's not going to go anywhere. So it is scary, it is overwhelming. I think even for myself, I the more I learn about it, the more weary I am and the more aware I am of what I'm using it for. But there's no need to be scared of it because, like we were saying at the end of the day, you still have the power to make the most of it and to use it for what benefits you. You know, it's within your power to use it for the things that you know that you struggle with, with the things that you just don't like doing, for the things that you think you could maybe do better with a little push that's just not available to you at the moment. It's there to support you with those things. You just have to use it for those things, you know. Nothing is gonna take away your intelligence, your creativity. And like when I did a course on AI earlier this year, we started a session with our trainer going online and using AI to analyze the job ads that were out there on LinkedIn for the last month in the UK for marketing roles, right? And he analyzed the job description for those job ads and asked AI how many of those responsibilities could actually be replaced by AI content creation and automation. And it gave it a number like 49%. It was something crazy. Almost like half of the marketing jobs available in the UK could potentially be replaced with AI. So when you hear something like that, you're like, that is so scary. But at the end of the day, I think it's about that change in mindset of it's only scary if you look at it like that. But if you look at it like, okay, if AI is gonna do 49% of my job for me, what else am I gonna bring to the table? What are the really cool stuff that I'm capable of doing that I just don't have the time to do right now that I can now bring to the table and just wow everyone and just be amazing. So yeah, just don't be scared of it and make the most of it, and I think everything's gonna be okay. And experiment. Always experiment. Always, yeah. Because it always just gets boring, you know.
Emma Smillie: 01:05:34
Well, thank you so much for joining me today. I have a lot, I probably had like a hundred more questions I could have asked you, but um, we can always do a part two.
Ines Costa: 01:05:42
Oh, yes, we'll see if it's like the public will demand for it. Oh, we must know more. That's great. Okay, thanks, I thank you so much. It was awesome.
Emma Smillie: 01:05:58
Thank you for listening to the response force multiplier from RSRL. Please like and subscribe wherever you get your podcasts. And stay tuned for more episodes as we continue to explore key issues in emergency response and crisis management. For more information, head to osrl.com. See you soon.
