International AI Narratives Research
A 300,000-word sample of how political leaders, the civil service, industry, advocates, academics, and the public are talking about AI across the EU, UK, and US, and why the public keeps ending up inside industry's worldview.
Across all 300,000 words, four narrative frames carry almost all the weight: Journey-Race, Journey, War, and Building. Each comes with its own logic about who’s responsible, what the “prize” is, and what good looks like.
Industry speakers use Journey-Race with striking discipline: same frame, same entailments, same logic, every time. Advocates and academics use a wide, inconsistent crate of frames, never settling on a shared narrative logic.
Public servants reach for Journey (“we’re on the path, making progress, navigating challenges”). Industry uses Journey-Race, a more specific cousin. To a legislator on a committee, both speakers sound like they’re saying the same thing, but only one of them has the goal defined.
The War frame raises the stakes but never names a credible enemy or a vision for victory. The public, looking for a baddie that justifies a war footing, lands on “foreign state actor”, at which point industry’s “we need our champion to win the race” suddenly sounds like the right answer to the public.
What if I told you there was something that radically rewired the way we understand the world? That makes up to 40% of the words we use, but less than 3% of people even notice.
Today we’ll be talking about how the future of AI will be determined not just by the technology, but by the words we use to describe it and to talk about it. Many technologists, policymakers, and advocates often describe wanting to see a future where AI is accountable, transparent, ethical, and open. More deeply, what they’re speaking about is a future where technology is used to make a more equitable and just society where everybody can thrive. But the words we’re currently using aren’t anchoring those ideas in people’s minds.
Over the next 30 minutes, I’ll outline how we can improve, and in doing so, how we can shape the way technology companies are building these tools, the way legislators are thinking about governing them, and how the public understands what those two groups are doing.
All of this builds on a year-long cognitive linguistic research project done with the Centre for the Future of Intelligence right here at the University of Cambridge, who have led AI narratives research globally. It also builds on my decade of experience working in Australian politics, where hundreds of focus groups illustrated that even small changes to our language can have enormous effects.
But this journey doesn’t begin in Silicon Valley.
So grab your crampons and let’s go back to 1787, when Geneva naturalist Horace-Bénédict de Saussure decided to climb Mont Blanc to answer a seemingly childish question: why is the sky blue?
After a gruelling 17-day hike across treacherous glaciers and icy cliffs, de Saussure and his team of 19 porters finally made it to the summit. They’d been carrying a bunch of cumbersome scientific equipment, including a device de Saussure had built himself called a heliothermometer, essentially a wooden box with a glass lid. He left it out in the snow for a couple of hours, and when he came back was struck by the fact that it had become scorching hot.
Writing in his diary, de Saussure concluded that there was something unique about this box: that the sun’s rays entering through that glass pane would heat the air and then trap it inside. He didn’t fully grasp the implications, but in 1824, Joseph Fourier concluded that the Earth’s atmosphere functioned much like one of these. It would heat the air, and that heat would stay inside.
Over the next 150 years, scientific concern about this grew and grew, and became a certifiable crisis by the 1970s. But for some reason, the public didn’t really understand or particularly care about Fourier’s theory of gas insulation, no matter the grave implications.
That all started to change in the 1980s, particularly in 1988, when NASA scientist Jim Hansen testified before the US Congress. Describing what was happening to the globe, he used the expression “the greenhouse gas effect.” It captured the public imagination, was widely used in media coverage of his testimony, and became a key concept in the first ever IPCC report in 1992.
So even though we had known the science of climate change for almost 200 years at this point, it wasn’t until we discovered a useful metaphor to describe it that the public was able to understand what was happening, who was playing a role in it, what we might be able to do about it, and the likely consequences if we don’t.
When topics are large and complex and hard to understand, such as climate change or the economy or AI, we really need a bridge, something that can take us from the known to the unknown. Metaphors are that bridge, and they’re often overlooked. They make up to 20% of all speech, 40% of all political speech, and they’re core to how we understand and process the world around us.
Medical science has demonstrated there is a small portion of our brain responsible for metaphoric reasoning. If it’s damaged or incapacitated, our entire cognitive system goes down. We’re unable to understand the world around us.
But it’s more than just being able to take the intangible and make it tangible. This process of translation also encodes within it a series of values and ideologies, or ways of seeing the world. And they’re also incredibly politically persuasive. Their subconscious nature means they often cut past people’s biases, a real power in our polarised world. They embed deeply in people’s minds and shape how they think over the long term and into the future.
These tools, amongst others, mean that metaphors are really good for facilitating conversations about complicated things we don’t understand. Looking at a Google Ngram viewer of the expressions “greenhouse gas effect” and “global warming,” we can see that the greenhouse gas effect as an expression became almost the carrier wave that allowed people to start thinking about and understand global warming.
This is also of incredible importance while we’re thinking about AI, because we’re currently in the process of determining the very words we’ll use to describe these tools. What even is it? Is it a co-pilot, an agent, a partner, an assistant, a servant, a tool, a teammate, an oracle, or a golem? They’re also critical for how we describe their implications and dynamics. Is it transparent or open? Is it a black box? Are we accelerating or decelerating, hallucinating or bullshitting? Are we aligning or diverging?
We’ll be making decisions every day about the kind of concepts we want to embed in the tools we are building, and they have consequences. The media are carrying this along as well. Almost every headline is metaphoric. AI is eating the world. There’s a high-stakes battle. There’s bubbles about to burst. Europe’s rushing to set rules. There’s a godmother of AI.
Understanding how these work is of critical importance to advocates who want to shape the future. It’s also critically important for the more technical and policy-orientated folk who want to ensure that governments can effectively regulate market power. As regulators are being established, they’re going to take a while to develop the sensing and the monitoring tools required in order to know what’s happening in these companies and in the market. It’s going to take longer still to have a sense of their powers, to start bringing cases, and start winning cases. And we know that the most important and powerful way of shaping market behaviour is through enforcement activity.
Yet we’re already seeing examples of compliance theatre, where in the EU, companies are rushing to label their models as open source to take advantage of much lower regulatory thresholds on those sorts of models. So while we’re waiting for regulators, establishing and reinforcing social norms is the most powerful tool that we have. And they’re critical for developers who are hoping to build ethical AI models and products. These ideas can ensure that you build meaningful community trust and that you get the social dividend for all of your ethical efforts.
But this is more than just simple metaphors on their own, or rather linguistic devices for that matter, similes, analogies, or what have you. All of these will compound together throughout a text, and in doing so, create a narrative. Metaphors are the building blocks of narrative. It is these narratives that are particularly persuasive and shape worldviews. But to be structurally sound, they need to have a cohesive and a familiar narrative logic.
What is a narrative logic? Narrative logic has two components. First and foremost, your metaphoric construction of an issue needs to adequately answer who is doing what, why, and how. Once you include all of those things, once you connect the dots, you’re able to establish cause and effect. X therefore Y. A because of B. X country introduced Y policy, therefore A company did B action. When you have all of these things, you activate a narrative space. A narrative space is a way of effectively structuring somebody’s vision of the world, framing the way that they reason through a problem.
We’ll come back to the idea of narrative logic in a bit because it’s of incredible importance, but I realise that up until now, things have been very theoretical. So I want to take you to a very tangible example.
This is one of my favourite little studies, done at Stanford about ten years ago, where 500 people were shown different newspaper reporting of crime in a local area. The stats were exactly the same for each group. The only thing that changed was the metaphoric construction of the problem. People who saw the beast narrative were seeing newspaper reports laced with expressions like, “this pack mentality is preying on the weak,” “they’re out there prowling for victims,” “we must tame our streets.” While the people exposed to the virus narrative were seeing things like, “this contagious behaviour has led to an outbreak of crime,” “we need to decontaminate it,” “we need to vaccinate our youth before it spreads any further.”
And you can see the different narrative logics that these construct. The beast narrative suggests that an individual aggressor is causing this harm, because they either choose to or they can’t help themselves. The natural consequence of that is to hunt or cage them. It’s very punitive. The virus narrative, on the other side, suggests that this illness is coming from the physical environment. No one’s choosing to get sick, but they nonetheless are. As a result, the natural response is to diagnose and treat the illness.
When they ran this study, they then asked all the folk who had read these media stories, “What do you think is the right answer here? What would you do to stop crime in this local community?” The people exposed to the beast narrative came to the view that if individuals were at fault, then the most natural conclusion is to increase law enforcement, to increase funding for police and prisons. It was very punitive. Meanwhile, folk who were exposed to the virus narrative were much more likely to come to the view that a social environmental factor was causing the increase in crime, and therefore social reform is the natural way of solving it, such as education and work to improve economic equity.
But this isn’t just something that exists in journal articles. It’s something that I saw time and time again through focus groups and representative public opinion surveys. Anchoring people in values, strategically choosing metaphors, and effectively creating a narrative logic that is both cohesive and a familiar leads to double-digit improvements in the way that people understand issues and what they think the right thing to do about them is.
One of the challenges, though, is that often people will describe this sort of work as framing, but there’s no unified definition of what that actually is, when a frame begins or ends, or the rules about what’s in it or what isn’t. And it makes it very hard to compare different research or different strategies.
This is something I’ve tried to address by developing a framework I call narrative frames. The typology builds upon a complex database of common metaphors throughout the English language. The people who have been building this over the last 30 years claim that it represents all of the metaphors in the English language. I don’t know if I would necessarily make that claim, but it’s definitely an attempt to be exhaustive. I then cross-reference that with a significant body of over a hundred different journal articles and other media commentary, which try to explore themes in frames.
So this is not an exhaustive list, but it does mean that we can start to create shared definitions and kind of get on with the job, rather than having to interpret every text individually. It also means that this methodology can be used by people from all sorts of backgrounds. Whether you’re a computer scientist, a policy expert, a journalist, or anyone else, you should be able to draw upon this and use it in your work. It is composed of 22 of these common narrative frames, which ultimately break down into 48 sub-narratives.
No one narrative is good, no one narrative is bad. The question is whether or not that narrative might reflect the values you’re hoping to communicate or entrench in your work, and whether or not it creates an accurate representation of the issue as you understand it.
Some are better suited to ideology or worldview discussions explaining the larger, bigger picture. Journey, natural world, game, war: these often come up when describing large complex systems and the way stories of our society come together. Others are better placed to talk about the mechanics or the processes within our laws or within the way the AI tools themselves work. Things like that include machine and view-camera. These describe specific cause-and-effect dynamics within a more specific context.
To look at how this might manifest, I found this example from the EU Parliament:
As the steady stream of AI progress surges forward, it erodes traditional boundaries. We must reinforce our society and prepare for a deluge of unintended consequences, just like we prepare ahead of natural disasters. We need to develop an emergency plan, inspect our safety systems, and if we have time, build new protective infrastructure.
When you start to break this down, you can see firstly there is an overt metaphor here: AI progress is a steady current. You then see what I describe as entailments, which are other expressions which aren’t inherently metaphoric, but they contribute to the metaphoric logic. They contribute to the narrative logic. So “surges forward,” “erodes,” “reinforce,” “prepare for a deluge.” These together with the metaphor reinforce the idea that some sort of weather event is happening and that that’s a bit of a natural disaster. Then underneath that, we have an overt analogy. Overt analogies are the things that people often recognise and appreciate. They’re the things that cut through into our consciousness, and it really reinforces the image here.
Returning to the principles of narrative logic, we can see something critical is missing. Who is causing this? Looking at this, you’re led to believe that maybe AI progress just woke up one day and decided to flood a valley. If we don’t identify who is causing a thing and who is being affected by a thing, we can’t really have a conversation about who may or may not be responsible for fixing it, or who may or may not see the benefits or harms.
A better message might be that if governments don’t move to shore up existing policies, the dam wall may well break, and in doing so, turn the now steady stream of AI progress into a surge which erodes traditional boundaries, and so on and so forth.
Which brings me to the first of two common failures. When you don’t complete the narrative logic, people will take shortcuts and come to conclusions that aren’t what you’re trying to communicate. Or your opponents, people with different agendas or different visions, will complete the narrative logic on your behalf, again leading to an impression that is not what you’re hoping to achieve.
This often manifests in two distinct common ways. Firstly, we don’t articulate what our vision is. What are the conditions for success, for victory, to win our prize? Or who is responsible for achieving it? It needs to be explicitly stated, and wasn’t in the example we just saw.
In this next example, which is a bit wordier but is still a condensed version from a well-known think tank and advocacy body, we see:
Social media rewrote the rules, taking every aspect of our society into its tentacles, holding us hostage. If you don’t have social media, you don’t exist. Now, behind the positive stories, we’ve got this weird creepy face again, hidden from view, then acting creepy at the journalists and trying to blackmail them. Behind that, another kind of monster. This monster is a set piece, because AI under the hood has grown, and this new monster is increasing its capabilities and may entangle itself with society again.
When we go through this, we see lots of examples of both entailment and metaphor, but more importantly, we see just lots and lots of different narrative frames being used. The earlier example was very cohesive. It really committed to the idea of a bad weather event. Here, there are lots of competing logics, lots of competing concepts. Which leads to the second common failure: when you have competing logics, they prevent any cohesive narrative space or worldview from forming.
Now I want to talk to you about the findings from the major research project I mentioned earlier. Drawing upon over 300,000 different words in a giant sample across the EU, the UK, and the US, I’ve been able to map out what the common narrative frames are being used by executive political leaders. Here we’re thinking about folk like the King or Rishi at the time, like Biden and Harris, or like von der Leyen in the EU.
We’ve also got legislative political actors. Here we’re thinking backbenchers or Congresspeople sitting on committees. A large body of this sample comes from committee meetings, which are ideal little universes because they have fairly frank exchanges and you get a good representation of all the different stakeholders that are concerned about an issue.
I’ve broken them up by advocates, academics, the civil service, big tech and industry, and also the public, where I’ve drawn upon a large sample of social media comments in response to the proceedings above. I can break these down by jurisdiction and sector if you are curious and want to reach out to me to do so, but I want to talk about the different dynamics that emerge.
Our executive political leaders are often using building or journey or war. (Don’t worry, I’ll talk in a bit of detail about what these actually are in a few moments.) Legislative political figures are using war, journey, or machine. Advocates and academics are using war, journey, ecosystem, building, and a bunch of other things. Academics and advocates use a huge crate of different metaphors. Civil service are fairly persistently using journey. Big tech and industry are fairly persistently using race. They’re often paired with substance or car, which, as I mentioned before, are useful to explain different facets of a larger concept. So you’ve got an ideological one and a more mechanical operational metaphor being paired together there. The public also use lots and lots of different metaphors, particularly war, mythical, machine, and transaction.
One thing worth touching on is that in typical speech, as I’ve mentioned earlier in this talk, we grab a whole bunch of different metaphors. We’re really fluid. Even in this talk, I’ve no doubt used a whole bunch of different competing ones. So the example I showed before is really just reflecting typical human speech. And that is also what we’re seeing in some of these groups.
The difference, though, is that civil service and big tech and industry speakers are extremely deliberate and intentional with the metaphors that they use, with the narrative frames that they use. There isn’t a lot of deviation. They’re very consistent, which suggests that they’re being chosen quite strategically, quite deliberately.
Let’s begin by looking at one up close, and I’m going to begin with Journey-Race. As we saw earlier, this is very popular with industry, and indeed it’s very common, particularly in the US and the UK.
Journey-Race, you can see manifest in the comments of Dario Amodei when he was speaking to the US Judiciary Committee:
As companies race to be the first, testing and auditing should happen at regular checkpoints, appropriate guard rails should be put in place… progress is driven by others who are not likely to slow down anytime soon.
We can also see it from other examples in the corpus, such as “in the sprint for market dominance,” “we need to overcome obstacles to innovation,” “we’re leading the pack in adopting AI technologies,” and “we’re on the fast track to AI compliance.”
It constructs a narrative space that has a number of rules that guide what we think might be possible, who’s responsible, or what might go wrong. Most importantly, it suggests that there is a shiny prize that we all want to pursue, and that this prize is an inherent moral good. It’s something that’s great. Obviously we want to win it. It also suggests that the race is already happening, and in that respect it’s very techno-determinist, particularly when you’re talking about AI.
It lets you have a conversation about opponents. They’re there, you need to keep moving, but it doesn’t really matter who they are. They’re just looming in the background. It lets you have a conversation about risks and rules, but those rules aren’t really there to determine whether we should be running the race at all, or perhaps if we should be doing something else, like spending time with our children. Those rules exist to ensure the moral legitimacy of the inevitable victor. They’re there to ensure that people don’t cut corners, they don’t dope, they don’t do anything else too creative.
It also pacifies and sidelines the public, suggesting that there is a champion who, due to their strength and fitness and experience, is the one who we as a group, as an in-group, have decided to nominate, will be the one who competes on our behalf. We want to see them succeed because their success is ours. And indeed, if we’re not on the team and we’re not there encouraging them along, we’re a bit shit, we’re not a team player, we’re not doing the right thing.
The next frame that is particularly relevant is Journey. Journey is the parent frame of Journey-Race and Journey-Quest for that matter, but it’s a bit more general. It’s less prescriptive about how it works. Merely, you need to be going from a place to another place with kind of purpose. And it is particularly common with the civil service.
We can see this manifest in the United Kingdom where Lindsay Chiswick, the director of intelligence for the Metropolitan Police, spoke saying:
Technology only takes you so far. It takes you to the point of potential identification. You’ve got to point them in the right direction… We’re there in the first place. We must turn things around. We’re approaching deployment carefully. We’re on a journey at the moment. It’s a lot of work to get us there in the right way.
We’ve all heard thousands of examples of this, typically from public servants. We’re on the path to success. We’re taking steps to address that problem. We’re navigating some challenges. We’re on the road to resolving this problem. We’re going to reach the next milestone any minute.
The problem with this is that by not articulating where we’re going, you’re not able to hold anyone accountable. You’re also not able to buy into their vision. It just creates a sense of momentum, but with no sense of purpose. It also doesn’t communicate where we might be along that path, or who might even be on the journey with us, or how long we’ve even been walking.
We’ve also seen another problematic pattern with this. It isn’t just common with the public service, because it’s often used by people in vice industries or other contexts to suggest progress is happening, but really to defend the status quo. There’s a significant body of climate communications research that says it’s the journey frame that is particularly popular and common with big polluters. They use it in a way to suggest that, oh yes, it’s very important that we move towards renewable technologies and we’re taking steps to do that. We’re heading in the right direction, as a way of essentially justifying the status quo.
I mentioned the risk of compliance theatre starting to emerge. I think that we need to pay really close attention to Journey, because as that plays out, this will almost certainly be the narrative frame used by people hoping to avoid any changes to their behaviour.
The next narrative frame which is particularly common and particularly powerful is war. You’d be forgiven for thinking that perhaps the war frame is especially common in the United States, and it certainly is very common in the United States, but it’s also very popular in the EU and in the UK and indeed anecdotally, I would say, Australia.
It manifests in a number of ways that, once you hear it, you can see it:
It’s exploding, leading to commercial invasions of privacy for profit. We’ve fallen victim to an algorithm. There’s nothing to do with liberating people. These threats are like the existential threats of AI because they could endanger the very existence of humanity.
These are from two different speakers in front of the EU’s special committee on AI. We also see that in other examples across the corpus, such as “we’re on the front lines of AI ethics,” “we’ve got to marshal our efforts and combat misinformation,” “this is a battleground of data privacy,” and “unregulated AI models are a ticking time bomb.”
War sets up a narrative space that is highly oppositional. At first it might seem like a strategic choice for a lot who are using it. It suggests that there are some core principles and values that must be protected, such as democracy or fairness or transparency. It implies that these are currently under threat, and that no one person can solve this on their own. Government, industry, small business, big business, civil society, your local community and mine need to come together to protect these things that we all communally hold dear.
But the problem with that is that it needs to have a baddie that is of such grave power and of such direct threat to those principles that it justifies a war footing. For the war to be well motivated and just, you need to have Sauron, you need to have Voldemort, you need to have Nazis.
And the problem we see in the data set is that advocates don’t have any shared sense of who this baddie is. Some of them will refer to industry: they’ll say, “Oh, Mark Zuckerberg’s bad, so you can’t trust him, we need to go to war,” or that Sam Altman’s terrible and we need to do something about him, or that big companies generally are bad. But the overwhelming majority of people in the United States, as we see through public opinion polling, and frankly also in the UK and in the EU, don’t necessarily feel like they are inherently morally evil characters. There’s significant distrust, there’s a sense that they don’t act in the interests of the community, but they’re not so evil that it warrants a war footing. Some portions of the community might feel that way, but it is not a widely held view.
So it doesn’t quite land, it doesn’t activate a narrative logic, and so the middle ground of the community keeps going through their list to think, “Well, this is a powerful narrative frame. How do I complete the narrative logic?” They go to the next thing on the list. Many advocates say, “Well, AI is the problem. AI is going to take your job. AI is stealing your copyrighted material.” And that has a degree of power. I get the impression that that was kind of where a lot of the “AI robots are going to kill us and take our jobs” anxiety came from. But as people have become more familiar with the models, that seems to have softened. People realise that these are abstract things, they don’t have personhood or agency in the way that you would need to justify this framing.
And so as people keep going down the list, they end up reaching the first actor in the public debate who could theoretically wage a war, who could theoretically be opposed to those core values, and could motivate a just war response. That’s how you end up with people thinking that maybe China is bad, and therefore something that we should take action about. I’ll come back to this because it’s a very important dynamic, but I won’t dwell on it quite yet.
Another really important dimension of the War frame is that in order for it to activate a narrative logic, it needs to have a clear condition for success. What does victory look like? How will we know when we’ve won? A big limitation of many of the advocates is that they don’t name this. They don’t say what their positive vision is or what the ideal circumstances are, merely that there’s a threat and we need to freak out about it.
War framing can be very effective at motivating a sense of solidarity and purpose. It can really powerfully shift people’s minds. But a significant body of climate communications research has revealed that when advocates don’t clearly outline what the conditions for success are, after 12 to 16 months the support and buy-in that you’re able to create dissipates. Not only does it dissipate, but trust in you and concern about the issue significantly decline, lower in fact than the baseline of where you were beforehand. So you can end up accidentally being the boy who cried wolf. It’s incredibly important to outline what you want to see. What does success look like?
The next frame I want to speak about is Building. Building is an extremely versatile frame that you can see represented in a number of different little manifestations:
By building on the G7 Hiroshima process and the global partnership on AI, we’ve reached a landmark agreement. We are laying the groundwork for success by building the foundations of the EU AI office. This sheltered situation risks creating siloed oversight. I’d rather see the risks earlier, so we’re able to close the barn doors before the horses bolt rather than after. It risks being the ceiling of protection, not the floor, locking in many unfortunate decisions.
We can also see that in other examples too, like “the blueprint for success,” “the cornerstone of our approach,” “we’re scaffolding a new strategy,” “sales are going through the roof,” and “society might collapse.”
It is extremely sophisticated and probably up there with war and journey, particularly Journey-Race, as having a really powerful ideology. It suggests that there is a desirable vision that we want to work towards, and that that vision has utility, often utility for you personally and us as a community. It suggests that we get there, we achieve that utility, we achieve that vision, often by working together. No one person has all of the skills to get there. It also suggests that we should be thoughtful and intentional from the start. We need a blueprint, we need strong foundations, and if you cut corners, if you rush, things don’t go that well.
It also suggests that it might degrade over time, and that we might need to do some renovations to make sure it remains fit for purpose. It also suggests that there may be an inside and an outside. We might want to close the barn doors, or unlock our future. We might want to keep out risks.
I’m going to take us back to the board and look at how these dynamics might interplay now that we have a better sense of what they are.
Industry is very powerfully using the Journey-Race narrative to suggest that there’s something really special, and we’ve got to get a move on to get there, and they have successfully completed the narrative logic to help people understand that. Meanwhile over here, advocates and academics are using war to suggest, “Wow, this is really high stakes. We should take this really seriously, and in the process, make sure we’re protecting some fundamental principles.”
The public think, “Holy crap, that sounds reasonable, and I’m a bit scared. I’m a bit thrown by this now.” But because advocates and academics haven’t successfully outlined the narrative logic, they haven’t successfully structured the narrative space for the public to comprehend what they’re suggesting. You just end up with the public freaking out: holy crap, we’re under siege, we’re under threat here, but I don’t know what success looks like. I don’t know what the positive vision we need to be moving towards is. I don’t know who’s a friend or a foe at a point that they would accept and believe.
And at the point they end up inevitably thinking, well, it must be some sort of foreign state actor, it ends up actually driving them right back into thinking that maybe big tech’s right. If we are in a race with a foreign actor, maybe we want our guys to win. Like, the whole idea is that they’re our champion and we’re there to cheer them on. So I guess we should be doing that.
In this process, advocates and academics in their attempt at raising the stakes succeed. But they fail to construct a narrative logic that meaningfully persuades or convinces anyone. And in that failure, they drive the public to embrace and share industry’s worldview. They undermine their own strategic objectives because of their incomplete narrative logic, and I would frankly also say, poor narrative frame choice.
Another dimension that I really want to speak about is the prominence of journey with the civil service and Journey-Race when we’re looking at industry, particularly because these are commonly being used in committee meetings before the legislative political actors who are trying to make decisions around what should our priorities be and what are the risks and the trade-offs. Because they’re using very, very similar frames, using very similar language, but because the civil service are not successfully establishing their narrative logic, but industry is, with an incredible degree of sophistication, you end up with legislative political figures looking down at both of these characters and thinking, they sure sound like they’re saying pretty much the same thing.
I’m sure many civil service listeners would think, well, we’re definitely not, actually. We think that we should be having some more structure and accountability, and that we actually probably need a lot more work on governance. But it doesn’t look like that, because we’re all just sort of making progress to some sort of goal, and the legislative political figures are thinking, well, what goal would that be? And industry is sitting right next to them saying, well, we know what goal that is. It’s the one we’ve just told you. Both of which leave them looking pretty much the same.
That’s all very interesting and revealing, but to conclude, I want to leave on a few points to make sure that you’re able to use this work to ensure that you’re more accurately communicating your understanding of the issue and how it all comes together.
As a quick recap, metaphor and figurative language fundamentally shape reasoning. Metaphor and figurative language are the building blocks of narrative. And narrative is important because it tells us who is involved in a story, where we might be going, and what is good or bad. When we answer all of those questions in a cohesive and a familiar way, we create a narrative space. A narrative space is what shapes our understanding of the world.
The two biggest failures I often see with people trying to construct a narrative space is that we either don’t outline our destination, the prize we’re pursuing, or the conditions for victory. How will we know when we’ve won or we’ve succeeded? We need to have a positive vision that we are appealing for people to achieve. Secondly, we also don’t often outline who the actors or the characters are: who is our friend, who might not be our friend, who has a different agenda and why. We need to clearly articulate who is a part of this story.
When we do that, and then we do it with consistency and discipline, both through a text and also over text for an extended period of time, each thing that we produce over months, we’re able to successfully shift the way people understand how things work and the values that are implicit or important in those tools.
Narrative frames as a methodology are a really, really important tool to help us achieve this. Because it lets us understand and have a frank conversation about which narrative best suits our values, which narrative best reflects the way that we do understand these tools to work. It also lets us identify what our opponents or our stakeholders or our friends are using. What are our competitors using? What are they trying to communicate? Are they succeeding or are they failing? And importantly, and this is critical: how does their narrative intersect or interplay with ours? It’s only by mapping this out that we can start to have really, really powerful and effective messaging.
If you want to see an AI future that is ethical, open, accountable, and transparent, where technology is used to build a more just and equitable society where everybody can thrive, then we need to make sure that the language we’re using cognitively reinforces these ideas and these values.
The future of AI is being written. Make sure you’re holding the pen.
Thank you.
Working on AI policy?
The question is whether they help your case, or quietly undermine it.
Whether you’re shifting public opinion, briefing a minister, or building a coalition, this is where to start.
Start a conversation
The methodology behind this research
This study applies the peer-reviewed Narrative Frames methodology, developed at the University of Cambridge, for decoding the metaphorical reasoning structures that determine whether communications on complex issues actually land.