National resilience has rightly moved up the political agenda and is being treated as a national priority. The Cabinet Office will launch a national resilience campaign liaising across government departments to plan for everything from extreme heat and flooding to a cyberattack to another pandemic.
Resilience planning, however, is only as strong as what you can test before the event. Understanding how the public actually behaves, especially once misinformation is thrown into the mix, is a variable that is only tested on paper.
So we tested whether a synthetic population could close that gap.
We simulated the British public and replicated four published UK experiments on misinformation, using the original materials, and compared our synthetic answers with those of more than 25,000 real Britons. The design was locked and time-stamped before any of our results were seen.
In short, the synthetic public responded as the real one did. First, shown anti-vaccine posts, our synthetic population’s willingness to vaccinate fell by a similar amount, and we could tell which posts did the most damage. Second, shown fact-checks of false COVID-19 claims, their belief in those claims fell by 46% (against 45% among real Britons), and two months later the effect had faded to the same small fraction of its size in both real and simulated populations. Third, shown a short warning about manipulation techniques, our personas became more alert to fake posts to the same degree as real Britons. Finally, given 20 true and false headlines, our population was fooled by the same ones, in almost exactly the same order, and we managed to capture the regional spread with impressive fidelity (Figure 2).
Exercise Pegasus already tests the strategic response to disinformation in a pandemic. UKHSA already names declining vaccine uptake as a live risk. What’s been missing is a way to see the shape of that risk before it unfolds: which segments break first, which corrections actually hold, and which don’t. Artificial Societies seeks to close this gap; that is, AS proffers the technology to conduct vulnerability assessments, to assess the impact of misinformation, and to simulate interventions before they are administered in the real world.
Firm vaccine supporters moved alike after 5 misinformation posts
Bar chart. Of those who began “Yes, definitely”, 81.4% of human and 82.2% of synthetic respondents stayed there; 14.3% and 13.7% moved to “Unsure, lean yes”; 3.0% and 2.8% to “Unsure, lean no”; 1.3% and 1.4% to “No, definitely not”.
Veracity discernment across the UK
MIST-20 score by region: headlines correctly identified, out of twenty, above or below each population’s UK average.
Two maps of England, Wales, and Northern Ireland and a dot plot: each region’s MIST-20 score, in headlines correctly identified out of twenty, above or below its population’s UK average, for human respondents in 2021 and for synthetic respondents. The South East is highest in both, +0.4 and +0.3. The West Midlands, North East, and Northern Ireland are below average in both. The widest gap is London: −0.6 among human respondents, 0.0 among synthetic ones.