This Algorithm Tells Us How World Leaders Are Reacting to Covid-19
Let’s play a game. I’ll give you two facts and one activity. Don’t scroll down the page to cheat!
Fact One: There is a facial expression analysis algorithm called Affectiva — it is a human perception AI which analyzes complex human states — based on a machine learning algorithm, it has analysed more than 9,258,714 faces.
Fact Two: This AI algorithm (which we will call face expression analysis) has been applied to three world leaders during their respective state of the nation addresses, which all three leaders made in the past week in regards to how their government will tackle Covid-19 in their respective countries.
The game: Look at the graphs below, which used the facial expression analysis AI, to analyse their facial expressions throughout their speeches addressing their nations. Then guess which graph of expressions corresponds to which leader.
Tip: Pay attention to the emotions of anger (brown line — the top row of each graph), contempt (the yellow line in the second row), disgust (the deep blue line in the third row of each graph), fear (the green line in the fourth row) and finally sadness ( the mustard line in the fifth row). On some phones, you may have difficulty seeing the graph clearly.
What world leaders did you guess each graph belonged to? Let’s compare the emotions more closely in the following table:*
*A note here, the results on the graphs above and the table differ slightly, as the table shows validated results, that is emotions that passed a threshold, whereas the colourful graphs above only visually show the emotions prior to validation.
You can see from the table that the leader in graph A spent 19% of the speech displaying anger — which was the main emotion displayed throughout their speech. 93% of their speech was in neutral affect, what that means is the leader’s face was mostly absent of emotion.
Graph B shows a leader who spend 29.56% of their address in anger. They also had displays of disgust, for 2.11% of their speech, with no joy or fear, and not positive affect at all. They had neutral affect for 95% of their speech.
Finally, graph C shows a very different result. It shows a leader who only spent 3.66% of their speech displaying anger, interestingly, they also showed the most significant amount of sadness at 3,91%. In the contrast with the other leaders— they showed joy for 15.63% of their speech. They also showed surprise for 16.4% of their speech. They were the only leader who had positive affect in a significant level throughout their speech, with 17% of their speech spent displaying positive affect, and 10% showing negative affect (such as sadness, anger etc), and the lowest neutrality.
What does this mean?
Well, when we watch a leader speak to us, in any circumstance, but especially when we are facing such uncertain times, we want to see leaders who exhibit humanistic traits. That is — we want leaders who also show sadness when something is sad (e.g. death and a pandemic), joy or smiling to reassure us things will get better, surprise that things have changed this quickly — and even anger that some people aren’t taking this seriously.
What happens when we don’t see a range of emotions? We feel polarised by our leaders. When a dominant emotion during a speech is anger — it is unsettling. Social contagion occurs, even through broadcasted speeches, and a population feel angry too.
So who were the leaders? Graph A was Boris Johnson (UK), Graph B was Donald Trump (USA) and Graph C was Jacinda Ardern (NZ). Let’s see more up close:
What does this mean for the management of Covid-19?
Well, we can see in the policies that these three leaders have put in place, that a lack of sadness (i.e. empathy), joy or smiling (i.e. reassuring a population), and fear (i.e. that the leader themselves is afraid and taking this seriously) can actually show some fundamental differences in how crises are handled.
Let’s compare the results of this facial expression analysis with the other objective measures of pandemic response.
- New Zealand has released a stimulus package equating to 4% of its GDP for the Covid-19 response. It includes covering wages for people who are required to self-isolate but cannot work from home, or those caring for relatives who are sick with the virus, even if they are not sick or do not test positive for Covid-19. “This package is one of the largest in the world on a per capita basis,” Grant Robertson, the finance minister, told reporters at New Zealand’s parliament on Tuesday. Their package has particularly catered for self-employed people, making it easy for people to get quick access to money if they have lost 30% or more of their income. They currently have 66 cases of confirmed Covid-19.
- The UK has announced a £30bn ($39bn) aid package which equates to 1.3% of their GDP, a giveaway for the coming year, made up of £12bn of immediate virus-related spending and £18bn of other measures. Some have criticised their aid package for being too little too late as many self-employed people are already out of work and struggling to meet their bills. They currently have 5,683 cases of covid-19.
- At the time of writing this, the US government is still confirming their aid package, with some hoping it could be over two trillion, but others criticising the government for the slow action. They currently have 28,521 confirmed cases of covid-19.
So, can this algorithm tell us how well national leaders are handling the Covid-19 pandemic? To some extent, you could say yes. There is at least some correlation between humanistic emotional displays and quick response and high GDP aid package.
Most of all — keep safe, wash your hands, stay at home & feel your emotions, we are all only human after all!
Connect with Nicola on Instagram, or for long-form exploration of ideas and feelings sign up to Nicola’s monthly newsletter, Et cetera, here.
