On September 13th, Sweden held its general election. Among the usual candidates, there was a new party: RegleraAI.nu (RegulateAI.now). This party was started only four weeks before the election, mostly out of pure frustration. AI is advancing at a staggering pace, and with it comes large societal transformations and even the risk of near-term human extinction. Despite this, half of the eight parties in the parliament did not mention AI even once in their election manifestos, and those who did made it painfully obvious that no one realized the gravity of the situation.
In Europe, there’s a strong precedent for refocusing the political discourse through the creation of single-issue parties. Examples include the pirate parties for copyright and privacy, the green parties for climate change, and the more recent wave of anti-immigration right-wing parties. In Sweden’s 2014 election, we even had a single-issue party around feminism. They didn’t make the national parliament, but everyone was talking about it, and it forced other parties to take the question seriously. We figured we could do the same.
We decided to almost exclusively focus on the prevention of existential risk. The party’s main demand is that Sweden should do everything in its power to pressure and encourage the US and China to agree on a pause deal. This may seem feeble, and we won’t deny that Sweden’s ability to affect such a deal might be small, but we don’t think it’s zero. By pushing hard enough within institutions like the EU and NATO, there should be enough leverage to move the needle in the right direction, even if just by a little bit.
So we got to work and ordered 200,000 ballot papers. As a new party, distributing these to the polling stations was our own responsibility. Thankfully, our numbers grew quite quickly, and it turns out that you can get really far with a couple dozen driven people. We managed to spread the ballot papers all across the country, with especially good coverage in larger urban areas. Our idea was that even if people didn’t vote for us, they’d see our logo, which was our website URL. This worked, and our website got thousands of visits. In addition, there’s some newsworthiness in a new party, and we managed to get a few articles in national and local newspapers.
Still, it’s hard to estimate the impact we had. Jacob Coxon’s heavily publicized departure from Anthropic happened right around the election. Swedish media picked this up fairly well. We probably played a role in that, but we don’t know how big. What we do know is that 390 people chose to vote for us. The bar for “wasting” a vote on a party with an almost-zero chance of making the parliament is quite high, so this points to a far higher total number of impressions. Even so, we obviously wish we could have reached further, and AI safety never fully broke into the political discourse as we had hoped.
Should this be replicated?
Starting a party isn’t risk free. We see it as a tradeoff between bringing attention to the problem and introducing a risk of polarization. We cannot afford this issue being buried in political conflict. A terrible outcome would be if AI safety followed the same tragic path as climate change and established a home for itself on one side of the existing political spectrum. For now, this does not seem to be happening, in Sweden at least. We were also careful to point out that we had no ties to any other parties and had no place on any existing spectrum.
With this risk in mind, we believe that this could be successfully replicated across other countries with multi-party systems. New attempts would ideally be launched more than four weeks ahead of election day, and with some more time and better planning, the reach could be much greater. People deserve to know what’s happening, and a political party can be a great way to move the Overton window while channeling latent commitment towards something good.
Comments and new perspectives are highly appreciated.
On September 13th, Sweden held its general election. Among the usual candidates, there was a new party: RegleraAI.nu (RegulateAI.now). This party was started only four weeks before the election, mostly out of pure frustration. AI is advancing at a staggering pace, and with it comes large societal transformations and even the risk of near-term human extinction. Despite this, half of the eight parties in the parliament did not mention AI even once in their election manifestos, and those who did made it painfully obvious that no one realized the gravity of the situation.
In Europe, there’s a strong precedent for refocusing the political discourse through the creation of single-issue parties. Examples include the pirate parties for copyright and privacy, the green parties for climate change, and the more recent wave of anti-immigration right-wing parties. In Sweden’s 2014 election, we even had a single-issue party around feminism. They didn’t make the national parliament, but everyone was talking about it, and it forced other parties to take the question seriously. We figured we could do the same.
We decided to almost exclusively focus on the prevention of existential risk. The party’s main demand is that Sweden should do everything in its power to pressure and encourage the US and China to agree on a pause deal. This may seem feeble, and we won’t deny that Sweden’s ability to affect such a deal might be small, but we don’t think it’s zero. By pushing hard enough within institutions like the EU and NATO, there should be enough leverage to move the needle in the right direction, even if just by a little bit.
So we got to work and ordered 200,000 ballot papers. As a new party, distributing these to the polling stations was our own responsibility. Thankfully, our numbers grew quite quickly, and it turns out that you can get really far with a couple dozen driven people. We managed to spread the ballot papers all across the country, with especially good coverage in larger urban areas. Our idea was that even if people didn’t vote for us, they’d see our logo, which was our website URL. This worked, and our website got thousands of visits. In addition, there’s some newsworthiness in a new party, and we managed to get a few articles in national and local newspapers.
Still, it’s hard to estimate the impact we had. Jacob Coxon’s heavily publicized departure from Anthropic happened right around the election. Swedish media picked this up fairly well. We probably played a role in that, but we don’t know how big. What we do know is that 390 people chose to vote for us. The bar for “wasting” a vote on a party with an almost-zero chance of making the parliament is quite high, so this points to a far higher total number of impressions. Even so, we obviously wish we could have reached further, and AI safety never fully broke into the political discourse as we had hoped.
Should this be replicated?
Starting a party isn’t risk free. We see it as a tradeoff between bringing attention to the problem and introducing a risk of polarization. We cannot afford this issue being buried in political conflict. A terrible outcome would be if AI safety followed the same tragic path as climate change and established a home for itself on one side of the existing political spectrum. For now, this does not seem to be happening, in Sweden at least. We were also careful to point out that we had no ties to any other parties and had no place on any existing spectrum.
With this risk in mind, we believe that this could be successfully replicated across other countries with multi-party systems. New attempts would ideally be launched more than four weeks ahead of election day, and with some more time and better planning, the reach could be much greater. People deserve to know what’s happening, and a political party can be a great way to move the Overton window while channeling latent commitment towards something good.
Comments and new perspectives are highly appreciated.