Voting in Switzerland takes place every three months. Fierce debates take place before referendums, and the tone can be particularly aggressive online. Insults, outright hatred and even threats of murder are not uncommon. This is a risk for democracy, says Sophie Achermann, director of Alliance F, the largest umbrella group representing women in Switzerland.
“It’s important to have tough but fact-based discussions,” says Achermann. “But hate on the internet prevents diversity of opinion. People are afraid of hate mail and prefer not to say anything. Ahead of the vote on the pesticide and clean water initiative in 2021, for example, some politicians so much hate mail that they didn’t want to appear in public, she says. It’s not just a Swiss phenomenon: Around the world, politicians are facing growing online hostility and threats, especially politicians. womenexternal link and minorities.
For this reason, Alliance F has developed an algorithm against hate speech. The algorithm is called “Bot Dogexternal link”, because – like a dog – it detects hateful messages on social networks and flags these messages. A group of volunteers then responds to each message. The idea is that hate on the internet should not go unchallenged and discussion can continue on a factual basis.
Bot Dog is still in the pilot stage. But his first attempts were successful: Researchers from the Federal Institute of Technology ETH Zurich and the University of Zurich monitored the pilot project and found that responses calling for sympathy for people who were victims of bullying speech hatred were particularly successful. Phrases such as “your post is very hurtful to Jews” led to the hateful speaker either apologizing or deleting the post.
In July, Bot Dog will make its official online debut. Anyone who wishes can participate in the project, says Achermann. Either by rating comments and helping the machine learning algorithm more accurately identify comments that contain hate speech, or by responding to posts marked as hate speech.
Science in Switzerland: women agents of change
We introduce you to some of the pioneering women who are making a valuable contribution to scientific research in Switzerland.
Is artificial intelligence really as smart as we think?
This content was posted on Feb 15. 2022 Feb 15 2022 Computers are making more and more important decisions for us. Should we leave them?
Cooperation between algorithms and people
The Bot Dog is just one example of how artificial intelligence (AI) can strengthen democracy. Dirk Helbing, professor of computational social sciences at ETH Zurich, sees immense potential in a digital democracy. He says he envisions participatory budgeting, where citizens can determine together how their city or community’s budget should be spent.
In the Zurich district of Wipkingen, this idea is already being tested. Using a digital platform, residents opted for city parks, a skate park and a street-food festival. Each of these projects is supported by 40,000 CHF ($40,250) of taxpayers’ money.
Another idea for digital democracy is that cities and regions around the world could join together to participate in competitions for sustainable economies, CO2 reduction or peaceful coexistence – a kind of city olympiad. The solutions and the associated data would be open source, therefore freely accessible to everyone, and could serve as the basis for AI models. “I think we’re moving towards a kind of participatory digital society,” says Helbing – a cooperation between AI and people. , as is the case with Bot Dog, where an algorithm and volunteers join forces to fight online hate.
AI has enormous potential for direct democracy, he says. But data-driven apps “also have enormous destructive potential for democratic societies based on the rule of law and human rights,” says Helbing.
Anticipate the government by smartphone
This content was published on Oct 27, 2017 Oct 27, 2017 Two new books examine how countries around the world, including Switzerland, are adapting to the demands of the digital age.
Not all good
Algorithms increasingly determine what information we receive, what products are presented to us and at what price. They determine what we see of the world and what issues we consider important. This increasingly guides our political thinking and voting behavior, says Helbing.
This was evident during Donald Trump’s presidential campaign in 2016, where his team used big data algorithms to create personalized content for his target groups. wanted it. Until they vote for our candidate,” is how Brittany Kaiser described her campaign work in the Netflix documentary “The Great Hack.” Kaiser worked for Cambridge Analytica, the company that dug into the private data of 87 million Facebook profiles before the US presidential election, sparking one of the biggest data scandals to date.
Even during Barack Obama’s second presidential campaign in 2012, voters were targeted with messages. Obama’s team collected as much data as possible and sent each voter a personalized message.
It is impossible to say with certainty the extent to which AI campaign methods contributed to the success of the candidates. But Helbing and his colleagues see great danger in these methods. Crucially, the combination of microtargeting and nudges with big data about our behavior, feelings, and interests could result in totalitarian power, Helbing says. Nudge is a psychological term and means to push people for a specific purpose or in a particular direction.
AI voting campaigns in Switzerland
AI needs one thing above all else: data. And the more there are, the better it works. In Switzerland, data protection and privacy are highly valued. Would campaign methods such as those used in the United States be possible here? Could Swiss parties get their message across to voters using so-called micro-targeting?
Lucas Leemann, whose work at the University of Zurich includes research into how public opinion can be measured using machine learning, doesn’t think that will happen.
He says that the situation in Switzerland is not comparable to that of the United States. This, he says, is partly because the kind of raw data used by the Trump and Obama campaigns is not available in Switzerland. “They had data points for almost every US citizen,” he says. “It’s completely different in Switzerland. ”
During his studies, Leemann briefly worked for a fashion company in the United States and not only had access to the database of customers – therefore name, address and date of birth – but also could see estimates of their annual income, how many children they had, what car they drove and whether they rented or owned their own home.
“In the United States, you can easily buy this data and it is also used for political purposes,” he says. “In Switzerland, as far as I know, this type of data is not used for political campaigns, or at least not. Again.”
Is “not yet” the key? Will this data soon also be on sale in Switzerland? According to Helbing of ETH Zurich, the use of data around the world is far more advanced than most people realize.
As an example, he cites the World Economic Forum’s (WEF) Center for Cybersecurity, headquartered in Geneva, which involves organizations and companies from around the world, including Amazon, MasterCard and Huawei Technologies. Helbing says that “a huge amount of data is collected there”. The data collected is used – with the blessing of the United Nations – for purposes such as the implementation of the 2030 Agenda for Sustainable Development.
Sign up for our free newsletters and get the best stories delivered to your inbox.
“It may be well-intentioned, but it begs the question of how this is implemented and what might be possible with all this data,” says Helbing. “As so many companies are involved, there is a danger that commercial interests could trump social interests.
Experts warn that laws and regulations must be created now to prevent AI from becoming a danger to democracy.
But to create operating rules, it is necessary to know exactly how the algorithms work. It is a problem. Platforms such as Facebook are a black box, says Anna Mätzener, director of AlgorithmWatch Schweiz. “We don’t know, in detail, how the algorithms work,” she says. “We also don’t know exactly what data is being collected about whom.”
The AI curating content on social media is a closely guarded secret. To find out how algorithms affect election campaigns, AlgorithmWatch has launched a research projectExternal link in cooperation with the newspaper Süddeutsche Zeitung. They called on hundreds of volunteers to donate data from their Instagram timelines ahead of the 2021 German elections. Volunteers were asked to subscribe to various party profiles. A browser plug-in recorded where and when content from these profiles appeared in users’ news feeds and passed this information to AlgorithmWatch.
The analysis showed that content from the far-right AfD (Alternative for Germany) ranked higher in Instagram’s news feed than content from other parties. Research could not determine why this is the case.
For Meta (formerly Facebook), owner of Instagram, this research was clearly unpleasant: the company threatened AlgorithmWatch with “more formal measures” if the project was not abandoned. As a result, data donations ended prematurely.
“Until this type of research is possible, we cannot say anything factual about the influence that AI-curated content on platforms has on society, particularly on the formation of political opinion. , and therefore directly on democracy,” says Mätzener.
Mussolini’s Swiss doctorate allowed to live
This content was published on Jun 24, 2022 Jun 24, 2022 University ignores requests to rescind honorary doctorate awarded to former Italian dictator Benito Mussolini.
Edited by Sabrina Weiss