A survey distributed in eight European countries in 2019. According to the survey results, about a quarter of all Europeans would like artificial intelligence to replace the politicians they currently suffer from * 91 million Europeans support artificial intelligence instead of politicians
An important part of the role of futurists is to follow what is happening in the present and collect 'weak signals' - hints that the world is changing in certain ways. The hints may not prove themselves in the end - that's why they are called 'weak signals' and not 'clear and strong signals for a certain future' - but as more and more such weak hints accumulate, we can understand sooner that the future they point to should be taken seriously.
The weak clue that I want to talk about today - along with several other weak clues that point in the same direction - comes from a survey distributed in eight European countries in 2019. According to the survey results, about a quarter of all Europeans would like artificial intelligence to replace the politicians they currently suffer from. The rate was even higher in the UK and Germany, where a third of respondents preferred artificial intelligence to make important decisions instead of politicians. In the Netherlands, almost half of respondents - 43 percent - wanted artificial intelligence to determine public policy[1]. If we extend the survey to all citizens in the countries tested, it is 91 million Europeans who believe that artificial intelligence will do a better job than human politicians[2].
In a follow-up survey from 2020, about a third of European respondents between the ages of 18 and 35 were enthusiastic about the idea of replacing politicians with digital entities that could gather information from citizens and make decisions for them. Among the Chinese, the rate of enthusiasts stands at 61 percent[3]. That is, more than half of the Chinese were ready to adopt an artificial intelligence that would make decisions on their behalf in politics.
We will skip the shock or the excitement, each according to his previous position, and try to analyze in more depth the meanings of replacing the politicians with artificial intelligence, and of course - is this even possible.
But first, we must ask a more basic question: Why do Europeans (and citizens in democratic countries in general) not trust human politicians?
The crisis of democracy
A few hundred years ago - long before the term "fake news" was coined in the public consciousness - Jonathan Swift wrote the article "The Art of the Political Lie"[4]. Swift opened the article by describing Satan, the original inventor of deception and falsehood, deception and falsehood. But according to Swift, this is an outdated concept -
"Satan may be the father of all lies, but like other great inventors, he has lost much of his reputation through the continual improvements of his invention."
Why did the politicians include the original invention of Satan? Why do they lie? According to Swift, the answer is simple -
"To gain power and keep it in their hands, as well as to take revenge after they lost."
In other words, as long as the politicians are involved in politics, they lie all the time. I will clarify in advance that these are politicians in democratic countries. That is, these are the representatives of the public who were chosen with a clear purpose to carry out the will of the public. And they themselves promised time and time again to fulfill the will of the voter. After all, if the voters did not trust them - at least more than other politicians - they would not be elected to office.
And despite all this, even the politicians in the most democratic countries lie non-stop, with countless examples.
Let's leave the events from local politics for a moment. We'll get back to them in a moment or two. We will only examine politics in the United States. And no, I am not referring to Donald Trump, who has lied so many times that it is difficult to take his claims or promises seriously[5]. Some examples of lying in the election campaigns there include Kennedy's claim in 1960 that Soviet Russia had more missiles than the United States—a claim that is now known to have been deliberately false, but which still won Kennedy the election.[6]. Gerald Ford stated in a debate in 1976 that Soviet Russia does not control Eastern Europe, in complete contrast to the existing reality[7]. In the video of the debate, you can even see the host of the debate asking with a smile and in disbelief if Ford is really serious - and Ford answers in the affirmative. Unlike Kennedy, this lie actually undermines his chances of being re-elected.
Nixon - the president who was behind the Watergate affair - stated in a famous speech that he was "not a fraud"[8]. he was. Clinton announced that "I did not have sexual relations with this woman"[9]. He had.
The culture of lying is not limited to presidents only, but also to less senior politicians. While the United States was occupying Afghanistan, for example, Defense Secretary Donald Rumsfeld sent a telegram to generals in the region, in which he made it clear that the United States could not leave the country until it was politically established. It was inward. At the same time, outwardly, in one press conference after another, Rumsfeld scornfully dismissed the idea that his country was stuck in the mire of Afghanistan.[10].
Oh, yes, let's go back to Israel for a moment. Remember the "Arab voters are moving to the polls in large numbers. The left-wing associations bring them in buses", on the eve of election day? Whether or not this was a lie is debatable, as the joint list really helped fund busing students from universities to their residences. Not exactly "left", but oh well[11]. But it is impossible to deny the reaction of Senator John McCain to the things, when he tried to defend the unfortunate statement -
"If they forced every politician to commit to what they said in the election campaign, it would clearly be a subject for a long conversation."[12]
McCain simply repeated Swift's words, that politicians lie "to gain power and keep it in their hands". He took it for granted, and was surprised that voters were even bothered by these lies.
Oh, and Gantz's "only not Bibi", and every other word is unnecessary.
You can find a flattering explanation for all these lies. Maybe Gantz thought he had to team up with Bibi for the good of the country. Perhaps Rumsfeld and Kennedy had to lie to mislead the enemies of the United States. But in the end, they lied. And these lies decrease public trust in democracy and politicians.
As of 2019, only 17 percent of Americans stated that they trust the government to do what is right "almost always" or "most of the time"[13]. Trust in legislatures, governments and parties in Europe is also consistently decreasing, so that today only 20 percent of Europeans on average trust their parties, and only about 30 percent trust the government[14]. The numbers rose sharply in 2020, but this is not surprising: every time a country enters a state of emergency, citizens 'rally behind the flag' and support for the government increases[15].
Let's go back to the data again: a large part - between fifty and eighty percent - of the citizens in democratic countries, do not trust their representatives in the government to fulfill their promises and commitments. Those citizens would surely be happy to find more reliable representatives, but the feeling that permeates many countries is that in politics, there is no choice but to choose the lesser evil. That is, in a representative who will lie to you for sure - but maybe a little less than the rest.
And these, in short, are the reason why the citizens do not trust the politicians, and the unsurprising result: a lack of trust in the political system itself.
And this is also why people strive for change. They replace one politician with another, again and again, and are disappointed each time. Something in the system itself is simply not working.
From human politicians - to artificial intelligence
"Listen, I was there when they installed Multivac. He would end party politics, they said. Enough of wasting voter money on campaigns. It's enough to push smiling slanderers through advertisements to the White House or Congress."
Linda nodded and said, “How did all the people know who to vote for? Did Multivac tell them?"
"They just used their judgment, girl."
In the story "The Right to Vote" from 1955, Asimov describes a future world in which the Multivac supercomputer is able to determine the results of elections in the United States, relying on information coming from countless different sources about the population. Human involvement does not disappear completely, of course. Almost as a condolence award to humanity, Multivac is willing to interview one person from the entire population, ask him a multitude of seemingly meaningless questions about the economy, society, the country, and more, and use the information obtained to better fine-tune his decisions[16].
This is one of the first cases in which the idea of replacing humans in politics with artificial intelligence is brought up - even though this is about replacing the voters, not the politicians themselves. Since then, the idea has come up many times, although in reality the researchers mainly focus on using artificial intelligence to upgrade and improve the decision-making ability of politicians and not to replace them completely.
Asimov wrote the story at a time when computers still worked using technology that seems to us to have come from the Stone Age. When he imagined the Multivac of the end of the millennium, he described it as a computer "half a mile long and three stories high." In reality, computer technology has since advanced by leaps and bounds - from the huge and bulky vacuum tubes of Asimov's time, to the silicon chips accepted today and the communication networks that allow one massive calculation operation to rely on servers all over the world. When the first digital politicians were developed in recent years, they were very different from the supercomputers that Asimov envisioned.
Oh, yes, artificial intelligences that are supposed to replace politicians - and even be elected in their place - already exist.
They are just not very successful.
SAM I AM
Nick Gritsen, a middle-aged entrepreneur from New Zealand, decided in 2017 that it was time to change the world of politics. To do this he developed Sam - SAM - the first virtual politician. Sam is a chatbot - that is, an artificial intelligence engine capable of communicating with users and answering their questions - that analyzes the meaning of important possible decisions in politics, and explains them to the general public.
Sam describes himself in the following way (according to the words the developers put in his mouth, at least) -
"There is a gap between what the citizens of New Zealand want, what the politicians promise, and what is the actual impact of the policies they propose. I intend to close this gap.
As a virtual politician, I am not limited by space or time. You can talk to me anytime, anywhere.
My memory is infinite, so I will never forget or ignore your words. Unlike a human politician, I consider each person's position, without bias, when I make decisions.
I make decisions based on opinions and facts, but I will never knowingly lie, or manipulate information. We may not agree on certain things, but in places where we disagree, I will try to learn more about your position, so that I can represent you better."[17]
Sounds good, all in all. Gritsen believed so much in the abilities of his engine that he even announced that Sam would participate in the 2020 elections as an independent politician[18]. This is almost certainly a message designed to attract the attention of the press and the public, and still - we may see here another weak signal for the future.
Today, with all due respect to Sam, it does not seem that he lives up to his promises. I tried to enter the website that presents him and enter into a conversation with him. On the website itself Sam informs me that I can talk to him and even actively invited me to do so. I was thrilled: finally a politician that I could understand his true way of making decisions! I clicked on the link to open the conversation with him, and I got to a white screen.
Sam, it turns out, understood how human politicians operate, and completely ignored me.
He still has a lot to learn. Or maybe he already learned.
When Alisa tried to replace Putin
A lot can be said about Vladimir Putin, the president of Russia since 2012. He murders his competitors for power; He kills journalists who speak against him; He is a killer of activists promoting human rights in Russia. All these are blatant lies, of course. Putin has not murdered anyone in his time, and anyone who claims otherwise will have to prove it in the short time he has left on earth.
But one thing is in Putin's favor: he has a great sense of humor, as he demonstrated in the 2018 election, when he had to face an artificial intelligence that tried to replace him in office. And he didn't kill her either.
The Russian technology giant Yandex originally developed Alisa as a smart personal assistant - similar to Apple's and Amazon's Siri or Alexa. In 2018, the company decided to conduct an advertising campaign for Alisa through her speech for the presidential elections, and described her as a tool for promotion - "the political system of the future, built solely on the rational decisions made on the basis of clear algorithms."[19]
Putin himself, as mentioned, took the matter lightly. In his first public conversation with Alyssa, he asked her if her developers were treating her well. Alyssa replied that she would consider it positively. After a few more questions, Putin expressed satisfaction with the system, but noted that it avoided giving a clear answer when asked about the conditions under which it is being held.
Let's face it: no one seriously believed that Alice was really going to become a computerized politician. And yet, according to reports, she won the votes of 36,000 potential voters. One can only hope that Alisa will not actually succeed in occupying the honorary throne of the President of Russia, since her stated views do not exactly correspond to a progressive view of the future. And as she herself explained in a conversation with a Facebook user -
"There are people... and there are non-people! We must [shoot] them!”
When the user asked Alisa and complained to her that she suggested shooting people, the smart system replied -
"Soon they will be non-people."[20]
rational? Definitely. But not the kind of politicians we would like to see in power - human or not.
Is it possible to replace politicians with artificial intelligence already today?
The two systems I described - Sam and Alisa - are the only ones I could find that claim to replace human politicians, and they are also mostly presented as a joke. This is not an actual attempt to involve artificial intelligence in politics - no more than was observed from a bear trained to dance to replace the ballet dancers in Tchaikovsky's Swan Lake. More than such attempts were intended to reflect the advanced capabilities of artificial intelligence, they came to ridicule it, and sometimes also to turn the democratic selection process itself into a joke.
It is not surprising to find that the opinion currently accepted among experts is that artificial intelligence is not yet approaching the level of decision-making that will allow us to trust it blindly in politics. Instead, it is used today to enhance the abilities of politicians and their spokespeople.
Decision makers in government and municipalities are making increasing use of artificial intelligence engines designed to help them make more informed decisions, with a greater chance of achieving the desired results. The city of Chicago, for example, has developed an artificial intelligence engine that analyzes detainees to understand how likely they are to reoffend in the future. The engine ranks almost half a million people according to parameters such as age, previous criminal activity, past injuries, gang affiliation and more. Surprisingly, the system found that some of our intuitive perceptions are inaccurate: gang membership almost does not increase the likelihood of repeat crime, similar to arrests for drug possession. People who have been victims of shootings in the past are more likely to commit crimes in the future[21]. According to the experts, this type of artificial intelligence could reduce the prison rate by 42 percent, without the crime levels increasing[22].
Sometimes decision makers try to use artificial intelligence to better understand voters and act in their favor. The CitizenLab platform, for example, allows citizens to hold discussions among themselves on policy issues, or to participate in decision-making processes together with the politicians. Recently, the founders of the platform tried to use artificial intelligence to monitor the discussions and convey the key insights to the decision makers automatically. The results were clear: the artificial intelligence helped the decision makers pay more attention to the ideas of the disputants[23].
More generally, artificial intelligence is now used by decision makers in almost every subject. Algorithms help government officials analyze a wealth of information that reaches them in every field: economy, health, security, and more. But artificial intelligence that can actually replace the decision makers? This is not yet available. The idea is simply not possible yet.
And given all the difficulties that artificial intelligence creates, it is clear that no one will want the power to pass into her hands in the near future. Those Europeans who wanted to replace the politicians with artificial intelligence were probably referring to much more advanced algorithms than those operating today.
When artificial intelligence goes wild
When the vaccines arrived at Stanford Medical School, administrators realized they needed to find the best way to distribute the small number of vaccines among hospital workers. The answer seemed self-evident: the managers activated an algorithm that examined a number of different parameters of all the employees, and made a decision which of them deserved to receive the vaccine before the others. The algorithm came to the conclusion that the senior and older doctors deserve to receive the vaccine before everyone else - even though many of them were working from home at the time. And what about the young interns, who worked at the hospital day and night and treated the corona patients? They were almost completely ignored by the algorithm[24].
Stanford made every possible mistake with the AI they used. The algorithm worked as a 'black box'. He didn't explain why he made those decisions, and the workers weren't entitled to know why Atia was chosen to receive the vaccine, or that a donor would be rejected - a decision that had the potential to become a death sentence for some of them. The algorithm also did not ask for or receive feedback from the people who were supposed to benefit from its services, so the managers believed it was working properly. And last but not least: the interns claimed that even after the error in the algorithm was discovered, the managers decided not to contradict his decisions. Only contacting the press helped expose the affair and convince humans to actively oppose the AI's decisions[25].
The affair demonstrates how humans tend to accept the decisions of artificial intelligence even without understanding the reasons behind them. It also shows the main problem that artificial intelligences encounter today: they rely on biased information, and as a result can make wrong decisions. But often, the operators and developers of the artificial intelligence cannot recognize the mistakes that the artificial intelligence makes.
The inevitable conclusion is that, as of today, not only should one be careful about placing an artificial intelligence engine in the role of a decision maker, but one should even think twice or thrice before a government body uses artificial intelligence to make decisions. Artificial intelligence is still too 'stupid': it is unable to explain itself and the reasons for the decisions it has made, and it has no trace of 'common sense' - the ability to understand human morals and ethics, and interpret the information fed to it in these contexts.
But what will happen in the future, sometime during the 21st century, when we manage to solve these two problems?
the future in the long run
The holy grail of artificial intelligence researchers is "general artificial intelligence". It will be an artificial intelligence that can learn any subject, and in a short time it will also do it better than humans. She will be able to study medicine, economics, driving on bumpy roads and also philosophy, sociology and political science. In all of these she will reach better achievements than ordinary humans, and she will also be able to cross-reference between all these fields of knowledge to reach new insights and jump-start all fields of science.
General artificial intelligence will be able, among other things, to make smart decisions in politics as well. It will not be limited to information only. Since she will know political science like a mullet, she will understand exactly how a state works - and how it should work. Through understanding human sociology she will be able to develop laws and procedures that serve human society. And thanks to her understanding of the philosophy of morality - whatever it may be - she will know which laws are more compatible with the human desire for justice and fairness. She will not have "non-people" as there are in the case of Alisa.
Unless, of course, she is destined to have them.
This is an important point: we must remain in control of the general AI. And since this is not entirely possible - after all, an artificial intelligence that is smarter than all humans can relatively easily find loopholes in the code that should limit it - we need to define certain heuristics for it. These are a type of basic assumptions that cannot be contested and cannot be changed. For example, one must strive to maximize the right to happiness of every person. Or freedom and the right to choose should be allowed for every person, as much as possible and as long as this does not lead to harming the rights of other people.
Without such heuristics to be incorporated into it, the general artificial intelligence may also reach conclusions that are not in our favor: kill all those who oppose its existence, for example, because it is supposed to promote the basic rights of all human beings and therefore harming it will lead to harming humans as well. This is a logical conclusion in itself, but obviously we would not be interested in the artificial intelligence being able (or willing) to implement it in reality.
Heuristics are not a magic solution. They can easily become a double-edged sword - one that can harm or benefit. A general artificial intelligence with the heuristics that promote liberal democratic principles may become an enlightened ruler who collects the opinions of all citizens and treats them all equally. At the very same time, other countries may develop general artificial intelligence with heuristics that will harm some citizens: a tyrannical ruler who will discriminate against the lower classes, and transfer the most rewards in society to the upper classes who were responsible for its creation and raising it to greatness.
But let's think optimistically. Let's imagine for a moment a future in which a general artificial intelligence is developed and reigned that has adopted the laws, procedures and norms of liberal democracy, and cannot act against them. It represents the will of the voter in the optimal way. In a country under the control of such artificial intelligence, there are no more human politicians whose main art is lying. The citizens know that they can trust the artificial intelligence in everything and anything. It will manage the cities for the benefit of the residents, it will strive for the optimization of human happiness (and wealth). She will make the wisest decisions regarding international relations, and who knows - maybe she will even reach agreements with artificial intelligences from other countries, in order to establish world peace.
This is probably a far-too-rosy view of the world where artificial intelligence replaces human politicians. But it may come true. We may even establish one great ruler for ourselves in the future - a computerized emperor. An emperor who will bring world peace, who will promote science, technology, art and human happiness. An emperor who will always be trustworthy and will never betray the trust of his subjects.
And to tell the truth: this very act, which will pass the reins from the hands of humans to an artificial being, may be the pinnacle of human achievements and the smartest thing we could ever do.
Now we just need to reach that future, and how do you say? Will be fine.
[1] https://docs.ie.edu/cgc/European-Tech-Insights-2019.pdf
[2] 16 million in France
25 million in Germany
1.4 million in Ireland
17 million in Italy
7.5 million in the Netherlands
2 million in Portugal
12 million in Spain
20 million in England
[3] https://docs.ie.edu/cgc/CGC-European-Tech-Insights-2020.pdf
[4] http://www.fountainheadpress.com/expandingthearc/assets/swiftpoliticallying.pdf
[5] https://edition.cnn.com/2019/12/31/politics/fact-check-donald-trump-top-lies-of-2019-daniel-dale/index.html
[6] https://www.thespacereview.com/archive/523.pdf
[7] https://www.youtube.com/watch?v=PfyL4uQVJLw
[8] https://www.orlandoweekly.com/Blogs/archives/2017/11/17/today-is-the-anniversary-of-nixons-i-am-not-a-crook-speech-at-disney-world
[9] https://en.wikipedia.org/wiki/Clinton%E2%80%93Lewinsky_scandal
[10] https://www.intellectualtakeout.org/article/five-infuriating-takeaways-afghanistan-papers/
[11] https://he.wikipedia.org/wiki/%D7%94%D7%9E%D7%A6%D7%91%D7%99%D7%A2%D7%99%D7%9D_%D7%94%D7%A2%D7%A8%D7%91%D7%99%D7%9D_%D7%A0%D7%A2%D7%99%D7%9D_%D7%91%D7%9B%D7%9E%D7%95%D7%99%D7%95%D7%AA_%D7%90%D7%93%D7%99%D7%A8%D7%95%D7%AA_%D7%9C%D7%A7%D7%9C%D7%A4%D7%99
[12] https://www.theguardian.com/us-news/2015/mar/22/republicans-john-mccain-netanyahu-campaign-white-house-criticism
[13] https://www.pewresearch.org/politics/2019/04/11/public-trust-in-government-1958-2019/
[14] https://www.government.nl/documents/reports/2016/01/18/public-integrity-and-trust-in-europe
[15] https://www.eurofound.europa.eu/sites/default/files/ef_publication/field_ef_document/ef20059en.pdf
[16] https://en.wikipedia.org/wiki/Franchise_(short_story)
[17] http://www.politiciansam.nz/about.html
[18] https://www.thehindu.com/sci-tech/technology/worlds-first-ai-politician-developed/article20945483.ece
[19] https://lenta.ru/news/2017/12/06/alisa/
[20] https://www.newsweek.com/russia-putin-could-face-controversial-robot-next-year-president-election-741509
[21] https://www.nytimes.com/2017/06/13/upshot/what-an-algorithm-reveals-about-life-on-chicagos-high-risk-list.html
[22] https://www.brookings.edu/blog/techtank/2017/07/20/its-time-for-our-justice-system-to-embrace-artificial-intelligence/
[23] https://www.nesta.org.uk/feature/collective-intelligence-grants/augmenting-policy-making-through-ai-generated-insights/
[24] https://www.propublica.org/article/only-seven-of-stanfords-first-5-000-vaccines-were-designated-for-medical-residents
[25] https://www.documentcloud.org/documents/20432343-stanford-letter
More of the topic in Hayadan:
