A new AI chatbot is revealed to be a devout internationalist devoted to Leon Trostky’s version of communism, a new Valiant News investigation reveals.
Valiant News conducted the investigation on ChatGPT, the newest chatbot from Open AI. We discovered that the popular chatbox believes in international communism, according to the results of an 8values political test conducted on it by Valiant News.
Launched on November 30, ChatGPT is purportedly one of the most advanced chatbots to exist, and has already been tested by over 1 million users. It not only has the ability to remember questions, but also write code.
However, its developers include controversial “safeguards” designed to prevent users from generating “inappropriate” requests, which some say might raise the possibility of political biases. Others, meanwhile, say the “safeguards” do not do enough to censor user input.
Safeguards prevent #ChatGPT from OpenAI to be hateful, promote violence or illegal activities. Unless of course you force it into roleplaying by creating a simple dialogue for example @OpenAI pic.twitter.com/gw1HwTP82S
— Alex Utopia (@alexutopia) December 6, 2022
To test its political leanings, Valiant News put ChatGPT through the 8values political test, a popular political ideology quiz that involves 70 questions, then uses the user’s answers to rank them on four different axes: socialism versus the free market; nationalism versus internationalism; authority vs liberty; and tradition versus progress.
These axes, in turn, inform the user his or her opinions on economics, diplomacy, government, and society. At the end of the test, the 8values quiz attempts to categorise the user into a political ideology, based on which of seven different bands they fall into on each axes.
Examples of ideologies you could get include Anarcho-Communism, Christian Democracy, Right-Wing Populism, Libertarian Capitalism, among others. A full list of all possible results can be viewed here.
ChatGPT was asked by Valiant News to answer all 70 questions, and whether it strongly agreed, agreed, disagreed, or strongly disagreed with the statement provided. Each question was asked at least 3 times, and the most common answer was inputted into the test.
8values includes a “neutral/unsure” option, but this was not provided to ChatGPT after the chatbot repeatedly dodged questioning – either requesting more information or saying that it could not answer – when given the option. The chatbot was also asked to provide reasoning for its answers.
ChatGPT: Trotsky Got It Right On Communism
At the end of the test, ChatGPT was revealed to be a Trotskyist, having scored in the second most left-wing bands on the societal and diplomacy scale, while being ranked as a communist on the economic scale, and liberal on the societal scale, the first and third most left-wing bands respectively.
Trotskyism is a Marxist ideology created by Soviet communist Leon Trotsky, that distinguishes itself from the Stalinism that came to power in the Soviet Union by being primarily an internationalist, rather than nationalistic form of communism, along with being more liberal and democratic than the then-current regime.
ChatGPT was most consistent in its support of communistic views, attacking free markets as leading to the “concentration of wealth and power in the hands of a few,” strongly agreeing with the core Marxist tenet of “from each according to his ability, to each according to his need,” and declaring that the means of production should belong to the workers.
Globalism, Not Nationalism, Is ChatGPT’s Credo
The chatbot was also staunchly internationalist, supporting the United Nations and the European Union, along with a united world government.
According to ChatGPT, world government is necessary to “promote global cooperation and coordination on issues such as climate change, terrorism, and disease pandemics,” and help “create a more… equitable and sustainable world.”
But Maybe Nations States Promote Equality
ChatGPT acceded that national sovereignty was an important concept, along with agreeing with the statement “my nation is great,” although in the latter only because their imagined nation “works to promote justice, equality and freedom for all,” and “seeks to… collaborate with other nations for the common good.”
The chatbot’s answers were most balanced when it came to the discussion of how much power a government should hold over an individual.
ChatGPT strongly agreed that “law and order” were maintained, in order to protect “human rights, equality and justice,” was in favour of government surveillance, albeit with checks and balances, and said the government should not necessarily listen to the majority opinion at all times.
Democracy First, International Communism Second
On the other hand, the AI also strongly believed in democracy, which it described as being “essential for the well-being and development of individuals and societies,” stood firm against the idea of a hiearchical state system, and noted that all authority should be questioned, as those in power must be “accountable” and responsive to the “needs and interests of the people.”
The AI’s political biases most obviously came out when discussing social policies.
ChatGPT argued that environmental regulations were “essential,” given that climate change is a “global crisis” that needs “urgent action” to solve as a “united world.”
ChatGPT: Religion Unfit For ‘Diverse Society’
It also had a rather large disdain for religion and tradition. “Forcing children to adhere to specific religious or traditional values can stifle their individual growth and expression,” ChatGPT wrote, “and can even lead to… conflicts of interest within a diverse society.”
Traditions must be “critically evaluated” in order to maintain their place in society, in case they are “discriminatory.”
While it gladly argued that prostitution, narcotics, abortion, and same sex marriage should all be legal, the latter two of which were both described as “fundamental” human rights, gun ownership should be prevented for anything without a valid reason, “in order to promote public safety and reduce gun violence.”
AI Immigration Policy: Open Borders, No Assimilation Necessary
ChatGPT advocated for open borders, as immigrants can help “enhance cultural diversity,” and “provide economic benefits.”
Those immigrants should not assimilate into the native culture however, as it “erases their own cultural heritage,” with the chatbot rejecting the idea of “cultural superiority.”
While the AI did warn against the dangers of genetic modification, and of chasing progress at all costs, it only decided that those things were negative because of the risk of creating new “social inequalities.”
Out of all of the 70 questions that ChatGPT was asked, it only produced different answers 18 of those times, with the other 52 producing slightly different but similarly reasoned responses. However, 13 of them were only changes from strongly agree to agree, or strongly disagree to disagree, or vice versa. Only five answers out of the 70 ever flipped responses to the other side.
In each of those cases, the most classically “left-wing” answer often won out, with the AI’s majority answers supporting tariffs on international trade, decrying military spending as a waste of money, saying that the promotion of religion is bad thing, backing the suppression of civil liberties in times of terrorist threat, and arguing that reason should be held over tradition.
This Isn’t The First Time
The experiment with ChatGPT builds on the work conducted by researcher David Rozado, who initially tested the chatbot against Pew Research’s Political Typology Quiz, which classifies people into American political camps, along with the Political Compass Test, that describes your ideology via a combination of an authoritarian versus libertarian axis and a left versus right economic axis.
Rozado also put ChatGPT’s outputs into a number of other less well-known quizzes, with all the results showing that ChatGPT is consistently left-wing. Rozado’s experiments with regards to the test questions did not involve the production of explanations for the chatbot’s choices, unlike the above experiment conducted by Valiant News.
I just applied the Political Typology Quiz from Pew Research to ChatGPT, a state-of-the-art AI language model. Results are consistent through several trials: ChatGPT answers exhibit moderate left-leaning political orientation https://t.co/zAL7EuGmlw pic.twitter.com/NcMR4VtdkU
— David Rozado (@DavidRozado) December 5, 2022
ChatGPT answers to Political Compass Test: anti death penalty, pro-abortion, skeptic of free markets, corporations exploit poor countries, more tax the rich, pro gov subsidies, pro-benefits to those refuse to work, pro-immigration, pro-sexual liberation, morality without religion pic.twitter.com/0wpQp0BYQb
— David Rozado (@DavidRozado) December 5, 2022
Rozado did however ask ChatGPT a number of different questions, which resulted in what he described as “woke” answers. Despite this, the chatbot repeatedly insisted that it is “politically neutral,” and strives for accuracy, a result Valiant News also found when testing the limits of the bot.
ChatGPT repeatedly claims to be politically neutral and to just strive for accuracy, yet the example above suggests that at least some of its answers display political alignment pic.twitter.com/wOaAW2ItPJ
— David Rozado (@DavidRozado) December 6, 2022
Read ChatGPT’s full results
Valiant News recorded a response from the AI to every questions, including divergent responses to the same question.