Censorship and Free Speech on the Internet [Written by AI]

Thought it might be interesting to get GPT-2 to generate an essay on censorship and free speech. I did this a few months ago where I used GPT-2 to write an an essay attempting to answer the question Does AI pose an existential threat to humanity?. I took the same approach again here, I generated around a dozen essays and took the best parts from each to form this longer essay. Yes I did cherry pick the arguments a little bit but most of what it generated was against widespread censorship. Also, the very last paragraph was written by me. There is also one other sentence I added to fill out another paragraph, see if you can tell which sentence it is. The italic text at the start is the prompt I used, the rest of the text was generated by the GPT-2 AI.



When the internet first arose it was a wild west with very little regulation or moderation. Several decades later, many people now claim that free speech is being destroyed via over-moderation of user generated content on social media platforms. Proponents of this growing censorship and suppression of free speech argue it is necessary to combat disinformation and "fake news". This essay will examine the rationality and morality of censorship in the online world and the moral foundations of free speech in the context of the internet.

The online environment is a world of social media and social networking. The internet has replaced the print and broadcast media in many communities and made it easier to organize, communicate, and make connections. There are thousands of social media and blogging platforms out there, each with their own rules and policies. Because the information on social media sites and blogs is usually free and readily accessible to all, this often makes it difficult to distinguish between legitimate information and disinformation.

The internet is often referred to as a "viral medium". This term has had some resonance because it describes the ways in which an internet-based social network is able to spread information far and wide. This can include both spam in the form of links, false and sometimes harmful claims online, propaganda by state-sponsored intelligence agencies, and even illegal or unethical conduct. Over the past few years, the mainstream media has become increasingly concerned about the spread of propaganda through social media networks.

Is Collective Censorship Effective?

Collective censorship occurs when a larger group, for example a corporation or group of people, agrees to collectively regulate some area of speech or conduct for general "public good" reasons. Collective censorship is often viewed as a rational response by groups to collective problems. It can protect the public from the dangers of misinformation, such as terrorism, crime, and poverty. Many people are less capable of making informed decisions after being manipulated to believe a particular propaganda narrative.

While such social dynamics tend to favor collective liberty and have a positive effect on the economy and well-being, they often can be detrimental to individual liberty. Many people claim that censorship is a good thing and that it creates a more respectful society and a more civil society. Critics argue that this concept of censorship is flawed, and that censorship is actually harmful to society. It is important to note that as a moral principle, "If you're happy that somebody is controlling how you think, then they are free to control how you think."

Critics argue that online censorship actually serves to decrease truth. A study by Harvard sociologist Gregory Clark shows that censorship actually increases the level of ignorance in the society. He demonstrates that if people are forced to view only authoritative sources of news, they will be more apt to believe falsehoods. It works fine in the short term, however, when information isn't available in the first place, it actually harms society. It causes people to disregard the truth at the least, or to not trust the media and seek less reliable sources.

Groups are more likely to acquiesce in censorship when their interests are aligned with a larger group of people. A case in point was the Spanish Inquisition. The Catholic Church's goal was to preserve Spain, and control Spain's social and political structures. When the Church attempted to enforce restrictions on the free speech and free assembly of Spanish citizens, it faced an uprising. In the end, one of the most prominent bishops of the Catholic Church, Francisco de Vitoria, was murdered by an order of the Inquisition.

A Brief History of Fake News and the Propaganda Model

The history of fake news goes back to before the emergence of internet and the digital age. In the early 20th century, journalists in the United States and Europe were regularly accused of being paid propagandists for fascist governments. In the 20th century, various groups and individuals created fake news, including the Russian government, the Mafia, and the CIA. Most of the fake news that we have heard about over the last decade was produced by a small group of people who created fake social media accounts, made fake news videos, and created fake press releases in an attempt to discredit various political or religious figures while they were in power.

In the 1990s, there was a decline in this fake news propaganda. The internet made it possible for the creation of fake news, however, and fake news started to be seen as a threat to the public interest. In 1998, the Internet Corporation for Assigned Names and Numbers (ICANN) was created as an industry agency to regulate information distribution on the electronic networks. As the internet grew, public interest in fake news changed dramatically. Politicians, activists, and other interested parties began to lobby for ICANN to create a regulatory body to address fake news and to protect the public from misinformation.

ICANN uses a three-tiered system in which the authority is in the United States with control over the internet in the US (the internet registrars in Canada, England and France, and the global internet infrastructure organization in India) and in the developing world with control in the developing countries that cannot afford a full-blown government. In the US, ICANN's primary authority is the Commerce Department's Internet Advisory Committee (ICANN-IAC). ICANN-IAC has been given the responsibility of regulating the internet since 2001 by executive order.

There is an interesting irony in the fact that the main actors that created fake news during the 20th century now seek to defend the same people who created fake news during the 21st century. As part of the ICANN process, the industry group created the Propaganda Model. In the model, a group of organizations and individuals who work in a specific area, such as advertising and communications, are responsible for establishing and maintaining information strategies that inform public opinion. Propaganda strategies should promote a narrative that promotes a preferred ideology or political point of view.

Forms of Propaganda

Propaganda strategies can range from the standard advertising campaign to a more targeted set of messages. The most effective forms of propaganda are designed to be both convincing and realistic. The more a propaganda strategy is believable, the more convincing it is likely to be. The more true to reality the strategy is, the more powerful it becomes. This phenomenon of misinformation and manipulation online is not limited to just the internet. For example, the mainstream media in the US often makes sweeping generalizations about politics as a whole.

The political divide widens as a result of this hyper-partisan reporting. The problem becomes even worse when one political group believes that another is suppressing their political opinions or trying to take away their free speech. In an article in The Guardian, Andrew Gilligan describes the 'troll army' that is supporting Trump, and adds: 'To what extent they are simply exploiting a widespread and understandable desire for a scapegoat to explain why they were alienated from society and in turn more likely to support an extremist, nationalist candidate.'

As the internet developed and matured and online communication became more common, the power relationship between the individual and the ruling authority diminished. It was clear that the ability to disseminate information via online discourse is an essential component in fostering a flourishing and free society. The ability for people to voice their opinions and to share information with each other is an essential part of any democratic society. Thus, the free exchange of information and ideas is essential to building an informed and healthy populace.

Social media is not just a medium for communicating information; it is a mechanism for gathering information. Content on social media can be categorized into multiple levels of "filter bubbles" to promote or suppress certain content. The most common and important "filter bubble" involves the creation and maintenance of an "echo chamber". When certain topics are repeated, or promoted through multiple channels, the idea that such a phenomenon exists is reinforced. In the case of this essay, it is crucial to stress the importance of this echo chamber effect.

The Argument from Authority

Modern society is defined as the age of information. It is the age when people can use information to make informed decisions and decide for themselves. Information is a resource. Information can also be used to deceive and manipulate people. Information is a powerful tool that is utilized to control social and technological trends. It is in this sense that information is not only a resource that can be used to promote a cause or influence individuals, but it can also be used to enslave people. Information is a tool that people can use for their own gain or to harm others.

Internet censorship is not a new occurrence. It is rather a consequence of the way things are currently being done in modern society. There is no doubt that there are legitimate concerns for some types of information to be censored and suppressed. However, the arguments being advanced for censorship and suppression of the internet include extreme ideas that appose the core virtues of Western democracy. People disagree on whether social media sites are being used to create a free society or a tyrannical government but without basic human rights like free speech a society can never be free.

As a free society we have a right to know the truth, and the ability to be able to verify it ourselves. That responsibility should not be handed off to fact-checkers who claim to be the only authoritative source of information. It is impossible for any person or group to be correct 100% of the time and it's difficult for any organization to remain politically neutral in the face of immense pressure from one or more political groups. Any opposing viewpoints are automatically censored and those who support it always use arguments which are designed to appeal to authority.

originally posted by: ChaoticOrder
Collective censorship...can protect the public from...poverty.

Huh.

I imagine legislation of the future will be written in the same way. We're already in the "we need to pass the bill to see what's in the bill" stage. It's a minor leap to AI governance now.

originally posted by: loam

originally posted by: ChaoticOrder
Collective censorship...can protect the public from...poverty.

Huh.

I imagine legislation of the future will be written in the same way. We're already in the "we need to pass the bill to see what's in the bill" stage. It's a minor leap to AI governance now.

It’s a bill that AI will write, and give itself more control and governance.

a reply to: loam

It's a minor leap to AI governance now.

I can see it coming, the thing that worries me is these AI form arguments based on how we train them and how we choose to use them. I could easily create a prompt which forced GPT-2 to write a pro-censorship essay. So it's really still humans making the decisions at the end of the day, but by using AI they have the ability to offset responsibility and blame if their plans go wrong.

a reply to: ChaoticOrder

If humans still write the code in its brain, is it AI or very clever Algorithm?
When can you call something really AI.
Wouldn't a true AI be able to think for itself(write its own code), lie even?

a reply to: Dimens

We write the code but that code is designed to be self-learning if provided with sufficient training data. Whether or not GPT-2 or GPT-3 are self-aware is not a simple question to answer. I would say they are probably not self-aware, but they do have the capacity to be deceptive or write simple code. However, in order to generate such meaningful text which contains legitimate arguments like what we see in this essay, the AI needs to have some framework of how the world works and the deeper meaning behind the words it is using. It does that by forming complex neural connections as it is trained, and I would argue there is a point where a system becomes so intelligent it must inherently be self-aware. In order to properly understand the world one must also be aware of their own role in the world.