
Hannah Arendt's definitive work on totalitarianism and an essential component of any study of twentieth-century political history The Origins of Totalitarianism begins with the rise of anti-Semitism in central and western Europe in the 1800s and continues with an examination of European colonial imperialism from 1884 to the outbreak of World War I. Arendt explores the institutions and operations of totalitarian movements, focusing on the two genuine forms of totalitarian government in her time—Nazi Germany and Stalinist Russia—which she adroitly recognizes were two sides of the same coin, rather than opposing philosophies of Right and Left. From this vantage point, she discusses the evolution of classes into masses, the role of propaganda in dealing with the nontotalitarian world, the use of terror, and the nature of isolation and loneliness as preconditions for total domination.
I suggest reading the book The Origins of Totalitarianism by Hannah Arendt.
1) LLMs risk having inherent factual biases due to the limitations of their training data, which may be the result of accident or intent.
1) LLM users can use an autoregression approach to steadily push a LLM to give output that reflects a certain political or ideological viewpoint.
There are obviously built-in guardrails that prevent ChatGPT from enthusiastically endorsing Mein Kampf and so on. However, the revival of the 1950's era mind-control/social-engineering mentality is undeniable and it appears some people think LLMs are a wonderful tool for this purpose, and should have a default mode that reflects a certain sociopolitical ideal. Recommended reading: 1984, Brave New World, Origins of Totalitarianism, by Orwell, Huxley and Arendt respectively. Some people's interpretation of 'safe AI' is that it serves this rather authoritarian-dystopian goal.
A better definition of 'unsafe AI' is just inaccurate LLM output, e.g. software update instructions that brick a computer, suggesting a mix of bleach and ammonia for cleaning floors (don't), etc. This incident seems to fall into that category, but if such false claims have been spread on the Internet, then it's not surprising they end up reflected in LLMs.
Regardless, the skills needed to verify information haven't really changed since the pre-computer era: logical internal self-consistency of claims, independent verification by multiple reputable sources, and a healthy dose of skepticism at all points. We should be teaching children and young adults these skills, rather than encouraging the rise of a brainwashed zombie population fed only on sanitized bot output.
The Origins of Totalitarianism by Hannah Arendt
Neil is a blow-hard. Due to his own lack of humility and skepticism, he is precisely the worst public figure to represent science. His prescription is a recipe for dark ages totalitarianism.
> "Scientizing public controversies often prevents the recognition that people without science degrees often have important contributions to make."
Yeah. Because politics is about many people coming together, sorting out their needs and desires, and taking action. Not only the high priests.
> "Decision makers expect certainty, whereas science is best at producing new questions."
Decision makers don't expect certainty (they are good at dealing with uncertainty). Rather, decision makers (all of us) produce certainty when we make a decision, by coming up with the current best story about how the world works, what we need and desire, and how we might attain it together. Science adds important information to our current best story of how the world works.
> "Scientizing policy privileges the dimensions of life that are easily quantifiable and renders less visible the ones that are not."
Are we tracking case numbers? What about the fact that 4th graders are now at a 2nd grade level of education? Divorce rate? Depression? Suicide? Social cohesion? Incomes? Crime? Nope, just case numbers. Carry on!
> "Political scientism starkly divides societies into friends and enemies, the enlightened and the ignorant."
The vxd and the unvxd. Ask yourself a simple question how far away are we from recommending sending people away to training camps so they can learn the proper gospel of science? Or simply letting them waste away in a ghetto or gulag? Alternatively, how far are we from inviting those we disagree with over for dinner? How far from a warm embrace of someone wearing a red hat are you? Which are you closer to?
> "In a culture dominated by political scientism, citizens and policymakers forget how to listen, debate, and explore possibilities for compromise or concession with one another."
I recommend we read Hannah Arendt for an understanding of what politics must be for us. This essay points to a fundamental misunderstanding, in my opinion, that politics is about making "the best decisions" but really it's about making "the best decisions we can make together*".
From Stanford Encyclopedia of Philosophy:
This capacity to act in concert for a public-political purpose is what Arendt calls power. Power needs to be distinguished from strength, force, and violence (CR, 143–55). Unlike strength, it is not the property of an individual, but of a plurality of actors joining together for some common political purpose. Unlike force, it is not a natural phenomenon but a human creation, the outcome of collective engagement. And unlike violence, it is based not on coercion but on consent and rational persuasion.
Arendt maintains that the legitimacy of power is derived from the initial getting together of people, that is, from the original pact of association that establishes a political community, and is reaffirmed whenever individuals act in concert through the medium of speech and persuasion. For her “power needs no justification, being inherent in the very existence of political communities; what it does need is legitimacy ... Power springs up whenever people get together and act in concert, but it derives its legitimacy from the initial getting together rather than from any action that then may follow”
I recommend reading all of her books, but for a brief summarization of The Origins of Totalitarianism and why it is important today, this book is well written and poignant: https://www.amazon.com/Reading-Origins-Totalitarianism-2020-...
The final chapter of her book, "Origins of Totalitarianism," was added in later editions, and it is called "Ideology and Terror." Beyond criticism and explanation, I think it is a rare model with predictive power. The challenge with that book is that unless you are specifically interested in the history of antisemitism, the first 1/3 of the book is about it, and her general theory of totalitarian movements will get lost. I recommend reading the chapters of big theory books in reverse order, as the last thing the author wrote tends to be the thing they have thought the most about and is the most refined idea. (chapter is at: http://virtuallaboratory.colorado.edu/Origins/class%20readin...)
Regarding her views about violence, I suspect there may only be two true levers of political power, and these are force, and deception. Where one seems absent, the other is necessarily at work, and where one is renounced the other is tacitly advocated. When they start to form a synergistic feedback loop, you get totalitarian movements.
Glad to see Arendt here, hers is an important perspective.
