{
"$type": "site.standard.document",
"content": "---\ntitle: \"Drinking from the bullshit firehose\"\ndescription: \"Draft notes on AI-generated bullshit in higher education---the problem isn't\n that it's wrong, it's that Brandolini's law means we'll be overwhelmed by the\n sheer volume.\"\ntags:\n - ai\n - teaching\n---\n\nhttps://stratechery.com/2022/ai-homework/\n\nhttps://en.wikipedia.org/wiki/Brandolini's_law\nhttps://en.wikipedia.org/wiki/On_Bullshit\n\nhttps://www.wired.com/story/chatgpt-fluent-bs/ (ChatGPT is fluent in bullshit\nbecause it's trained on (human-created) bullshit)\n\nhttps://www.cnn.com/2021/11/16/media/steve-bannon-reliable-sources/index.html\n(\"flood the zone\")\n\nhttps://twitter.com/LuminanceBloom/status/1600598003391266816 (snarky zizek\ntweet about the absurdity of the whole thing)\n\nhttps://twitter.com/yoavgo/status/1602026029979164675 (write an essay and\ncritique the resulting prose)\n\nhttps://medium.com/essays-and-non-fiction/i-am-a-bullshit-artist-6638eaedb300 (I\nam a Bullshit Artist)\n\nWhy Meta’s latest large language model survived only three days online\nhttps://www.technologyreview.com/2022/11/18/1063487/meta-large-language-model-ai-only-survived-three-days-gpt-3-science/\n\n---\n\nAI language models have advanced to the point where they can generate text that\nis convincing and difficult to distinguish from human-generated content. This\nhas led to the rise of AI-generated bullshit, which is defined as text that is\ndesigned to deceive or manipulate without providing any real information or\nvalue.\n\nBrandolini's law states that the amount of energy necessary to refute bullshit\nis an order of magnitude bigger than to produce it. In other words, the main\nproblem with AI-generated bullshit is not that it will be factually incorrect,\nbut that educators will be overwhelmed by the sheer amount of it.\n\nBetter AI-generated bullshit detection tools might seem like the solution, but\nit is likely to be an endless arms race between the bullshit generators and the\ndetectors. Instead, our society needs to get better at dealing with bullshit at\nscale.\n\nOne way to do this is to teach our students how to recognize and evaluate the\nquality of written content. This not only helps them avoid being deceived by\nAI-generated bullshit, but also enables them to become more discerning consumers\nof information in general. Educators can also play a role by partnering with\nstudents to generate and detect bullshit, and by providing guidance on how to\nnavigate the overwhelming amount of information available today.\n\nIn conclusion, the use of AI-generated bullshit in higher education is becoming\nmore common and presents both benefits and drawbacks. By being aware of its\nexistence and potential impact, educators and students can make informed\ndecisions about its use and work towards building a society that is better\nequipped to deal with bullshit at scale.\n\n---\n\nLet's try again. The three points I want to make are:\n\n1. AI language models are now at a point where they can generate convincing\n bullshit (here I'm using bullshit in the technical sense, following Harry\n Frankfurt) with minimal human effort\n2. Brandolini's law tells us that the main problem here isn't that the text will\n be wrong, but that educators will be overwhelmed.\n3. Better AI-generated bullshit detection tools aren't the answer, because it'll\n always be an arms race between the bullshit generators and the bullshit\n detectors.\n4. What's clear is that our society needs to get better at dealing with bullshit\n at scale, and fast. And this is what we should be teaching our students.\n Which means we get to partner with them in generating the bullshit, but also\n training them to recognise it (and become more discerning consumers of\n information in the process).\n\nCan you write the above points into an article for a general audience?\n\n---\n\n_this was the initial one_\n\nI need to write an article on the use of AI-generated bullshit in higher\neducation. I'm not interested in saying either \"this is amazing\" or \"this is\nterrible\". The points I want to make are:\n\n1. AI language models are now at a point where they can generate convincing\n bullshit (here I'm using bullshit in the technical sense, following Harry\n Frankfurt) with minimal human effort The use of AI-generated bullshit in\n higher education is becoming more common There are potential benefits and\n drawbacks to using AI-generated bullshit in higher education It is important\n for educators and students to be aware of the existence and potential impact\n of AI-generated bullshit in order to make informed decisions about its use In\n recent years, AI language models have advanced to the point where they can\n generate text that is convincing and difficult to distinguish from\n human-generated content. This has led to the rise of AI-generated bullshit,\n which is defined as text that is designed to deceive or manipulate without\n providing any real information or value.\n\nOne area where AI-generated bullshit is gaining popularity is higher education.\nIn some cases, educators are using AI tools to generate papers and other written\nassignments that are indistinguishable from human-generated content. This can\nsave time and effort for both students and educators, but it also raises ethical\nconcerns.\n\nOn the one hand, AI-generated bullshit has the potential to be a valuable tool\nin higher education. By automating the production of written assignments, it can\nfree up time for educators and students to focus on more important tasks, such\nas research and critical thinking. It can also help students learn to identify\nand evaluate the quality of written content, which is an important skill in the\ndigital age.\n\nOn the other hand, there are also potential drawbacks to using AI-generated\nbullshit in higher education. For one thing, it can be used to cheat on exams\nand assignments, undermining the integrity of the education system. It can also\nlead to a proliferation of low-quality content that does not provide any real\nvalue or knowledge.\n\nIt is important for educators and students to be aware of the existence and\npotential impact of AI-generated bullshit in higher education. By understanding\nthe potential benefits and drawbacks of using AI tools to generate written\ncontent, they can make informed decisions about whether and how to use these\ntools in their own work. It is also important for educators to develop\nstrategies for detecting and dealing with AI-generated bullshit, in order to\nmaintain the integrity of the education system and ensure that students are\nlearning valuable information and skills.\n\n---\n",
"createdAt": "2026-05-13T23:14:47.610Z",
"description": "Draft notes on AI-generated bullshit in higher education---the problem isn't that it's wrong, it's that Brandolini's law means we'll be overwhelmed by the sheer volume.",
"path": "/blog/2022/12/07/drinking-from-the-bullshit-firehose",
"publishedAt": "2022-12-07T00:00:00.000Z",
"site": "at://did:plc:tevykrhi4kibtsipzci76d76/site.standard.publication/self",
"tags": [
"ai",
"teaching"
],
"textContent": "Draft notes on AI-generated bullshit in higher education---the problem isn't that it's wrong, it's that Brandolini's law means we'll be overwhelmed by the sheer volume.",
"title": "Drinking from the bullshit firehose"
}