The AI Con by Emily M. Bender and Alex Hanna
2025-Jul-11, Friday 12:55 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
altamira16This book was a very well done AI skeptic book that was rooted in deep knowledge of the history of artificial intelligence. It brought to light some interesting points that I had never thought about, and it never descended into a rant.
It gets into the history of AI, and a lot of that discussion is rooted in the type of probabilistic models that I learned about in grad school. It is discussing n-grams, Markov, and so on.
There is a discussion about how AI is an attempt to break labor and gets into a more detailed history of the Luddites. The Luddites were craftsmen, and machines were replacing their hard won skills with an inferior product. The machines that were doing this were also dangerous to their operators.
Various people involved in AI feel like there should not be any AI policy until it is thoroughly discussed, but the authors propose that existing laws should be used to limit the use of AI in areas where it can do harm. They quote Michael Atleson, an attorney within the FTC Division of Advertising Practices:
The book was extremely critical about the use of AI in making medical decisions and in law. Law has to do with the nuance of language, and generated language that no human really thinks through does not have the same nuance.
There were also good arguments for limiting the use of AI in education.
While some technology in education is important, a lot of technology in education is designed to give an inferior education to poor kids and union-bust.
One thing that I did not know was that the little Gemini summary on a Google search uses 10-30 times more energy than search before this feature was added.
The authors see both AI doomers and AI boosters as two sides of the same coin. Both of these groups believe that the AI will become smarter than humans. The outcome is the only thing that they differ on.
The group that wants to consider the data used to train the models and the impacts that AI has on the present really does not want to get lumped in with AI doomers that think that the AI is going to eventually get so smart that it will destroy humanity. They are rooted in reality while the doomers are not. There was some criticism of how Vice President Harris was trying to get the people concerned with the present impact of AI to work with the doomers.
There were a lot of references Karen Hao's work. How has recently released the book "Empire of AI." Hao is an AI journalist specifically focused on OpenAI.
It gets into the history of AI, and a lot of that discussion is rooted in the type of probabilistic models that I learned about in grad school. It is discussing n-grams, Markov, and so on.
There is a discussion about how AI is an attempt to break labor and gets into a more detailed history of the Luddites. The Luddites were craftsmen, and machines were replacing their hard won skills with an inferior product. The machines that were doing this were also dangerous to their operators.
Various people involved in AI feel like there should not be any AI policy until it is thoroughly discussed, but the authors propose that existing laws should be used to limit the use of AI in areas where it can do harm. They quote Michael Atleson, an attorney within the FTC Division of Advertising Practices:
Your therapy bots aren't licensed psychologists, your AI girlfriends are neither girls nor friends, your griefbots have no soul, and your AI copilots are not gods.
The book was extremely critical about the use of AI in making medical decisions and in law. Law has to do with the nuance of language, and generated language that no human really thinks through does not have the same nuance.
There were also good arguments for limiting the use of AI in education.
In August 2020, thousands of British students, unable to take their A-level exams due to the COVID-19 pandemic, received grades calculated based on an algorithm that took as input, among other things, the grades that other students at their schools received in previous years. After massive public outcry, in which hundreds of students gathered outside the prime minister's residence at 10 Downing Street in London, chanting "Fuck the algorithm!" the grades were retracted and replaced with grades based on teachers' assessments of the student work.
While some technology in education is important, a lot of technology in education is designed to give an inferior education to poor kids and union-bust.
One thing that I did not know was that the little Gemini summary on a Google search uses 10-30 times more energy than search before this feature was added.
The authors see both AI doomers and AI boosters as two sides of the same coin. Both of these groups believe that the AI will become smarter than humans. The outcome is the only thing that they differ on.
The group that wants to consider the data used to train the models and the impacts that AI has on the present really does not want to get lumped in with AI doomers that think that the AI is going to eventually get so smart that it will destroy humanity. They are rooted in reality while the doomers are not. There was some criticism of how Vice President Harris was trying to get the people concerned with the present impact of AI to work with the doomers.
There were a lot of references Karen Hao's work. How has recently released the book "Empire of AI." Hao is an AI journalist specifically focused on OpenAI.
