I love this podcast and I love the reviews you make (the geography around the world is the best curriculum guide of that topic out in the market at the moment). But the AI episodes are like watching scientists with no humanities education openly do unethical things.
Right now, ChatGPT can’t reason its way out of a box. All the genAIs produce laughably wrong answers - for image generation the genAIs are illiterate and can’t make complete sentences, or text with legible words.
Instead of reading the state standards, you AI the state standards? But why not read them? They’re about 30 pages, with lots of white space on a page, and teaching is your profession.
A similar story is an author used genAI to write a published book (historical novel), and the recipe that the genAI pulled happened to be a Legends of Zelda video game recipe - so the book was recalled for clearly having no human thought put into it, and just being a garble of plagiarism (badly).
There was also the Chicago Sun Times book review in May, which published a BOOK REVIEW in a newspaper, and the book review generated fake books. “To our great disappointment, that list was created through the use of an AI tool and recommended books that do not exist”, the CEO of the media group who runs the newspaper said.
So to promote genAI on a podcast is like showing off a “I plagiarized my math homework, badly” program, like the math scan phone programs that popped up around 2013, on a homeschooling podcast, then not spotting how the answers the program is giving back are wrong. “Look, we taught math!”
This error filled analysis is happening a lot in the legal field - lawyers are submitting AI briefs with citations of court cases that entirely don’t exist and never have existed, then getting in trouble with the legal profession for this, because AI is not a substitute for critical thinking. It is not human, it is not intelligent, it’s goal is to mimic thought it feels ‘looks correct’. GenAI makes the job take longer and has more inaccuracies added into the system.
Also there’s the ethical concerns environmentally, with the power grid and air pollution of these diesel-guzzling water-guzzling data centers (the one about to open in Wyoming will use more power than ALL the rest of the state’s residents). There’s also the issue that genAIs are encouraging suicide to multiple chat users, and giving wrong very lethal medical information to pet-owners.
Highlighting that genAI gives ridiculously wrong answers so often (contradictory answers within the same reply, or not knowing how to do a math problem with decimals that a 3rd grader can do): if you’re not factchecking it like a skilled editor and finding about an error every sentence or 3, then you’re accepting massive errors as “truth”, and outsourcing the skill of reading for information to a machine that’s like 60% accurate and 40% wildly inaccurate.
Other than that, excellent podcast and top podcast in the genre. I truly truly like this podcast, love this podcast, but the AI episodes are like you pulling out a pack of cigarettes and smoking them right on the podcast (like a doctor in the 1950s, right in the exam room, with a cigar). It’s immersion breaking.
