Aurich Lawson
Think about you are a podcaster who commonly does fast 10- to 12-minute abstract critiques of written works. Now think about your producer offers you a number of pages of nothing however the phrases “poop” and “fart” repeated time and again and asks you to have an episode concerning the doc on their desk throughout the hour.
Talking for myself, I would have hassle even understanding the place to start out with such a activity. However when Reddit person sorryaboutyourcats gave the identical immediate to Google’s NotebookLM AI mannequin, the result was a surprisingly cogent and interesting AI-generated podcast that touches on the character of artwork, the philosophy of consideration, and the human want to make which means out of the inherently meaningless.
Analyzing Poop & Fart written 1,000 times – Creating meaning from the meaningless
byu/sorryaboutyourcats in notebooklm
Once I asked NotebookLM to review my Minesweeper book last week, commenter Defenstrar smartly asked “what would occur if you happen to fed it a much less engrossing or effectively written physique of textual content.” The reply, as seen right here, exhibits the attention-grabbing instructions a contemporary AI mannequin can go in whenever you let it simply spin its wheels and wander away from an primarily unmoored start line.
“Generally a poop is only a poop…”
Whereas Google’s NotebookLM launched over a year ago, the mannequin’s recently launched “Audio Overview” feature has been getting a variety of consideration for what Google calls “a brand new method to flip your paperwork into participating audio discussions.” At its coronary heart is a LLM much like the sort that powers ChatGPT, which creates a podcast-like script for 2 convincing text-to-speech fashions to learn, full with “ums,” interruptions, and dramatic pauses.
Experimenters have managed to trick these AI-powered “hosts” into what appears like an existential disaster by telling them that they aren’t really human. And investigators have managed to get NotebookLM to talk about its own system prompts, which appear to deal with “going past surface-level data” to unearth “golden nuggets of information” from the supply materials.
The “poop-fart” doc (as I will be calling it for simplicity) is a fairly attention-grabbing check case for this type of system. In spite of everything, what “golden nuggets of information” may very well be buried past the “floor stage” of two scatological phrases repeated for a number of pages? How do you “spotlight intriguing factors with enthusiasm”—because the unearthed NotebookLM immediate suggests—when the doc’s solely oft-repeated factors are “poop” and “fart”?
Right here, NotebookLM manages to make use of that full lack of context as its personal start line for an attention-grabbing stream-of-consciousness, podcast-like dialog. After some throat-clearing about how the viewers has “outdone itself” with “a novel piece of supply materials,” the ersatz podcast hosts are fast to match the repetition within the doc to Andy Warhol’s soup cans or “minimalist music” that may be “surprisingly highly effective.” Later, the hosts attempt to glean some which means by evaluating the doc to “a contemporary dadaist prank” (pronounced as “daday-ist” by the speech synthesizer) or the vase/faces optical illusion.
Inventive comparisons apart, NotebookLM’s digital hosts additionally delve a bit into the psychology behind the “very human impulse” to “seek for a sample” on this “unintentional Rorschach check” and our tendency to “attempt to impose order” on the “data overload” of the world round us. In virtually the identical breath, although, the hosts get philosophical about “confront[ing] the absurdity in looking for which means in all the pieces” and counsel that “generally a poop is only a poop and a fart is only a fart.”