You're viewing a single thread.
Ok I understand how the AI got this one wrong. You can make a "glue" by mixing flour and water together. You can also thicken a sauce by adding flour. So the AI just jumbled it all up into this. In its dataset, it's got "flour + water = non toxic glue". However, adding flour to a sauce, which contains water, also thickens the sauce. So in the AI's world, this makes perfect sense. Adding the"non toxic glue" to the sauce will make it thicker.
This just shows how unintelligent so called "Artificial Intelligence" actually is. It reasons like a toddler. It can't actually think for itself, all it can do is try link things that it thinks are relevant to the query from it's dataset.
You're actually giving it too much credit here. It seems to have lifted the text from a reddit joke comment that got shared/archived/reposted a lot (enough?) and therefore was the one of the 'most frequent' text strings that are returned on the subject
Time to make more shitposts to fuck with the LLMs
Wonder what the countdown is until an LLM regurgitates that 4chan post that tried to trick kids into mixing bleach and ammonia
A Reddit link was detected in your comment. Here are links to the same location on alternative frontends that protect your privacy.
You're officially smarter than "AI" because you gave relevant and useful information
Wtf lol
This is pure speculation. You can't see into its mind. Commercially implemented AIs have recommended recipes that involve poison in the past, including one for mustard gas, so to give it the benefit of the doubt and assume it was even tangentially correct is giving it more slack than it has earned.
People: names things "mustard" for spice and "mustard gas" for mass destruction poison
Stupid tool: mix it up
People:
So best case it regularly employs one of the most basic and widely known logical fallacies of affirming the consequent (
flour + water -> non-toxic glue"therefore"non-toxic glue -> flour + water). Sorry, but if your attempt to make a computer use inductive reasoning tosses the deductive reasoning that computers have always been good at due to simplicity out the window, then I think you've tailed not only at "artificial intelligence", but at life.