Fuck AI

I didn't remember this so I looked it up.
According to GoogleGPT she purposely gave that recipe to teach the town how to make mustard gas but made a little whoopsie because instead of everyone getting to make their mustard gas it actually makes Chloramine.

And that it's a recurring theme of the show that people get the recipe wrong. Aw shucks, just not teaching people the correct way to make mustard gas.

I considered making a search engine based on verified real people, but I have not enough time or resources to create such a thing.

Finding human content is going to be like getting pre WW2 steel

One can only hope a modern library of Alexandria exists, cause soon they'll rewrite all information with ai on the internet and you won't even have sources anymore to countercheck ai results

I'm starting to think that ordinary people only provide the illusion of intelligence, too

Advanced predictive text

The scientists never called it an AI in the first place. It is still a Machine Learning Model. Zero intelligence behind that. The term was abused by idiotic managers. And one of them (at least) is already talking about AGI.

Coherent sentences have been a shorthand for "intelligent and thoughtful" for a very long time. Breaking that and forcing people to think about what they're reading is going to be extremely hard.

But people won’t invest trillions of dollars into Simulated Intelligence!

Just like everything on the internet it was trained on

Well it's trained off how people on the internet speak so....

ChatGPT has memories if you are logged in. I told it on my account to never use emojis or talk with any sort of personality. It saved it as a memory and it hasn't made me want to smash its face in after that.

CORRECTED STEP 2: THE SOAK OF RIGHTEOUSNESS

The tone of the whole "oops I tried to kill you" response is inappropriately unserious...like "not the witchy potion we want"? If a person was trying to be cute and stuff after making such a serious error, I think we would all be justifiably very annoyed.

The problem is not that this person asked chatgpt for cleaning tips (tbh it's pretty cringe to call someone lazy and stupid for trying to learn something 🙄 Have you seriously never looked up how to clean something weirdly specific? And I suppose those who weren't lucky enough to have parents who taught them how to adult properly are lazy and stupid when they try to learn?)

The bigger problem is that LLMs are being used to create content for the web. So now someone who knows they can't just mix any old chemicals together is going to Google whether bleach and vinegar are safe to mix and find a bunch of websites that have contradictory info.

These people, whether they use LLM to search or to create content, aren't even the root of the problem. Expecting that everyone is tech savvy enough to understand the limitations of generative AIs and how untrustworthy they can be is an unrealistic standard, especially in a world where everyone and their brother is using them and they seem like miracles of technology.

The responsibility lies with the companies that keep touting this technology as something it is not and who refuse to put meaningful limitations on them, and with governments who are dragging their feet in regulating them.

One of the ways people use chatGPT is like a search engine. Would you consider someone lazy and stupid for goggling how to clean correctly?

Bleach + ammonia = chloramine

Bleach + vinegar = chlorine gas

Both are harmful

You shouldn’t combine bleach with basically any acid, because the same reaction can occur to free the chlorine in bleach and create chlorine gas. Vinegar is a weak acid so it will probably create less by volume than a similar amount of ammonia would, but you don’t want to breathe it.