"The dumbest person you know is being told 'you are absolutely right' by an AI right now."

[slotos@toot.community]

@CressSalad @AlSweigart Is this the new version of „blockchain elections”?

[cresssalad@mastodon.social]

@slotos @AlSweigart

Did not know about this until I read an article - thats interesting:

Blockchain-Based Voting Systems - Blockchain Council

Blockchain voting systems offer secure, transparent elections, reducing fraud and enhancing voter confidence.

Blockchain Council (www.blockchain-council.org)

Personally I think we just as a species/planet need to get human emotions out of the picture in making big decisions - like digging up all the last oil, starting wars etc. We have to have some level headed advice and decisions, OK, we can put human gate-post on it. But it should never be 1 individual or party. I would even say for local issues it can be left alone!

[dabeaz@mastodon.social]

@AlSweigart And if you can't spot that person... well, you know.

[rupertreynolds@hachyderm.io]

@AlSweigart In the spirit of "know your enemy" I asked (an LLM coding assistant) to write a C99/x11 function I could call when an alarm or countdown triggers in my music player. I explained that the app is in dark mode and I'd like it modal, with a few pastel colours to be eye-catching. I forget the exact words.

Never once did it say "That's a terrible idea!"

It ran first time. But let's be frank it *was* a terrible idea

It perpetrated this:
(the coloured bands move slowly, BTW)

[okuna@social.tchncs.de]

@AlSweigart In the US half of the adults do it too

[kasperd@westergaard.social]

One needs to be able to judge whether the output from an AI is trustworthy or not. Anyone who cannot correctly make that judgement isn't smart enough to be using AI.

[okuna@social.tchncs.de]

@ozzelot @SnoopJ @AlSweigart once a doctor told me I have to give up drinking, cigarettes and too much meat.

I never drank, never smoked and was vegetarian.
Now I am prepared for the next doctors visit to oblige to his request
SCNR

[okuna@social.tchncs.de]

@kasperd @n_dimension @AlSweigart in my AI classes I usually start with this:
I dont need AI.
If I have no clue, I can’t use AI because I am not sure it is lying. If I know the topic, I dont need AI anyway.

[okuna@social.tchncs.de]

@bbbhltz @AlSweigart did you see the outcry when openAI closed chatgpt 4o or so?
UNBELIEVABlE

[okuna@social.tchncs.de]

@dabeaz @AlSweigart Only a Buddha recognizes a Buddha

[bbbhltz@framapiaf.org]

@Okuna @AlSweigart My students told me about it and my mother in law complained about it to my partner. There was certainly a moment of panic. At the time my mother in law was using ChatGPT to prepare a legal thing so she was upset.

[the_wub@mastodon.social]

@AlSweigart Look on the bright side. MAGA types with nascent anti-vaxer tendencies will have their conspiracy theory based suspicions confirmed.

It might take a generation or two but, given that there will inevitably be more pandemics, these pandemics will tend to reduce the number of people of voting age that promulgate MAGA style politics.

[chu@climatejustice.social]

@AlSweigart

Chatgpt is probably making Trump's war plans right now

[bradthornborrow@thecanadian.social]

@AlSweigart @briankrebs Sam Altman?

[hydropsyche@ecoevo.social]

@criticalangle @AlSweigart chatbots don't know what "facts" are

[kasperd@westergaard.social]

Sometimes I know enough about a topic to verify the correctness of an answer. In those cases there is less than 10% chance an AI can give me a useful answer to the question. But there is a non-zero chance.

One example from my own experience was that I wanted to know the git equivalent of this command hg log -r 'children(.)'. I have searched for an answer using search engines. I have asked some people who are promoting use off git. All of that lead me to suggestions that would only produce incomplete outputs. And for years nobody I asked could suggest a fix for this.

Eventually I asked an AI, which initially gave me the same answers I had seen before which did not work. After pointing out 2-3 times that the suggested solution didn't work, the AI eventually managed to give me a useful suggestion. It hadn't produced a fully working solution, but it had given me a different approach that I was able to turn into something working by myself. The key point was to use the command git cat-file --batch-all-objects --batch-check as part of my solution.

I wouldn't have managed to find that solution without the help of AI. But using the AI certainly wasn't a quick and easy way to a solution either. I had to go through loads of useless output from the AI to pinpoint that one tiny piece of information that was useful.

So my conclusion is that current LLM are sometimes better than nothing, which is a pretty low bar to meet. And they are very far from the all-knowing infallible oracles that proponents say they are.