ChatGPT can get in the fucking bin. Fuming.

[noblegiraffe]

There was a thread on twitter asking people to see what ChatGPT said about them, so I tried it with my MN username. After correctly identifying that I was a well-respected member of the MN community I decided to ask it some more direct questions.

Anyway, ChatGPT confidently told me that I was an openly Conservative voter who regularly expresses right wing views and then that I voted for Nigel fucking Farage in 2015.

What. The. Fuck.

I have to say that any remaining faith I had in AI has been shattered by this.

Better ways to control the information highway, and build and design far faster and far better and combined with human intellect, potential increase our knowledge and innovation 100 fold

Glad @ComeTheFckOnBridget has answered but still surprised when people don’t know. Particularly annoying when people (not meaning you but people IRL) bang on about the environment without a clue about the most damaging things.

I also asked Copilot how I voted and it said "I couldn’t find specific information about how the Mumsnet user “parkrun500club” voted in the last UK General Election. Voting records are private, and unless someone publicly shares their voting choices, this information isn’t available."

It clearly isn't scraping data to the same extent as Chat GPT.

That's understandable, my knowledge of data centres is old mines and underwear ones at the moment, however I do know of some traditional built in warehouses etc

A chatgpt query uses up to 25 times more energy than a google one.

We could do that.

However we won't ...

Who cares about the truth ? Now that form of words exists, it will be factored out, built up, refactored and incorporated into ChatGPTs next iteration of "the real world".

Who knows - asking ChatGPT about SerendiptyJane (UK security service asset)in a few days might just resolve to ...

JUst asked it about one of my previous usernames: Mumsnet member REDACTED is known for her infamous post about keeping a pee-soaked mattress for her grandchildren to sleep on when they visited. The post, which was shared on the AIBU (Am I Being Unreasonable) forum, described how she continued to use a mattress that had been urinated on by a grandchild without cleaning it, and she insisted it was still fine for use. This sparked a significant amount of backlash from other Mumsnet members, who found the situation unhygienic and unsuitable for children. The post became a notorious example of unusual and controversial advice on the platform, often referenced in Mumsnet discussions as part of its internet folklore.

Which is odd, because I don't have any granchildren....
Load of bollocks isn't it?

I could say if only students used common sense like that, but then it might be harder to spot…
At least the howlers make marking amusing.

Garbage in, Garbage out.

It's becoming rapidly apparent that training AI on the entirety of the internet, when the internet is objectively terrible, might not have been a good idea.

Although in the case of my MN posts it was 'Excellence in, Garbage Out', so perhaps the AI is at fault here too.

You and @OliveWah need to form a support group. Why are some posters having these outrageous stories made up about them?!

That would really hack me off, that I was so badly and factually incorrectly was reported.

Hah, just saw this on twitter.

Although I'm surprised that what was on the internet before AI filled it with slop was considered higher quality. There was plenty of slop there already!

AI is literally a load of pointless hot air.

And it's named badly. It's artificial stupidity not intelligence.

Chatgpt is quite good depending on what you copy into it as I've been copying eg encyclopedia information and asking the ai to expand on the information etc and it depends on the extra settings as to how you want chatgpt to respond with your questions etc.

It's just clever (for some values of "clever") pattern matching done big and fast. The sort of thing that impresses some men I guess.

It is already using it's pattern matching to winnow out the credulous.

You can also do that with images. Asking the AI to expand out, e.g., a photo of party of a room to show the rest of the room. Or if you want to make a small rehearsal space look like a concert hall you can ask the AI to "continue" the seats etc beyond the frame you give it. All great. There is no guarantee of "truth" though - the concert hall is not real, the room is a guess. When people use it to get expanded info, it's no more reliable than asking a random bloke down the pub (it might be true, it might not be but it will be said with confidence).

i think id trust Ai alot more with the information than the majority of mumsnet (pub) information, yes its a starting point but as when doing intelligence analysis as long as other sources match your original intell then it stands to reason of the accuracy of it

https://eastangliabylines.co.uk/technology/astonishing-results-but-is-this-intelligence/

You can develop complex algorithms to analyse data. You can create "learning algorithms" that collect data, analyse data etc and improve based on the feedback of users (this would work in more controlled environments like specific workplaces). That's not what Chat GPT is designed to do at all though. Its designed to entertain, and "pass" the Turing test. If I said "exercise is good for you, it gets colder in winter, 911 was an inside job" - do you genuinely think my third statement is made more true because the first two were? People make that mistake about other people (influencers) all the time. Lets not do the same for Chat GPT.

the third one may be less bias because of the amout of datasets it can pull from , you can run the physics, the various other details you can refine etc overall chatgpt may even give better intelligence analyst than what the main stream media spin is, especially how the cold war era showed how cozy the main stream media was with various aspects so whos to say that chatgpt could be more accurate unless of course its restricted in some ways of certain info

besides documentary's are the same argument, two documentary's on eg water tornados then one documentary on conspiracy's the first two would be most likely to be fully accurate, the third very debateable @biscuitandcake

It's axiomatic that "AI" can (and will) lie. Worth bearing in mind.

One thing that "AI" is good at for now (I predict this will fade once the spivs get their sticky little mitts on it, so enjoy it while it lasts) is to provide a name for an item from a description. Especially useful if you are seeking something niche.