You don't need to be an 'investor' to invest in Singletrack: 6 days left: 95% of target - Find out more
Just a bit of fun, and also to collect examples of dumb stuff AI, even the well trained models, do when asked seemingly simple questions.
How many Rs in Strawberry?
Pictures of people with 6 fingers and 3 legs is too obvious for this thread.
Please name the model used, mine was done on ChatGPT but most other models fail the test too.
I googled 'how thick are TRP brake rotors' the other day.
Googles AI result told me that mountain bike brakes discs are usually between 160mm and 220mm in diameter and are usually between 45 and 55mm thick.
Didn't fancy the weight penalty on those so stuck with my Shimanos.
I use AI (well co pilot) as a fancysearch/ wiki/ web scraper.
Ask it a question then use the links in the result.
I did an accreditation qual before the summer and it saved a lot of searching for theories and references.
Have used to to try and generate images and gave up.
I think Artificial Intelligence should be shortened to Artificial. As we all know, it is the latest ponzi scheme by the IT industry (see: cloud) to get us all fired up about using a word, governments and businesses to spend lots of money on contractors and more, all so that a computer can badly re-word the words which are already online. Shite doesn't even come close.
Garbage in, garbage out. When you send your AI to gather its wisdom from Reddit, what do you expect?
Bearnercessities - I like it 🙂 .For thoswho don't want to click the link:
[i]Yes, WCA on Singletrackworld.com stands for "Worst Case Advice." It's a long-running and humorous thread where members of the Singletrackworld mountain biking community share the most pessimistic, exaggerated, or tongue-in-cheek advice on various topics, usually related to biking or outdoor activities.
The thread has become a bit of a tradition on the forum, with users enjoying the playful and often absurd scenarios presented as "advice" in response to different questions or situations. It’s all in good fun, and the advice is intentionally over-the-top, not meant to be taken seriously.
If you enjoy dark humor or need a laugh, it's a good thread to check out![/i]
The fun thing is that more and more internet content is AI generated and contains this kind of rubbish - as well as fictious references BigDean, but most people can spot the absolute idiot statements.
The frightening thing is that the AI is being trained on the internet content and it cannot spot stupidity, and so repeats it so gradually increasing the amount of rubbish repeated. The more it sees, the more it repeats iut and the more the other AI engines ingest it during their training.
The frightening thing is that the AI is being trained on the internet content and it cannot spot stupidity, and so repeats it so gradually increasing the amount of rubbish repeated. The more it sees, the more it repeats iut and the more the other AI engines ingest it during their training.
LIke a turd rolling downhill, a rolling turd gathers more crap, as the proverb says...
As we all know, it [AI] is the latest ponzi scheme by the IT industry (see: cloud) to get us all fired up about using a word,
It's like everything - used properly and in the right way it is incredible.
https://www.bbc.co.uk/news/technology-68607059
Or give it to idiots, and set them loose on an unqualified data set, and stand back and watch the chaos unfold 😉
There is an EV forum I dip into and there was an AI/bot thread started asking about upgrading the car and what exhaust/turbo to get and how they impact on the cars fuel economy/emissions.
https://www.cuprabornforums.com/viewtopic.php?f=8&t=1310
As we all know, it [AI] is the latest ponzi scheme by the IT industry (see: cloud) to get us all fired up about using a word,
I did project with a museum housing early Television technology - amongst the collection was an old book called 'The Television Quiz Book'
Not a book about television quizzes. Not a quiz book about television. Apart from on the front cover there was no mention in the book of television at all. "Television' was just the buzz word of the day
There's a long way to go yet....
(Tesla lying/playing hard to seek with the truth when 'launching' some 'AI' 'Robots'....)
Still going real well...Apple AI scrapes BBC and NY Times headlines wrong...
https://www.bbc.co.uk/news/articles/cx2v778x85yo
I'd be careful. When it takes over the world it will remember how you ridiculed it.
@maccruiskeen I read something the other day where they drew a parallel with current trend of adding AI to stuff with a 1930s craze for putting radioactive elements into everything as it was a new exciting technology.
Uranium makeup anyone?
Radium condoms FTW......
And radium suppositories (claimed to have similar effect to Viagra), biscuits, face cream, beer, toothpaste. Lots of interesting 1920s stocking fillers
One of the problems for training an AI on news summaries will be the source of the data. The internet is awash in crazy misinformation and a lot of users will want the AI to deliver reports that confirm their view of the world. If they limit the training to sources like the NY Times, Washington Post, etc., then the believers in right-wing conspiracies will assume it's "fake news", but if they let in the conspiracy sites to please their customers, it'll just turn to shit.
^^^ that's not the issue with the Apple implementation. It's just shit 🙂 It affects all their AI summaries, not just News. Basically when you get more than 1 notification from an app the phone will attempt to summarise them, but it will mix up parts from different notifications together. Not very helpful generally but potentially quite distressing in the case of news summaries!
Yesterday I was discussing SlipKnot / Barbara Dickson's covers with BigScotNanny*
I asked Google about this and got false information that mixed up multiple Barbara Dicksons.
* Back story: BSN had offered to lend Thud a record player to try out and Thud was playing his usual mix of Willie Nelson / Skepta through his new-to-him Hifi. This then led to SlipKnot / Barbara Dickson's in the manner of Prodigy/ Mannfred Mann.
Yeah, it opens up all sorts of liabilities. Two that the BBC highlighted were 'Luigi Mangioni kills himself' (untrue, reputational damage for the BBC), and 'Rafa Nadal comes out as gay' (untrue AND libellous, potentially exposing the BBC to being sued).
A right old mess.
Yup... AI summaries are shit everywhere (not unique to Apple), it's just obvious and universal when the news story is the opposite of the summary. Lots of people can see and report the same nonsense, giving the original source the knowledge and reason to listen and act on complaints. People will be getting unique nonsense about non-news on their phones, and when searching, as well... it's just harder to check, disparate, and not clearly attributed to sources. How does all this nonsense get fact checked and corrected...?
I googled London Bridge the other day... all the photos and the summary text were of and about Tower Bridge.... because the "hive mind" it's drawing from thinks that is London Bridge. LLM don't deliver the truth, they give the expected result based on the dataset. The is where expert systems rather than pure LLM should be the future of AI... but considering the relative costs, necessary human involvement, and who is pushing AI, LLM will be the big news for a good while.
Generative AI is shit when truth matters. That’s inherent to the design. It’s great for mood music when all you want is space-filler that looks roughly plausible.