I’m confused a little by the LLM’s and datasets here.
OpenAI and Reddit have their partnership for training, so I would have assumed the pizza glue answer would have come from an OpenAI result, but Google doesn’t use OpenAI. How did Google give an answer that was clearly scraped from Reddit?
There is a great Futurama episode that visualises this. The Late Philip J. Fry - Futurama S06E07
This is by the wonderfully weird artist: davidtherobot
Fuck cancer
I believe you when you say you’re not anorexic, or ever had anorexia, but as a 43 year old man, who is 185cm tall, my healthy range is 63-86 kilos.
I am 82 kilos give or take a meal, and if I lost 15 kilos I would be very, very slender, and that would still keep me well in the “healthy” range.
What I’m saying is the healthy range, to me at least, should be taken with a pinch of salt, and it’s shithouse that your diagnosis has vexed you for over ten years.
ChatGPT has entered the teenage years.
This is a really interesting issue within the industry. I work within the film and television industry, both in Australia and the UK, and have never been credited for the roles that I do that, some would argue, should be credited for.
But when you break down something like a mega-Hollywood-super-feature-blockbuster-film like Oppenheimer, the amount of ancillary and auxiliary bodies working on a movie like this, would amount to a credit role that would be very hard to keep track of.
For example, do you credit the data wranglers who might be assisting the DIT, or the courier service who delivers dailies to the post house? Or the team of edit assistants who make proxies for OCM? Do you credit every person who was on craft catering for the entire production run?
On one hand, I can see why it is “consolidated” to team leads, or heads of departments, but personally I think that ANYONE who worked on the film SHOULD be credited, as their time is equally as important as everyone else’s.
I have learnt to accept that my name won’t be appearing in any credit role, despite the hours I put in to make sure it happens, and I am okay with that, but totally understand and support anyone who feels like they should get the recognition… as I said at the start… interesting issue.
Thank you for the link. I didn’t realise that Google had a deal with Reddit as well, which explains why it was clearly indexed from Reddit.
I agree that AI doesn’t have a clue what an accurate response is. It’s just not sentient enough to differentiate between shitposting and fact. I also totally agree that an answer given from a search result HAS to be accurate, and we’re heading down a path of a misinformation super highway if LLMs are trained on incorrect data.