This isn’t going to hurt Google’s antitrust cases at all… Noooo sir.
This isn’t going to hurt Google’s antitrust cases at all… Noooo sir.
Saw one the other day with Jennifer Anniston. Good enough that it took a second to realize it was deep fake audio and video.
Ignoring the content of the words, they probably spaced it out like that so they could still see out the rear view mirror.
That whooshing sound, in case you were curious, was the point sailing right over your head.
This is how to tell someone you haven’t checked grocery prices lately without actually telling them you haven’t checked grocery prices lately. A box of mediocre pasta alone is going to cost you $1.75. A jar of Preggo will run you another 2.50. So 4.25 for an I hate life spaghetti and marinara meal.
The sauce they make probably doesn’t come out of a jar of reconstituted tomato paste and dried seasonings either.
If you buy decent ingredients you are looking at $3 for the pasta and $9 for the sauce. Or $12 for an “ok for a home cook” spaghetti meal with no protein.
Restaurant serving sizes (for better or worse) are usually 2x+ larger than you would serve at home. Rent isn’t free for the restaurant either. Or labor. Or utilities. Or equipment. Etc. General rule of thumb is that a restaurant needs to charge 3x raw food costs to cover expenses.
So your I hate life pasta would need to be priced at $6.50 and your ok for home but not something I would be happy with getting at a restaurant pasta would need to be priced at $18.
They’re supposed to be good a transformation tasks. Language translation, create x in the style of y, replicate a pattern, etc. LLMs are outstandingly good at language transformer tasks.
Using an llm as a fact generating chatbot is actually a misuse. But they were trained on such a large dataset and have such a large number of parameters (175 billion!?) that they passably perform in that role… which is, at its core, to fill in a call+response pattern in a conversation.
At a fundamental level it will never ever generate factually correct answers 100% of the time. That it generates correct answers > 50% of the time is actually quite a marvel.