AI has never been 100% accurate to the documents it pretends to reference, so the fact that anybody could even consider reading it like it's not pure gibberish is mind boggling to me.

I sure would hope that those AI overviews are cached. Querying an LLMs for every repeated search sounds very wasteful

Me after losing 6 days of testing broken theories from an ai search when I could have just spent an hour watching a video that a human made that was fully functional from the start.

That said: No doubt that human spent way longer discovering the technique and then making the video: this should also be considered as a reality that creating something good takes time. And humans are still better at creating scalable content just give them the proper R&D time ffs.

This is nice but I never figured out the search suggestions query string.

I gave Claude a document to proofread today. It came back with an error. Clause said I used "told" instead of "tell." The error ended up being Claude's error, as I used "tell," and it was just confused.

This was the first time I ever used Claude. Just wanted to see what it was like. It did apologize, in fairness. But yeah, it's clearly fallible. I guess I don't understand it well enough to get how it read the document incorrectly. I'm baffled.

When the old new ways become the old ways and i cant keep up with the new new ways

It's because it's a statistical model that guesses the next word. You asked it to read a document and formulate a response. So, based on responses in the dataset to the documents in the dataset, it chose the words and phrases which were most likely to be in a response.

Statistically, the most likely result the machine could come up with, was that you had used the wrong word. Factually, you did not.

Encarta FTW

If it didn't do the AInsearch repeatedly they couldn't use the billions of daily searches to artificially boost their AI engagement metrics and further inflate this bubble.

Don't you dare duck with me!

LLMs fundamentally can't learn. They will continue to fail the same thing over and over until they guess and check into a response that appears coherent and correct.

That and TVTropes of course

That website was such a timesink. Amazing.

And had a useful answer about 1 in 20 times.

I have it disabled through an extension on my phone's Iceraven browser. Can't bother reading those. I need the Wikipedia/website info.

Wikipedia has an app. It's half of why I Google stuff. Usually a good place for quick answers, and an ok place to start digging in.

I'm half amazed at how bad Google has gotten, but that's probably just me remembering the early days of web searches.

Oh it is, it has the same issues as the first few iterations of Chat GPT if you ever used AI in the early days

If you need nuance and background, wikipedia is NOT the place to get that. Maybe you can get some good links at the bottom of the page to real information, but that can be hit or miss also.

It would have been a legal play IF the lineman had checked in with the referee as an "eligible" receiver before the play. But evidently he screwed up and report in. It's not a common play in the NFL, but every team has such a play on hand.

So that's kind of a "it needs context" question.

Let them waste money on generating an AI overview I'll happily skip on my way to the wikipedia link.

Yeah exactly. The AI was just wrong and said it's legal as long as you hand it off behind the line of scrimmage. I actually still have the screenshot he sent. Notice the 2nd bullet point.

I like the old ways because I also see/learn about unexpected things.

I've replaced search engines with Wikipedia as the first step in looking for information. skimming an article is much faster than opening and closing a dozen sloppy s.e.o. 'd websites :p

Or Qwant. It's Euroopan and doesn't even have the AI option.