Google still recommends glue for your pizza | Hacker News Reader

Google still recommends glue for your pizza

theverge.com

44 pointslaktak22 comments

This exactly scenario is why our company is so afraid to put AI into production without the results being completely clear that it could be wrong. But if there’s even a chance that it could be wrong, why are we offering it to the user? How much due diligence does the user need to do? Does the benefits outweigh the cons?

The AI should be required to cite the source. But that wont work for LLM though as they are just random words put together in a way that is statistically probable.

I have asked chatgpt for sources, and it gave me links which actually worked.

In this case the problem is caused by citing sources - Google finds a list of links using their existing search tech and then produces a response using RAG based on the search results.

That’s where the problem is - it’s originally citing a Reddit post where someone recommends it as a joke, then Business Insider through citogenisis from the original story.

Pure LLMs (no RAG) don’t make this mistake - Claude will tell you it’s a bad idea and will taste bad.

In the past it was called sloppiness, today less and less people seem to care.

> But if there’s even a chance that it could be wrong, why are we offering it to the user?

Corporations push code written by fresh junior devs into production every day, breaking stuff that could cost them tens of thousands. Do they care? On paper, very much so, in practice, they dgaf.

Recent and related:

Google AI recommends adding Elmer's glue to pizza cheese after scanning Reddit - https://news.ycombinator.com/item?id=40448074 - May 2024 (135 comments)

Google scrambles to manually remove weird AI answers in search - https://news.ycombinator.com/item?id=40475578 - May 2024 (368 comments)

AI Overviews: About last week - https://news.ycombinator.com/item?id=40530833 - May 2024 (61 comments)

Its a step up from the dangers of gastric band surgery

Gluing your lips together while eating a pizza can only be good for your health and a great way to start dieting.

Google Answers directly from our advanced, human-level AI (artifical idiot).

Google, supposedly full of AI experts, doesn't understand what current AI is good for and its limitations. A lot of AI critics (and enthusiasts) expect AI to be a know-it-all that never fails and provides authoritative factual answers. But current technology AI isn't good at all for this.

I use AI everyday for giving me opinions, starting points for re/search, reviewing my work and brainstorming. AI is fabulous for that and was a life changer for me. Hallucinations are actually welcome because they provide new perspectives. And when AI fails me, that's not different from getting an unhelpful perspective from a friend.

I wish more people were focusing on what LLMs do best: augmenting human abilities rather than replacing them.

But then how do I replace that pesky labor? My MBA didn’t show me any other way to create value!

Just off-shore it, of course! Remember, there are two meanings for "AI".

Yes but do the letters 'AI' stand for, Jen?

I think Google understands perfectly that current AI is great at increasing the revenue produced by selling ads. Limitations are irrelevant if they have no impact on selling more ads. It takes a long time for the cycle of brand destruction to impact reported revenue.

If we still had Gödel I think he might prove that any sufficiently advanced intelligence must make mistakes. Just my intuition.

If we still had Gödel, I think he might prove that statistics can not be an effective substitute for cognition and logic.

Just my intuition.

Gemini: "Hey, don't diss it till you've tried it ;)"

Does this snippet feature only work in chrome?

The reason Google is uniquely bad at this is because Google poisoned its own well years and years ago. Google is SEO and SEO is Google. The whole company is inextricable from the concept of ranking based on "what generates the most advertising value and eyeballs", and not "what is actually verifiable, reasonable, evidence-based, or legitimate".

Google banked on those two things being similar enough in the early days of the internet to generate billions of dollars, but strayed so far from "don't be evil" that they no longer see that they were instrumental in pushing the two apart.

Google is actually better at being accurate then gpt. It's just they are starting over with the LLm's it seems and trying to play catch up by completely bypassing it's search logic.

> Google is actually better at being accurate then gpt.

Do you mean their snippets before the LLM era? They were shortcut crap, too. I remember when searching for the Dalai Lama, it showed: "Nationality: Chinese". But of course the current version is even worse.

Today

"When does the metro close?" Ask a colleague

Another one, with a laptop open "I can Google it"

He googles it, reads the preview and i immediately think "...i think that's wrong" and it was, in fact, wrong.

I mean, that's so simple and specific but it still got it wrong.

The ai boom is also a boom for mediocre journalists

Reply on news.ycombinator.com