Alexandra Merideth Erin [she/her] - in my Brain Damage arc: Brilliant. Fixing LLMs' inaccuracy by using already existing search technology to look up the real answer and then blend that answer into whatever the LLM generates, so the generated answer has a better chance of resembling the real answer, which was looked up but not shown to the querent.

Brilliant. Fixing LLMs' inaccuracy by using already existing search technology to look up the real answer and then blend that answer into whatever the LLM generates, so the generated answer has a better chance of resembling the real answer, which was looked up but not shown to the querent.

“The LLM is still guessing answers, but with RAG [retrieval augmented generation], it's just that the guesses are often improved because it is told where to look for answers…” “There's still no deep understanding of words and the world…”

Can a technology called RAG keep AI models from making stuff up?arstechnica.com The framework pulls in external sources to enhance accuracy. Does it live up to the hype?

Just like NFTs before them, all the ways people come up with to make LLMs useful involve stapling them to functional programs that don't need them and could do the job better without the unnecessary layer of overhyped scamware

"Can we make randomized text output tend towards resembling the truth" would be an interesting experiment if it weren't environmentally ruinous, but the best outcome is never going to be intelligence, nor as effective and efficient as a search engine

hey I heard you like googling so I googled your answer and then googled the answer I googled so you could google while you're googling your google

This is like those reality shows about dodgy restaurants, where it turns out they don't really have a kitchen and are just ordering takeaway from nearby competitors and serving it to their customers and pretending they cooked it themselves.

I am so exhausted by absolutely everything techbro adjacent.

No innovation, only slivering parts down

Just pouring billions of dollars into a bottomless pit.

Reminds me of early Silicon Valley episode where the failed app startup guy had the idea to make an app that remembered where your car was in a parking garage. “Why don’t you just write it down?”

I am not advocating for this as a solution, but it's baffling to me that Google didn't just implement it this way in the first place. When you have a tool that already can find the right answer, why wouldn't you just feed that answer to the natural language machine?

couldnt you save a step by just discarding the LLM

I'm literally not disputing that. But if the goal is "look we have the shiny new AI thing that kind of sounds like a person" that's not an option.

i'm pretty sure Google *did* do this; the results from their AI nonsense that people were laughing about recently were all taken from posts on various websites seemingly based on their pagerank or similar the problem is that popular isn't always right

Sounds like an unnecessary middleman, big tech.

And Chad the product manager was given more stock options after they explained this at the meeting

should call it BANDAID

Mechanical Turk cubed.

So what happens when the search results are just flooded with shitty LLM garbage?

I honestly can't imagine it mattering that much

This is so ass-backwards. Wonders never cease.

Congratulations, you made a search engine with more steps, and it STILL doesn't give the user what they want.

efficiency!

The llm's yearn to become machine centipedes

Aren't those search engines using LLMs to "enhance" their search results? Anyways this sounds like a royal kludge.

It's like +1 Magic LLM of Getting the Right Answer! Everybody likes getting a +1 on their dice rolls. (*sigh*)

Even if it’s more accurate, it’s still “borrowing” work done by actual people without their permission!

it's really telling that they only now got the idea to check their work. i mean i never need to edit or verify the correctness of my work, because i only shit out excellence

dang if only there were a way for people to look things up using a search engine without incinerating megawatts of computer power real shame 😅

Search engine uses AI to look up information ↓ AI uses search engine to refine results ↓ Search engine uses AI to look up information ↓ AI uses search engine to refine results ↓ Search engine uses AI to look up information ↓ AI uses search engine to refine results ↓ Search engine uses AI to ...

Sham looking less legitimate by the day #AiiHype

So… they reinvented Siri? That’s what I’m getting out of this. Siri, only worse.

Post