Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: Gemini Pro 3 imagines the HN front page 10 years from now

dosaygo-studio.github.io

Post reply on HN
Print view View original

911–920 of 1001 posts

Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now

#911

Post by Workaccount2 »

Aurornis wrote: post #876
ActivePattern wrote: post #869

Earlier quoted context omitted.

I've never heard the caveat that it can't be attributable to misinformation in the pre-training corpus. For frontier models, we don't even have access to the enormous training corpus, so we would have no way of verifying whether or not it is regurgitating some misinformation that it had seen there or whether it is inventing something out of whole cloth.

> I've never heard the caveat that it can't be attributable to misinformation in the pre-training corpus. If the LLM is accurately reflecting the training corpus, it wouldn’t be considered a hallucination. The LLM is operating as designed. Matters of access to the training corpus are a separate issue.

I believe it was a super bowl ad for gemini last year where it had a "hallucination" in the ad itself. One of the screenshots of gemini being used showed this "hallucination", which made the rounds in the news as expected.

I want to say it was some fact about cheese or something that was in fact wrong. However you could also see the source gemini cited in the ad, and when you went to that source, it was some local farm 1998 style HTML homepage, and on that page they had the incorrect factoid about the cheese.

Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now

#912

Post by doe88 »

jll29 wrote: post #825

AI professor here. I know this page is a joke, but in the interest of accuracy, a terminological comment: we don't call it a "hallucination" if a model complies exactly with what a prompt asked for and produces a prediction, exactly as requested. Rater, "hallucinations" are spurious replacements of factual knowledge with fictional material caused by the use of statistical process (the pseudo random number generator u…

Exactly! At first this is the precise reason I didn't click through as I thought from the title, a page must have been somehow outputted/hallucinated by error, but luckily I then saw the number of votes, revised my choice and saw a great page.

I'm partial though, loving Haskell myself (as a monad_lover) i'm happy it wasn't forgotten too :)

Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now

#914

Post by leecarraher »

julienreszka wrote: post #870
articlepan wrote: post #843

Earlier quoted context omitted.

I agree with your first paragraph, but not your second. Models can still hallucinate when temperature is set to zero (aka when we always choose the highest probability token from the model's output token distribution). In my mind, hallucination is when some aspect of the model's response should be consistent with reality but is not, and the reality-inconsistent information is not directly attributable or deducible fr…

That's because of rounding errors

i agree, not just the multinomial sampling that causes hallucinations. If that were the case, setting temp to 0 and just argmax over the logits would "solve" hallucinations. while round-off error causes some stochasticity it's unlikely to be the the primary cause, rather it's lossy compression over the layers that causes it.

first compression: You create embeddings that need to differentiate N tokens, JL lemma gives us a bound that modern architectures are well above that. At face value, the embeddings could encode the tokens and provide deterministic discrepancy. But words aren't monolithic , they mean many things and get contextualized by other words. So despite being above jl bound, the model still forces a lossy compression.

next compression: each layer of the transformer blows up the input to KVQ, then compresses it back to the inter-layer dimension.

finally there is the output layer which at 0 temp is deterministic, but it is heavily path dependent on getting to that token. The space of possible paths is combinatorial, so any non-deterministic behavior elsewhere will inflate the likelihood of non-deterministic output, including things like roundoff. heck most models are quantized down to 4 even2 bits these days, which is wild!

Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now

#915

Post by stingraycharles »

Workaccount2 wrote: post #894
hbn wrote: post #890

Earlier quoted context omitted.

"Hallucination" has always seemed like a misnomer to me anyway considering LLMs don't know anything. They just impressively get things right enough to be useful assuming you audit the output. If anything, I think all of their output should be called a hallucination.

We don't know if anything knows anything because we don't know what knowing is.

This is just something that sounds profound but really isn’t.

Knowing is actually the easiest part to define and explain. Intelligence / understanding is much more difficult to define.

Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now

#916

Post by rat9988 »

iberator wrote: post #97
dragonwriter wrote: post #38

Earlier quoted context omitted.

Consider the subject of the post, not just the username in isolation.

I don't get it

The user seems to be struggling at leetcode, so he needs some persistence to get there until he is able to do it.

Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now

#919

Post by GFischer »

thomasm6m6 wrote: post #598
pseudosavant wrote: post #342

Earlier quoted context omitted.

This is one of the greatest LLM creations I've ever seen. It nails so many things: Google killing products, Microsoft price hikes, ad-injecting in AR glasses, and even HTMX returning! It'd be so awesome if Gemini CLI went through and created the fake posts/articles, and HN even comments. Perhaps a bit much to ask of it?

Here it is: https://sw.vtom.net/hn35/news.html I downloaded the original article page, had claude extract the submission info to json, then wrote a script (by hand ;) to run feed each submission title to gemini-3-pro and ask it for an article webpage and then for a random number of comments. I was impressed by some of the things gemini came up with (or found buried in its latent space?). Highlights: "You’re probably…

This is great.

Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now

#920

Post by Miner49er »

asveikau wrote: post #205
Jaygles wrote: post #157

I opened this, walked away from my computer, then came back and clicked on the Debian 18 link wondering how the hell did I miss 14-17

They also already used Debian trixie for 13.0. I wonder if it's an intended joke that they ran out of names and needed to re-use some.

There's no way they run out of names for a good while. There's over 50 Toy Story characters at this point.
Post reply on HN