Earlier quoted context omitted.
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…
SQLite 4.0 Release Notes: https://sw.vtom.net/hn35/pages/90098234.html Another absolute gem: Columns now support "Vibe" affinity. If the data feels like an integer, it is stored as an integer. This resolves the long-standing "strict tables" debate by ignoring both sides. Also: SQLite 4.0 is now the default bootloader for 60% of consumer electronics. The build artifacts include sqlite3.wasm which can now run bare-meta…
Show HN: Gemini Pro 3 imagines the HN front page 10 years from now
931–940 of 1001 posts
Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now
#932Earlier 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…
> In my mind, hallucination is when some aspect of the model's response should be consistent with reality By "reality", do you mean the training corpus? Because otherwise, this seems like a strange standard. Models don't have access to "reality".
This is an explanation of why models "hallucinate" not a criticism for the provided definition of hallucination.
Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now
#933AI 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…
"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.
If you pick up a dictionary and review the definition of "hallucination", you'll see something in the lines of "something that you see, hear, feel or smell that does not exist"
https://dictionary.cambridge.org/dictionary/english/hallucin...
Your own personal definition arguably reinforces the very definition of hallucination. Models don't get things right. Why? Because their output contrasts with content covered by their corpus, thus outputting things that don't exist or were referred in it and outright contrast with factual content.
> If anything, I think all of their output should be called a hallucination.
No. Only the ones that contrast with reality, namely factual information.
Hence the term hallucination.
Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now
#934AI 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…
Terminology-wise, does this read like a better title instead?: Show HN: Gemini Pro 3 generates the HN front page 10 years from now
Generates does not convey any info on the nature of the process used to create the output. In this context, extrapolates or predicts or explores sound more suitable.
But nitpicking over these words is pointless and represents going off on a tangent. The use of the term "hallucination" reffers to the specific mechanism used to generate this type of output. Just like prompting a model to transcode a document and thus generating an output that doesn't match any established format.
Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now
#935Earlier quoted context omitted.
This common response is pretty uninteresting and misleading. They simply predict tokens? Oh. What does the brain do, exactly?
It does exactly the same, predicts tokens, but it's totally different and superior to LLMs /s OTOH, brain tokens seem to be concept based and not always linguistic (many people think solely in images/concepts).
That is an absolutely wild claim you've made. You're being way to presumptious.
Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now
#936Earlier quoted context omitted.
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…
Comedy gold. The whole "Right to Human Verification" Act bit sound eerily plausible and realistic, but then we get this bit: "A recent Eurobarometer survey showed that 89% of Europeans cannot tell the difference between their spouse and a well-prompted chatbot via text." Also I bet this will become a real political line in less than 10 years: "A European citizen has the right to know if their customer service represe…
Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now
#937Earlier quoted context omitted.
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…
SQLite 4.0 Release Notes: https://sw.vtom.net/hn35/pages/90098234.html Another absolute gem: Columns now support "Vibe" affinity. If the data feels like an integer, it is stored as an integer. This resolves the long-standing "strict tables" debate by ignoring both sides. Also: SQLite 4.0 is now the default bootloader for 60% of consumer electronics. The build artifacts include sqlite3.wasm which can now run bare-meta…
This is brilliant. Well done.
Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now
#938AI 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…
Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now
#939This is awesome, but minor quibble with the title - "hallucinates" is the wrong verb here. You specifically asked it to make up a 10-year-in-the-future HN frontpage, and that's exactly what it did. "Hallucinates" means when it randomly makes stuff up but purports it to be the truth. If some one asks me to write a story for a creative writing class, and I did, you wouldn't say I "hallucinated" the story.
Re: Show HN: Gemini Pro 3 imagines the HN front page 10 years from now
#940Edit: See also this current thread: Auto-grading decade-old Hacker News discussions with hindsight - https://news.ycombinator.com/item?id=46220540