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

A new Google model is nearly perfect on automated handwriting recognition

generativehistory.substack.com

Post reply on HN
Print view View original

241–250 of 328 posts

Re: A new Google model is nearly perfect on automated handwriting recognition

#241

Post by DrewADesign »

tyre wrote: post #212
krackers wrote: post #125

Earlier quoted context omitted.

Every time a model is about to be released, there are a bunch of these hype accounts that spin up. I don't know they get paid or they spring up organically to farm engagement. Last time there was such hype for a model was "strawberry" (o1) then gpt-5, and both turned out to be meaningful improvements but nowhere near the hype. I don't doubt though that new models will be very good at frontend webdev. In fact this is…

My guess is that there are insiders who know about the models and can’t keep their mouths shut. They like being on the inside and leaking.

I’d also bet my car on there being a ton of AI product/policy/optics astroturfing/shilling going on, here and everywhere else. Social proof is a hell of a marketing tool and I see a lot of comments suspiciously bullish about mediocre things, or suspiciously aggressive towards people that aren’t enthused. I don’t have any direct proof so I could be wrong, but it seems more extreme than a iPhone/Android (though I suspect deliberate marketing forces there, too,) Ford/Chevy brand-based-identity kind of thing, and naive to think this tactic is limited to TikTok and Instagram videos. The crowd here is so targeted, I wouldn’t be surprised if a single-digit percentage of the comments are laying down plausible comment history facade for marketing use. The economics might make it worthwhile for the professional manipulators of the world.

Re: A new Google model is nearly perfect on automated handwriting recognition

#242

Post by cowboy_henk »

ACCount37 wrote: post #204
risyachka wrote: post #201

Earlier quoted context omitted.

Its always amusing when "an app like windows xp" considered hard or challenging somehow. Literally the most basic html/css, not sure why it is even included in benchmarks.

Those things are LLMs, with text and language at the core of their capabilities. UIs are, notably, not text. An LLM being able to build up interfaces that look recognizably like an UI from a real OS? That sure suggests a degree of multimodal understanding.

UIs made in the HyperText Markup Language are, in fact, text.

Re: A new Google model is nearly perfect on automated handwriting recognition

#243

Post by lern_too_spel »

roywiggins wrote: post #132

My task today for LLMs was "can you tell if this MRI brain scan is facing the normal way", and the answer was: no, absolutely not. Opus 4.1 succeeds more than chance, but still not nearly often enough to be useful. They all cheerfully hallucinate the wrong answer, confidently explaining the anatomy they are looking for, but wrong. Maybe Gemini 3 will pull it off. Now, Claude did vibe code a fairly accurate solution t…

This might be showing bugs in the training data. It is common to augment image data sets with mirroring, which is cheap and fast.

Re: A new Google model is nearly perfect on automated handwriting recognition

#244

Post by jofla_net »

terminalshort wrote: post #143
jstummbillig wrote: post #40

Earlier quoted context omitted.

I remain confused but still somewhat interested as to a definition of "novel", given how often this idea is wielded in the AI context. How is everyone so good at identifying "novel"? For example, I can't wrap my head around how a) a human could come up with a piece of writing that inarguably reads "novel" writing, while b) an AI could be guaranteed to not be able to do the same, under the same standard.

If a LLM had written Linux, people would be saying that it isn't novel because it's just based on previous OS's. There is no standard here, only bias.

Cept its not made Linux (in the absence of it).

At any point prior to the final output it can garner huge starting point bias from ingested reference material. This can be up to and including whole solutions to the original prompt minus some derivations. This is effectively akin to cheating for humans as we cant bring notes to the exam. Since we do not have a complete picture of where every part of the output comes from we are at a loss to explain if it indeed invented it or not. The onus is and should be on the applicant to ensure that the output wasn't copied (show your work), not on the graders to prove that it wasn't copied. No less than what would be required if it was a human. Ultimately it boils down to what it means to 'know' something, whether a photographic memory is, in fact, knowing something, or rather derivations based on other messy forms of symbolism. It is nevertheless a huge argument as both sides have a mountain of bias in either directions.

Re: A new Google model is nearly perfect on automated handwriting recognition

#245

Post by throwup238 »

otherdave wrote: post #219
throwup238 wrote: post #2

I really hope they have because I’ve also been experimenting with LLMs to automate searching through old archival handwritten documents. I’m interested in the Conquistadors and their extensive accounts of their expeditions, but holy cow reading 16th century handwritten Spanish and translating it at the same time is a nightmare, requiring a ton of expertise and inside field knowledge. It doesn’t help that they were of…

Where can I find these Conquistador documents? Sounds like something I might like to read and explore.

See here: https://news.ycombinator.com/item?id=45933750

Re: A new Google model is nearly perfect on automated handwriting recognition

#246

Post by DangitBobby »

otabdeveloper4 wrote: post #235
yomismoaqui wrote: post #207

Earlier quoted context omitted.

I implemented a receipt scanner to Google Sheet using Gemini Flash. The fact that it is ”intelligent" it's fine for some things. For example I created structured output schema that had a field "currency" with the 3 letter format (USD, EUR...). So I scanned a receipt from some shop in Jakarta and it filled that field with IDR (Indonesian Rupiah). It inferred that data because of the city name on the receipt. Would it…

> Would it be better for my use case that it would have returned no data for the currency field? Almost certainly yes.

Except in setups where you always check its work, and the effort from the 5% of the time you have to correct the currency is vastly outweighed due to effort saved from the other 95% of the time. Pretty common situation.

Re: A new Google model is nearly perfect on automated handwriting recognition

#247

Post by gf000 »

ninetyninenine wrote: post #165
snickerbockers wrote: post #104

Earlier quoted context omitted.

uhhh can it? I've certainly not seen any evidence of an AI generating something not based on its training set. It's certainly smart enough to shuffle code around and make superficial changes, and that's pretty impressive in its own way but not particularly useful unless your only goal is to just launder somebody else's code to get around a licensing problem (and even then it's questionable if that's a derived work or…

>I've certainly not seen any evidence of an AI generating something not based on its training set. There is plenty of evidence for this. You have to be blind not to realize this. Just ask the AI to generate something not in it's training set.

Like the seahorse emoji?

Re: A new Google model is nearly perfect on automated handwriting recognition

#248

Post by gf000 »

taneq wrote: post #209
snickerbockers wrote: post #76

Earlier quoted context omitted.

Generally novel either refers to something that is new, or a certain type of literature. If the AI is generating something functionally equivalent to a program in its training set (in this case, dozens or even hundreds of such programs) then it by definition cannot be novel.

OK, but by that definition, how many human software developers ever develop something "novel"? Of course, the "functionally equivalent" term is doing a lot of heavy lifting here: How equivalent? How many differences are required to qualify as different? How many similarities are required to qualify as similar? Which one overrules the other? If I write an app that's identical to Excel in every single aspect except tha…

I think the importance is the ability. Not every human have produced (or even can) something novel in their life, but there are humans who have time after time.

Meanwhile, depending on how you rate LLM's capabilities, no matter how many trials you give it, it may not be considered capable of that.

That's a very important distinction.

Re: A new Google model is nearly perfect on automated handwriting recognition

#249

Post by HarHarVeryFunny »

famouswaffles wrote: post #239
HarHarVeryFunny wrote: post #230

Earlier quoted context omitted.

You seem to be ignoring two things... First, the obvious one, is that LLMs are trained to auto-regressively predict human training samples (i.e. essentially to copy them, without overfitting), so OF COURSE they are going to sound like the training set - intelligent, reasoning, understanding, etc, etc. The mistake is to anthropomorphize the model because it sounds human, and associate these attributes of understanding…

You’re restating two claims that sound intuitive but don’t actually hold up when examined: 1. “LLMs just mimic the training set, so sounding like they understand doesn’t imply understanding.” This is the magic argument reskinned. Transformers aren’t copying strings, they’re constructing latent representations that capture relationships, abstractions, and causal structure because doing so reduces loss. We know this no…

> This is the magic argument reskinned. Transformers aren’t copying strings, they’re constructing latent representations that capture relationships, abstractions, and causal structure because doing so reduces loss.

Sure (to the second part), but the latent representations aren't the same as a humans. The human's world that they have experience with, and therefore representations of, is the real word. The LLM's world that they have experience with, and therefore representations of, is the world of words.

Of course an LLM isn't literally copying - it has learnt a sequence of layer-wise next-token predictions/generations (copying of partial embeddings to next token via induction heads etc), with each layer having learnt what patterns in the layer below it needs to attend to, to minimize prediction error at that layer. You can characterize these patterns (latent representations) in various ways, but at the end of the day they are derived from the world of words it is trained on, and are only going to be as good/abstract as next token error minimization allows. These patterns/latent representations (the "world model" of the LLM if you like) are going to be language-based (incl language-based generalizations), not the same as the unseen world model of the humans who generated that language, whose world model describes something completely different - predictions of sensory inputs and causal responses.

So, yes, there is plenty of depth and nuance to the internal representations of an LLM, but no logical reason to think that the "world model" of an LLM is similar to the "world model" of a human since they live in different worlds, and any "understanding" the LLM itself can be considered as having is going to be based on it's own world model.

> Saying “it understands nothing because autoregression” is just another unfalsifiable claim dressed as an explanation.

I disagree. It comes down to how do you define understanding. A human understands (correctly predicts) how the real world behaves, and the effect it's own actions will have on the real world. This is what the human is predicting.

What an LLM is predicting is effectively "what will I say next" after "the cat sat on the". The human might see a cat and based on circumstances and experience of cats predict that the cat will sit on the mat. This is because the human understands cats. The LLM may predict the next word as "mat", but this does not reflect any understanding of cats - it is just a statistical word prediction based on the word sequences it was trained on, notwithstanding that this prediction is based on the LLMs world-of-words-model.

Re: A new Google model is nearly perfect on automated handwriting recognition

#250

Post by fragmede »

roywiggins wrote: post #132

My task today for LLMs was "can you tell if this MRI brain scan is facing the normal way", and the answer was: no, absolutely not. Opus 4.1 succeeds more than chance, but still not nearly often enough to be useful. They all cheerfully hallucinate the wrong answer, confidently explaining the anatomy they are looking for, but wrong. Maybe Gemini 3 will pull it off. Now, Claude did vibe code a fairly accurate solution t…

and then, in a different industry, one that has physical factories, there's this obsession about getting really good at making the machine that makes the machine (product) being the route to success. So it's funny that LLMs being able to write programs to do the thing you want is seen as a failure here.
Post reply on HN