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

Gemini 3

blog.google

Post reply on HN
Print view View original

981–990 of 1001 posts

Re: Gemini 3

#981

Post by orly01 »

recitedropper wrote: post #411
orly01 wrote: post #326

Earlier quoted context omitted.

Why should it have to be expensive computationally? How do brains do it with such a low amount of energy? I think catching the brain abilities even of a bug might be very hard, but that does not mean that there isn't a way to do it with little computational power. It requires having the correct structures/models/algorithms or whatever is the precise jargon.

This is the million dollar question. I'm not qualified to answer it, and I don't really think anyone out there has the answer yet. My armchair take would be that watt usage probably isn't a good proxy for computational complexity in biological systems. A good piece of evidence for this is from the C. elegans research that has found that the configuration of ions within a neuron--not just the electrical charge on the…

My whole point is that it maybe possible to do perception using a lot of computational power, or alternatively, there could be another kind of smart ideas that allows to do it in a diferent way with much less computation. It is not clear it requires it.

Re: Gemini 3

#982

Post by jsmo »

simonw wrote: post #641
torginus wrote: post #634

Earlier quoted context omitted.

Considering how important this benchmark has become to the judgement of state of the art AI models, I imagine each AI lab has a dedicated 'pelican guy', a a highly accomplished and academically credentialed person, who's working around the clock on training the model to make better and better SVG pelicans on bikes.

That would mean my dastardly scheme has finally come to fruition: https://simonwillison.net/2025/Nov/13/training-for-pelicans-...

This reply needs more upvotes ^_^

Re: Gemini 3

#983

Post by jarjoura »

oezi wrote: post #501
epolanski wrote: post #464

Earlier quoted context omitted.

Imho Gemini 2.5 was by far the better model on non-trivial tasks.

To this day, I still don't understand why Claude gets more acclaim for coding. Gemini 2.5 consistently outperformed Claude and ChatGPT mostly because of the much larger context.

Gemini 2.5 and now 3 seem to continue their trend of being horrific in agentic tasks, but almost always impress me with the single first shot request.

Claude Sonnet is way better about following up and making continuous improvements during a long running session.

For some reason Gemini will hard freeze-up on the most random queries, and when it is able to successfully continue past the first call, it only keeps a weird summarized version of its previous run available to itself, even though it's in the payload. It's a weird model.

My take is that, it's world-class at one-shotting, and if a task benefits from that, absolutely use it.

Re: Gemini 3

#985

Post by decide1000 »

We hire a developer to build parsers for a complicated file format. It takes a week per parser. Gemini 3 is the first LLM that is able to create a parser from scratch, and it does it very well. Within a minute, 1-shot-right. I am blown away.

Re: Gemini 3

#986

Post by tkgally »

hbbio wrote: post #930
tkgally wrote: post #885

Earlier quoted context omitted.

Thanks for the suggestion. I’m not sure why I didn’t include an OpenAI model in my first round. Here’s the updated page with GPT-5.1 results added: https://gally.net/temp/20251107pelican-alternatives/index.ht... As your example shows, GPT-5 Pro would probably be better that GPT-5.1, but the tokens are over ten times more expensive and I didn’t feel like paying for them.

Thanks for adding! Extending beyond the pelican is very interesting, especially until your page gets enough recognition to be "optimized" by the AI companies. It seems both Gemini 3 and latest ChatGPTs get a deep understanding of the representation of SVGs that seems a difficult task. I would be incapable of writing a SVG without visualizing the result and a graphical feedback loop. PS: Would be fun to add "animated"…

I would also be unable to write SVG code to produce anything other than the simplest shapes.

I noticed that, on my page, Gemini 3.0 Pro did produce one animated SVG without being asked, for “#8Generate an SVG of an elephant typing on a typewriter.” Kind of cute, actually.

As for whether the images on the page will enter LLM training data: In the page’s HTML are meta tags I had Claude give me to try to prevent scraping:

Who knows if they will work, though.

Re: Gemini 3

#987

Post by JohnKemeny »

CMay wrote: post #633

I was sorting out the right way to handle a medical thing and Gemini 2.5 Pro was part of the way there, but it lacked some necessary information. Got the Gemini 3.0 release notification a few hours after I was looking into that, so I tried the same exact prompt and it nailed it. Great, useful, actionable information that surfaced actual issues to look out for and resolved some confusion. Helped work through the logic…

This ad was brought to you by DeepMind™. Changing people's lives.

Re: Gemini 3

#988

Post by fragmede »

jacquesm wrote: post #921
red75prime wrote: post #877

Earlier quoted context omitted.

Unlike general public the models can be trained. I mean if you train a member of general public, you've got a specialist, who is no longer a member of general public.

Unlike the general public though, these models have advanced dementia when it comes to learning from corrections, even within a single session. They keep regressing and I haven't found a way to stop that yet. What boggles the mind: we have gone for so long to try to strive for correctness and suddenly being right 70% of the time and wrong the remaining 30% is fine. The parallel with self driving is pretty strong here…

> the killer feature for me is an indication of the confidence level of the output.

I don't think I did something special too ChatGPT to get it to do this, but it's started reporting confidence levels to me, eg from my most recent chat:

> In China: you could find BEVs that cost same or even less than ICE equivalents in that size band. (Confidence ~0.70)

Re: Gemini 3

#989

Post by fragmede »

jorvi wrote: post #775
windexh8er wrote: post #740

Earlier quoted context omitted.

If we've learned anything so far it's that the parlor tricks of one-shot efficacy only gets you so far. Drill into anything relatively complex with a few hundred thousand tokens of context and the models all start to fall apart roughly the same. Even when I've used Sonnet 4.5 with 1M token context the model starts to flake out and get confused with a codebase of less than 10k LoC. Everyone seems to keep claiming thes…

Same. I've been needing to update an userscript (JS) that takes stuff like "3 for the price of 1", "5 + 1 free", "35% discount!" from a particular site and then converts the price to a % discount and the price per item / 250 grams. Its an old userscript so it is glitchy and halfway works. I already pre-chewed the work by telling Gemini 3 exactly which new HTML elements it needs to match and which contents it needs to…

Honestly, in my biased unscientific testing, what gives is that Gemini isn't actually all that good. I mean, it's fine. but it's not actually good.

Re: Gemini 3

#990

Post by mlrtime »

dotancohen wrote: post #957
mlrtime wrote: post #890

Earlier quoted context omitted.

>documented problem in Home Assistant this evening. All it would take is 3-5 lines of YAML. The model failed miserably. I think we're all still safe. This is mostly because HA changes so frequently and the documentation is sparse. To get around this and increase my correction rate, I give it access to the source code of the same version I'm running. Then instructions in CLAUDE.md on where to find source and it must u…

Feel like sharing that prompt? I have a feeling that the phrasing on the "must use source code" part needs to be just right.

It's not really magic, in my project folder I will git clone the source code of whatever I'm working on. I will put something in the the local md file like:

Use ./home-assistant/core for the source code of home assistant, its the same version that I'm running. Always search and reference the source when debugging a problem.

I also have it frequently do deep dives into source code on a particular problem and write a detailed md file so it only needs to do that once.

"Deep dive into this code, find everything you can find about automations and then write a detailed analysis doc with working examples and source code, use the source code."

Post reply on HN