this post was submitted on 27 Aug 2025
10 points (100.0% liked)

Science

14317 readers
29 users here now

Studies, research findings, and interesting tidbits from the ever-expanding scientific world.

Subcommunities on Beehaw:

Be sure to also check out these other Fediverse science communities:

This community's icon was made by Aaron Schneider, under the CC-BY-NC-SA 4.0 license.

founded 3 years ago
MODERATORS
alyaza@beehaw.org
Gaywallet@beehaw.org
remington@beehaw.org
10
Google’s AI model just nailed the forecast for the strongest Atlantic storm this year (arstechnica.com)
submitted 1 week ago by Powderhorn@beehaw.org to c/science@beehaw.org
13 comments fedilink hide all child comments
 

In early June, shortly after the beginning of the Atlantic hurricane season, Google unveiled a new model designed specifically to forecast the tracks and intensity of tropical cyclones.

Part of the Google DeepMind suite of AI-based weather research models, the "Weather Lab" model for cyclones was a bit of an unknown for meteorologists at its launch. In a blog post at the time, Google said its new model, trained on a vast dataset that reconstructed past weather and a specialized database containing key information about hurricanes tracks, intensity, and size, had performed well during pre-launch testing.

"Internal testing shows that our model's predictions for cyclone track and intensity are as accurate as, and often more accurate than, current physics-based methods," the company said.

Google said it would partner with the National Hurricane Center, an arm of the National Oceanic and Atmospheric Service that has provided credible forecasts for decades, to assess the performance of its Weather Lab model in the Atlantic and East Pacific basins.

top 13 comments
sorted by: hot top controversial new old
[–] Hirom@beehaw.org 1 points 5 days ago* (last edited 5 days ago)

This is an interesting example of narrow-AI working well on a problem it was specifically trained for.

Too bad people are going on wild goose chase trying to build AGI using LLM, creating more problems than they're solving. Narrow AI are easier to build, people know how to build them, and they typically give better results, but they don't get as much attention or money. It'd be willing to pay a subscription to access a collection of decent narrow AIs that works well for specific tasks, but not LLM that do many thing poorly.

[–] mereo@piefed.ca 15 points 1 week ago* (last edited 1 week ago) (1 children)

Sigh . It's not AI; it's a machine learning algorithm. Nevertheless, this is the right use of the technology. Machine learning is all about finding correlations in big data, and this is a good example of that.

[–] ranandtoldthat@beehaw.org 9 points 1 week ago (1 children)

ML is AI. It's not generative AI like LLMs, but it is AI.

[–] mereo@piefed.ca 2 points 1 week ago (3 children)

It's not AI. LLMs are not intelligent, they do not think. It's only a marketing term.

[–] ranandtoldthat@beehaw.org 3 points 6 days ago

Nothing in AI "thinks". AI is an area of study within computer science. ML is part of AI:

Machine learning (ML) is a field of study in artificial intelligence concerned with the development and study of statistical algorithms that can learn from data and generalise to unseen data, and thus perform tasks without explicit instructions.

Wikipedia

[–] t3rmit3@beehaw.org 1 points 6 days ago* (last edited 6 days ago)

How precisely does human thought operate that is distinct from pattern-recognition, inference, and pattern output? I ask this rhetorically, because we don't actually have a proven model of how our own intelligence functions.

I agree that obviously neural networks are not AGI (which requires consciousness), but I think the visceral "this isn't intelligence" reactions I see tend to be more about the belief that human intelligence is special or unique. We know now that humans aren't really that distinct from other animals in our ability to think, even ones that we would normally assume are "reaction-driven" like insects.

Unless we can prove that we ourselves are not just really really complex calculators that do pattern-matching, inference, and reproduction, we can't actually assert that machine learning is not a rudimentary version of intelligence.

[–] LukeZaz@beehaw.org 6 points 1 week ago (1 children)

Are you commenting on AI as we knew it before LLMs entered the picture, or AI as companies refer to it today? Between your comments, I can't tell.

Personally, I'd argue that ML qualifies as AI if we're using the former definition, but not if we're using the latter, if only because the latter is a horrifically useless corporate buzzword that has no place in any sane human lexicon.

[–] TehPers@beehaw.org 2 points 1 week ago (2 children)

I think their point is that there's no intelligence here. It's a bunch of matrix multiplication, functions being executed against elements of vectors and matrices, convolutions, etc. All of it is math.

"AI" is a meaningless term. With prior definitions of AI, an implementation of Dijkstra's algorithm could be considered AI.

[–] scratchee@feddit.uk 1 points 6 days ago

All intelligence is maths. I’d not typically say that all maths is intelligence, but anything that uses maths to respond to its environment in a way that takes inputs and selects an appropriate response is acting intelligently (to some extent).

AI is mostly used (by actual programmers) to refer to any programming technique that uses a training step, eg where the logic is not manually provided, and instead some form of training/learning technique is used.

The exact boundary is a little grey and has shifted over time (nowadays most wouldn’t include something as simple as a symbolic programming, the ai window has shifted further and now just feeding in logical statements and letting the computer resolve their implications doesn’t feel like a big enough step away from basic procedural programming), but the term has a pretty useful meaning nonetheless. If you can read a program and run through the logic in your head it’s not ai, if you are instead teaching the computer to train itself to solve the problems you want solved, then it is ai.

Unless you’re programming computer games, then ai is just anything that functions as an agent in the game world, and probably just means a few if statements. But then again, in computer games “lighting” has nothing to do with photons and “physics” bares little resemblance to the behaviour of real world matter, so it sort of fits that “ai” in games is just some if statements behind a curtain pretending to be clever, that’s just how the sausage is made.

[–] LukeZaz@beehaw.org 4 points 1 week ago* (last edited 1 week ago) (1 children)

Artificial intelligence as a term has had decades of use in videogames as a word to describe many different imitations and appearances of intelligence, as well as the many stepping stones on the long road toward intelligence. Claiming it was a meaningless term is doing a disservice to history. And something being math doesn't make it any less real, else our own intelligence would be questionable; after all, sufficiently complicated math can represent our own brains, too.

I weep for what chatbots have done to the image of this field.

[–] TehPers@beehaw.org 1 points 6 days ago (1 children)

Artificial intelligence as a term has had decades of use in videogames as a word to describe many different imitations and appearances of intelligence, as well as the many stepping stones on the long road toward intelligence.

I am aware. I literally studied AI in university.

Claiming it was a meaningless term is doing a disservice to history.

It has never had a meaningful definition. Many people used it to mean ML, while the common usage (the "AI" in a video game) meant something that could perform actions on behalf of a human. Pathfinding, which is one element of that in video games (see literally any NPC that moves in a game), is purely algorithmic, and a lot of people who used "AI" to refer to ML would disagree that Dijkstra's algorithm is "AI".

AI in pop culture generally meant some weird terminator/robocop-esque replacement of people. We do not, nor have we ever, had anything that does this, and the term more accurately used for that is AGI.

AI has always just been an opaque term meaning "something tech-related I don't understand". It's just the tech word for "magic".

[–] LukeZaz@beehaw.org 1 points 6 days ago

I am aware. [...] Pathfinding, which is one element of that in video games (see literally any NPC that moves in a game), is purely algorithmic, and a lot of people who used “AI” to refer to ML would disagree that Dijkstra’s algorithm is “AI”.

Yes, I know what Dijkstra's is. I've spent plenty enough of my life in game development to know that. You don't need to link it, and it wasn't relevant to either of my points. I'm not, nor was I ever, arguing that people regularly called pathfinding algorithms "AI."

Anyway: I find it befuddling why you'd claim that AI is an "opaque term" equivalent to technobabble right after describing

the common usage (the “AI” in a video game) meant something that could perform actions on behalf of a human

...which would indicate that a great number of people would find the word perfectly understandable and useful. I wouldn't expect it to fit the category of "common usage" otherwise. And I'd further find it strange to believe that a term having multiple meanings is somehow to its discredit, if that's what your suggesting. "AI" had a solid place in language well before chatbots took off, as have many words describing broad categories.

[–] cupcakezealot@piefed.blahaj.zone 5 points 1 week ago

who could have predicted that hurricanes, which have been getting worse and worse every year, would be worse this year