speerer

It prompts "gigantic alien tentacles that destroy the world."

I suspect you're reading the base prompt, but the video also demonstrates adding further prompts to change what is happening as the scene progresses.

The part where the LLMs (actually, VLMs) outperformed the CNN was that the LLMs were better at avoiding false negatives (see the quote below). That kind of Type 2 error is the most dangerous one since it persuades people not to get advice. Since the LLMs are designed to be general-purpose and public-facing, the headline could be flipped to: "A specialised 10-year old CNN is a worse public facing tool than a general frontier LLM". Which would surprise nobody.

                                 | GPT |Claude | CNN |  
    Unsafe lesions kept (of 100) | 97 | 96     | 77 at its preset [sic] threshold; 98 or 99 when matched |

A rare occasion where the AI response was an improvement on the human-written ones.

Sri Lanka?

South Korea?

I agree with you, but the article doesn't. FWIW, I've been in the specific scenario (explaining things neutrally to politicians) several times.

If I had an opinion on an issue but no dog in the fight; and if I were to explain my opinion to politicians in the hope they might learn from or even share it -- I may well be "lobbying", but I don't believe I'd be corrupt.

Yes! https://discourse.ubuntu.com/t/gleam-and-the-value-of-small/...

...And he animates his presentations by hand-drawing each frame transition, which speaks volumes.

[edit - this might have been ambiguous. I meant that it shows he really cares about the fine details.]

I think the last sentence of your comment is the most exciting thing about this whole article!

There are at least three enormous gulfs here:

First, just understanding music and even jazz as a prerequisite to being able to do this.

Second, being able to do this using machines instead of a piano itself, which is unaffordable for a lot of people.

Third, the capturing of the je ne sais quoi of musical style, which is almost impossible to articulate.

So for me bringing this all together, it allows you to capture an improved understanding of different styles without needing to invest a huge amount of time because you can feed the same input into multiple style models and see how it diverges. That's a control which can't usually be obtained!

As a technical feat, what AI research allows across domains is a remarkable thing.

What a disgusting and unworthy comment.

The cut-and-stick nets aren't but the rest most certainly are! Look up "The Complete Book of Origami Polyhedra". I've made some of them myself: they use only square paper without tearing, cutting or sticking.

To be fair, the grandparent comment being replied to has a username beginning with Mr.

> ... particularly concerned about is social media. Most images and videos uploaded to most social media is re-compressed by the target platform. This is a door to tracking that is far too easy for social media platforms to open.

This has been going on for a while with Facebook. They seem to embed custom metadata tags so that images shared outside the platform can be traced back:

https://stackoverflow.com/questions/31120222/iptc-metadata-a...

I'm not sure why one would even strike metal against a crucible! It's a container for liquid metal. One of the outputs shows it being smashed by the manoeuvre, which is probably the most realistic outcome of all of them.

Sorry, I'm not trying to nitpick. I'm just joining in because I'm interested in how the models dealt with the request.

> Why do they insist on every new phone being bigger than the phones before them?

Because every time, they want it to be the next big thing.

The original source is here: https://www.nature.com/articles/s41467-026-76614-0

> Thylacine crania are far larger than expected for their body mass, as determined by their centroid size (Fig. 3a), and exceed the absolute cranial sizes of far larger marsupials such as red kangaroos (Supplementary Fig. 5). While the thylacine weighs ~17 kgs34, its cranial size (within ±10%) is similar to that of much larger species, between 24.5 and 66.7 kg (Fig. 3a). In contrast, most placental carnivorans that are close to the thylacine in overall cranial shape (Figs. 1b and 2a) are jackals and foxes weighing 6–14 kgs (Fig. 3b). The cranial size of the thylacine is on par with much heavier predators such as the grey wolf, African wild dog (Lycaon pictus), puma (Puma concolor), and leopard (Panthera pardus).

...

> The large size of the thylacine cranium has not been integrated into interpretations of its function, possibly because morphometric analyses generally remove absolute size43,44 and biomechanical comparisons usually focus on comparative, rather than absolute, generation of bite force44. However, a large cranium counteracts the weaker structure of a long and narrow rostrum by increasing the absolute magnitude of rostral resilience to stress and bite force. This is because larger crania are more resilient to intrinsic and extrinsic loads3,44 and carry more muscle3 compared to smaller crania that share the same shape. It is also possible that the wide zygomatic arch compared to other canids, particularly in larger thylacines (Fig. 4e), plays a role in buttressing the jaw joint against the leverage of the long jaw3,45, but this remains to be tested. Overall, the large size of the thylacine cranium thus indicates a specific mechanism of prey capture and/or immobilisation optimised for fast, high-impact bites whose force was dissipated across a large volume of cranial bone.

Far worse from my perspective is that it's not a fixed citation that it's not falsifiable. It's just a mutable, generalized source, which is stochastic in nature.

It may as well say "(Socrates, probably)"

Just an aside on the design rather than the content:

I was wondering for a while about the greenish and yellow bars on the margins of the page, but I'm pretty sure that those are padding to make the document work well on both US letter paper and A4 at the same time! The text area is A4 proportions, so if it were printed, zoomed to size on that paper, the sidebars would simply disappear.

No doubt this is a common technique that I'm just noticing for the first time.

...but yes, some comments are performative.

It does in the sense that ineffable things are unable to be described in language (conceptually: unknown to language).

I think normally it's just that someone has the thought and shares it for discussion. Not every comment is performative.

This is so similar to human decision-making though. First I my innate experience to approximate to what I expect is right, then I map that to the truth.

It's the same for so many things:

- reading documentation (what do I expect this function to be called?)

- finding clothes in a shop (something long-sleeved and light)

- picking the fridge for dinner

- finding a book in the library...

so many analogues where I'm not coming cold to a choice.

Genuine question - are you copying the Claude phraseology for effect (in which case you captured it brilliantly), or is there a more mundane explanation?

Exciting! I needed this last week, and now here it is. Thanks.

I share the annoyance about phone detection previees over-promising. But I'm pleased this comes with perspective correction and lighting correction together.

> FAQ

> Will you store my images?

> We are running this service on a small server. We literally cannot afford to store any images. If that is what you need, there are more suitable services out there.

This does not answer the question - it doesn't say "No." or that images are permanently deleted from server on burn expiry, only that there are (currently) limits. So slightly worrying that there might be selective retention.

Is this correct? I was fairly certain that the correlation between random data points approached 0.

Is it also plausible that the relationship between true and percieved is wholly independent (as this presumes)?

There's some interesting commentry at https://www.techtimes.com/articles/322754/20260803/norway-id... , which suggests that the widespread outage is due to a single point of failure:

> ID-Porten: When One Gateway Controls Everything

> At the center of the disruption is ID-porten — the national login gateway operated by Norway's Digitaliseringsdirektoratet (Digdir), the government agency responsible for public-sector digitalization. ID-porten functions as the single sign-on portal through which Norwegian citizens authenticate themselves to access public digital services.

It seems to be a corporate service provider that's a dependency for many other services.

Hi, I've been working on handwritten blogging too! It would be great to compare notes. I might also drop you a line on another platform.

I think normally when people say Chinese made in this way, what they're really communicating is that there's no (meaningful) brand. All they know about it is that it is from China.

Having subscribeed for years, I find it very good on that front. I haven't detected any political bias, and I suspect they would say that fairness and honesty is what the "journalism" half of "slow journalism" means.