jshmrsn

It’s not the LLMs themselves, it’s the training data that they are trained on being supplemented by synthetic video from simulations / 3D rendering.

Don't know if you'll ever read this, but for the record, I honestly misread your original comment and wrote my snarky comment (which was too snarky even if you wrote what I thought you wrote) without double checking. Sorry.

Also for what it's worth your point turned out to be very true with yesterday's Image 2.6 release, where I was all happy to switch to it because the pricing was the same, only to later realize that prices would increase ~3x because my use case was using the "low" setting before, and the 2.6's low is ~3x the price of 2.5's low.

2019: We have no idea if computers will ever able to talk like a human in natural language. 2026: We have multiple magical know-everything oracles that can speak any language better than most humans and produce pages of output for fractions of a cent, but the model names are a bit confusing so let's just not bother.

In case this is helpful to you:

There's a back button, which if you just went into a child folder from a parent folder, then pressing the back button will go to the parent folder.

If you explicitly want to go to the parent folder, clicking on the name of the current folder in the dropdown in the top center of the open dialog, will reveal a list of folders that is the hierarchy from current folder on top, to root folder on bottom.

You can also press Command + Up Arrow to go to the parent folder.

To edit the path as text, press Shift + Command + G, which will show a dialog with the current path, and allow you to edit it as text with autocomplete support.

Bonus tip: While in the file open dialog, if for some reason you want open the current directory in a full normal Finder window, press Command + R.

His last blog post was well written, to the point, and sobering. The friends website measuring his lines of code and pull requests is making me feel sick to my stomach. I feel reminded to reflect on how I would measure my own life.

Precisely. That has always been prohibited both by technical measures and by the App Store review guidelines. Especially on Apple, but also on Google Play / Android as well.

Well, I am trying to inform myself because I did understand the cookie popups to be connected to GDPR. As far as I can tell from my reading, your assertion that "cookie popups have nothing to do with GDPR" is not true.

From my reading, it seems that while cookie permissions first became an explicit EU law concept in a 2009 amendment to ePrivacy Directive (not GDPR), companies were able to get away with passive consent banners (not popups).

It was GDPR's new definition of consent which then retroactively strengthened the existing ePrivacy Directive cookie consent to explicitly require user action to give consent (i.e. popups, banners large enough to push users to interact with them, etc.).

Unless your point is that GDPR has nothing to do with popups because the companies could just not use non-strictly-necessary cookies and therefore not need a popup, but I think that's a stretch to jump from there to "nothing to do with GDPR".

https://gdpr.eu/cookies/

https://wp-gdpr.eu/gdpr-cookie-consent-2026/

https://eulawanalysis.blogspot.com/2022/01/consent-and-cooki...

Well cheated would definitely imply fraud. “Faking it” as in “fake it till you make it” is more like pretending you know about a topic until you learn enough on the job to participate competently.

Roll call from the Assembly vote: https://legiscan.com/CA/rollcall/AB2047/id/1702219

Mad Max 1910

Am I missing something, or do this article’s purported vulnerabilities rely on an assumption that an attacker already has enough access to your system that the attacker can modify files which your code is referencing by path? Isn’t this more of an escalation vector than a vulnerability in itself?

I’m trying to understand the practical takeaway.

I am working on a modal code editor project that you might find interesting then. It also operates on an AST directly, which is represented as UI nodes which closely resemble normal text layout. Email in profile if you’d like to give it a try and possibly give early feedback (still in early development).

Considering that a fleur-de-lis involves somewhat intricate curves, I think I'd be pretty happy with myself if I could get that task done in an hour.

Given a harness that allows the model to validate the result of its program visually, and given the models are capable of using this harness to self correct (which isn't yet consistently true), then you're in a situation where in that hour you are free to do some other work.

A dishwasher might take 3 hours to do for what a human could do in 30 minutes, but they're still very useful because the machine's labor is cheaper than human labor.

If you haven’t already, try going to Personalization settings, change tone to “Efficient”, and set Warm, Enthusiastic, and Emoji to “Less”. While not fundamentally solving the issue, I do prefer it over the baseline behavior, to the extent that I miss having a similar setting in Gemini.

I enjoyed reading it. Whether one believes the future will look like this fictional/hypothetical one, it encourages the reader to think about what would need to become true for this future to be plausible.

That has not been established in the courts, at least not precisely enough to assert that for sure this project isn’t copyrightable.

“ But the decision does raise the question of how much human input is necessary to qualify the user of an AI system as the “author” of a generated work. While that question was not before the court, the court’s dicta suggests that some amount of human input into a generative AI tool could render the relevant human an author of the resulting output.”

“Thaler did not address how much human authorship is necessary to make a work generated using AI tools copyrightable. The impact of this unaddressed issue is worth underscoring.”

https://www.mofo.com/resources/insights/230829-district-cour...

I don’t mean to suggest it alone would tip the scales. And I agree the hope for international intervention is dimmer than it ever has been. But it would be one thing on the scales, as it has been in Ukraine as well. While there has not been direct military intervention in Ukraine, the support that has been provided relies on political popularity, and Russia’s endangering of Zaporizhzhia has contributed to the disdain of and attention towards Russia’s invention.

I wish Taiwan’s reactors were never shut down in the first place, and I hope Taiwan can hold out long enough to get it started back up again. It’s a step towards being able to withstand a blockade (Taiwan lacks oil, gas, and goal resources, so it relies on imports). If PRC chose to attack a nuclear power plant, it might give the necessary pressure for international intervention.

For what it’s worth, I’ve personally walked around the nuclear containment area on Orchid island and swam in the waters around it. It’s a well managed and nice place.

Scale AI is a provider of human data labeling services https://scale.com/rlhf

This is true again for the most advanced fighter aircraft, except the active hand is now a computer.

If the machine can decide how to train itself (adjust weights) when faced with a type of problem it hasn’t seen before, then I don’t think that would go against the spirit of general intelligence. I think that’s basically what humans do when they decide to get better at something, they figure out how to practice that task until they get better at it.

I am familiar with ‘it’ as a default closure input from Kotlin. From a quick search, that in turns seems to be inspired by Groovy.

Some of the text that the LLM is trained on is fictional, some of the text that its trained on is factual. Telling it to not make things up can tell it to generate text that’s more like the factual text. Not saying it does work, but this is a reason how it might work.

Did that model also factor in risk of damage, liability, and normal wear and tear that a tenant brings over a vacant unit?

If the “sound” is an internal perception, then noise cancelling headphones would not help at all. They might make it worse by quieting any background sounds that could otherwise help cover up the internally produced sensations.

An idea I hear often listening to talks about LLMs, is that training on a larger (assuming constant quality) and more various data leads to the emergence of grater generalization and reasoning (if I may use this word) across task categories. While the general quality of a model has a somewhat predictable correlation with the amount of training, the amount of training where specific generalization and reasoning capabilities emerge is much less predictable.

I can only speak from my own internal experience, but don’t your unspoken thoughts take form and exist as language in your mind? If you imagine taking the increasingly common pattern to “think through the problem before giving your answer”, but hiding the pre-answer text from the user, then it seems like that would pretty analogous to how humans think before communicating.

There’s a website designed for language learning from watching YouTube captions with inline translations and dictionary lookup. It also has support for searching videos by subtitle content. But it has a limited index and isn’t free for all features. I thought its source was available but I can’t find it now… https://languageplayer.io/

I get the point that the US already put people on the moon… but how can you possibly make the leap that there can be no scientific value to additional unmanned laboratories and instruments landing on the moon? Especially since this represents increasing the number of countries who can contribute to this scientific endeavor? If the US elects a president who is not interested in lunar science or has economic problems, then the whole world must wait for the US to decide to resume lunar missions?

An overview of the scientific instruments onboard:

“ Lander payloads: Chandra’s Surface Thermophysical Experiment (ChaSTE) to measure the thermal conductivity and temperature; Instrument for Lunar Seismic Activity (ILSA) for measuring the seismicity around the landing site; Langmuir Probe (LP) to estimate the plasma density and its variations. A passive Laser Retroreflector Array from NASA is accommodated for lunar laser ranging studies.

Rover payloads: Alpha Particle X-ray Spectrometer (APXS) and Laser Induced Breakdown Spectroscope (LIBS) for deriving the elemental composition in the vicinity of landing site.

Chandrayaan-3 consists of an indigenous Lander module (LM), Propulsion module (PM) and a Rover with an objective of developing and demonstrating new technologies required for Inter planetary missions. The Lander will have the capability to soft land at a specified lunar site and deploy the Rover which will carry out in-situ chemical analysis of the lunar surface during the course of its mobility. The Lander and the Rover have scientific payloads to carry out experiments on the lunar surface. The main function of PM is to carry the LM from launch vehicle injection till final lunar 100 km circular polar orbit and separate the LM from PM. Apart from this, the Propulsion Module also has one scientific payload as a value addition which will be operated post separation of Lander Module.”

https://www.isro.gov.in/Chandrayaan3_Details.html

Wikipedia defines panspermia as living organisms being spread throughout the universe. An unreviewed paper about carbonic acid in outer space does not directly have a connection to panspermia. The topic of panspermia is being pushed by the article.