GranularRecipe

What I find interesting is the implicit priorisation: explainability, (human) accountability, lawfulness, fairness, safety, sustainability, data privacy and non-military use.

Is this a new policy? Otherwise, why this sudden and broad implementation so that "suddenly none of the IT employees at the agency could do their jobs properly anymore" (according to the source).

Removing admin from people who don't need it is 100% the correct thing to do according to any IT guidelines you could quote. And of course, every single user will whine that they're special and really really need it.

You assume that "suddenly none of the IT employees at the agency could do their jobs properly anymore" is whining and not substantial?

Shouldn't be least privilege principle a culture (a standardised and automated process) and not something that happens ad hoc?

The first month after March equinox has great religious significance across many cultures (in the temperate zone), e.g. Pesach. When different cultures meet, they tend to syncretise.

That's the trade-off between efficiency and resilience. Another example are the egg prices in US. The industrialisation of agriculture made it more brittle.

Where did you register yourself as an expert? Asking for a friend.

In covalent bonds, it's two electrons one pair. Metallic bonds are more polyamorous.

This is not available in many countries. Is it due to legal issues, to preserve server capacity or a marketing gag?

I don't know how Google is internally organised, but maybe Google Labs does not have a dedicated legal team. And out of fear for possible legal implications, they just block it in EU. It's quite understandable, because the relevant legislations (AI Act etc) are all very new, not been tested in courts and due to the lack of stare decisis, you can't rely on a uniform application of the law in all jurisdictions, so they just block it. Frustrating, but I prefer they move fast rather than safe. And from their perspective, it means not releasing it in the EU.

For those who want to check the complete conversation: https://gemini.google.com/share/6d141b742a13

It's not an inshallah, it's a masha'Allah.

Math notation consists of symbols and syntax (order of symbols). How to arrange the Chinese characters (syntax) is very different between different languages.

I would argue that alphabetisation would increase the use of regional languages because it would be possible to write down your speech, instead of relying on Mandarin or phonetical transcription with Chinese characters.

It's not a language. It's a writing system used to write Classical Chinese for 3000 years and Mandarin for 150 years. When I was in Hong Kong, I was surprised that the written language is Mandarin (in traditional form).

I don't think these two forms of'amnesia-induced' typos are so different. "Typos" happen quite often, especially when typing on a smartphone and selecting the wrong character. And people learn to read it correctly. It's sometimes used intentionally, e.g. 歪果仁.

I think it's also an issue of translation. "Dialect" in English is variety of a language. 方言 (fangyang, translated as dialect) in Chinese is a regional language or speech. Linguistically, Cantonese and other varieties of Chinese are recognised to be part of the Chinese language group / family.

Cf https://baike.baidu.com/item/%E6%B1%89%E8%AF%AD/22488993?fro...

The chart is dumb (the axes badly labelled). It's a power-law graph (used to explain the Pareto principle). I think the message is that a large percentage of interventions are not cost-effective, and only a small number are.

Why do you think it's a fallacy? Do you think that he would have returned to China regardless of his treatment in the US, or do you think that his contribution is insignificant to China's missile programme?

Or put it differently, how would it (counterfactually) not be a fallacy?

A few comments mention that the policy is about internalising the cost of externalities.

In this concrete case, I wonder how do you price it correctly absent a market?

Generally, I wonder how far you can extend the concept of internalising externalities without severely limiting individual liberties.

E.g. health care (in countries with a public health care system). We already try to price in the social costs (burden on public health insurance) of smoking and drinking with taxes and other limitations. How about other activities, like biking without helmets (legal in some countries), extreme sports or consumption of coffee, meat etc? Should we tax and limit activities that are "negative externalities" and provide tax benefits for "positive externalities"?

I agree with his sentiment, but his wording is rather offensive to barbarians. Profiteering is enabled by the fine civilisational invention called "intellectual property".

It really unnerves me, that we call this AI errors "hallucinations". Hallucination is an error in perception (false positive stimuli). LLM cannot hallucinate because they have no sensory organs (other than training data).

I think the closest psychological conditions are confabulation (memory error) and delusion (false beliefs).

Let me expound my arguments more verbosely:

The author set himself an ambitious goal of webbing an intricate story about Western physics from Aristotle to Lagrange. Good story-telling starts with something familiar, seemingly truthful to the reader, that is yet (partially) wrong or incomplete (unknown to the reader). And through the story, the reader gains wisdom by being surprised of something new that (partially) falsifies their current body of knowledge.

The author starts with something that the modern-day reader knows to be wrong, and ends with a dubious conclusion (that somehow the variational principle validates Aristotelian teleology).

A good story would start with the Newtonian mechanics, the prevailing view among those who did not study physics at the university. The author does not need to heed my amateur advice: He can attack strawmen, i.e. valid yet superfluous, off-target criticisms.

Yes, I was referring to the opening of the article which ridicules (trolls) Aristotelian physics. No, I did not only read the beginning. I also skimmed over the rest.

Arguing against Aristotelian physics in 20th / 21st century is like pounding a rotten strawman. Does anyone know a modern (influential) Aristotelian philosopher who holds views the writer is criticising?

Valid criticism. I removed the snipped "what eIDAS is about" from my comment. I think the criticism raised is not (solely) against digital identity, but about possible security risks associated with it.

For those wondering why some cybersecurity experts are concerned, from Wikipedia (https://en.wikipedia.org/wiki/EIDAS#Man-in-the-middle_attack...):

In 2023, a change was proposed to eIDAS that would allow any EU government to surveil all internet communication, even when encrypted.[..]

The proposal would force internet companies to place a backdoor in web browsers to let them perform a man-in-the-middle attack, deceiving users into thinking that they were communicating with a server they requested, when, in fact, they would be communicating directly with the EU government. The EU government would then read and change their messages before passing the possibly modified message on to the intended recipient.

This is not only an issue for shallow networks. As far I know, both points apply to all feed-forward networks regardless of depth.

What the ratio for the number of parameters required to learn some complex function between a shallow network and a deep network (preferably as a function of the complexity)?

The reason for deep learning is that shallow networks are very hard (or impossible) to train. In that sense, long time of training is evidence for shallow networks.

The right of the general public to read the unpublished novels might override the wish of a dead person. Maybe.