Spanish cities by seaside have a pattern of putting the sea on the "south", even when it's not there.

The next subtle Skynet will thank you for putting this in its training set.

> but LLMs have a place in vast number of places where we'd previously never use a classifier because it was too much effort.

Absolutely! But that doesn't rule out that there are AIbros that will tell you to use an LLM, even if the task could be done with a static analyzer with a few rules. Use an LLM because it's AI!

You seem to be quite fixated on a personality type rather than the underlying technology.

Not really, although when a certain personality type dominates a scene, people tend to conflate the two. Especially when they are louder than the rest.

Now, about the wifi part, why isn't there an Home Assistant extension that detects already room presence magically exploiting all of this??

There is: https://community.home-assistant.io/t/espectre-wi-fi-motion-...

There is with Zigbee: https://github.com/orgs/home-assistant/discussions/960

But this relies on Hue. I would like something at HA level directly, aka a zero-effort solution if you already run a zigbee swarm either directly controlled by HA or via zigbee2mqtt

It can be useful if you lost the keys for $REASON while hiking or doing any activity and you can phone somebody else who has access to the API and ask them to unlock and remote start the car. It doesn't happen that often, but it happens, and in the big scheme of things it's more likely to happen [1] that somebody stealing it via a vulnerability in the car maker web services.

Or at least as a car owner, this is my risk assessment exercise output.

[1] Yeah, I just pulled this off my ass, don't think there are trustworthy stats either in one direction or the other.

Don't remote-started cars still require the presence of the key or fob to shift the car into drive?

My 2012 Chrysler certainly does.

I think it depends on the make and model. Teslas don't, for example.

You don't need access to move a vehicle, you can lift it up and move it. Like when you park in a place you shouldn't.

In my experience, this kind of developer was always the one taking shortcuts, generating bad code and tech debt and in general just thinking about the ticket they were working on and nothing else.

Mmmh no? 90kWh battery packs are not that common. An M3 or MY LR AWD have an 82kWh. Some Mercedes and some Chinese model have a 100+kWh battery but the most common setup is between 60 and 70.

Anyway I agree that in emergency conditions it can go a looong way for a house.

My bmw has 108

> Now if you want to read about e.g. Ukraine/Gaza/Iran/etc, you can read first person accounts from people on the ground, get imagery of things seconds after they happen, and so on.

That's the good side of the Internet (and smartphones) and what all of us early adopters always said the Internet will be great for. But then there is the other side of the coin: fake news, fake account, coordinated disinformation campaigns from foreign state actors, bad taste pranks etc. Some of those existed before with classic journalism as well but the scale at which things are now done was simply impossible before.

So, having journalism and journalists as a trusted source, and filter, that can already digest and filter out the fake from the truth are still needed. Or even more than before, I dare to say. The issue is that there are too many monetary incentives to not act like that.

How do you expect journalists to be able to figure out the truth? And how do you not see them as a single point of failure? Let's take an example from the past. The Vietnam War was started because of a false flag. [1] This was a time when there was a much more adversarial relationship between the media and the government/intelligence agencies than there is today. But even in that context, they mostly unquestioningly reported the lie. Because how else could they have known?

In modern times I'll again take the glass half full side here. The truth of things tends to come out over time, because it's the only 'narrative' that can hold up under scrutiny from any angle. And even in the anonymous internet age, there's still the issue of reputation. It's perhaps more important than ever now a days because you can look at some entity's predictions or statements from one time period, see how it played out over the next e.g. 6 months, and see whether they were onto something, or blowing smoke.

I think this is why it's so hard for the government to drum up support for war now a days. They have no ability to control the narrative and so framing invading some country half way around the world as being 'defensive' is a narrative that usually collapses in rapid order. Even during Vietnam it started becoming clear that the narrative around the Gulf of Tonkin was dodgy, but it's a process that took years. By Iraq we were down to lies starting to be exposed within months. Now a days it'd probably take about 24 hours, to the point that with Iran didn't even bother with trying to manufacture consent and we saw the result of that a few days ago.

[1] - https://en.wikipedia.org/wiki/Gulf_of_Tonkin_incident

> How do you expect journalists to be able to figure out the truth?

And how are YOU expected to figure out the truth? Journalists should be akin to scientists and researchers: specialized in their field, with better connections and expertise and be able to follow a trail to reach ground truth. Now, can and will be they fooled? Absolutely, and there should be still some peer-review of what they publish.

Also this is an ideal scenario, and it's not what happening right now, or at least it happen only for a handful of media outlets all over the world. This model still support (and expect) having biased publications - I mean, it's just how we humans are - and you as a responsible citizen are still required to look at least at a couple of sources and form your own opinion.

I do agree with you that the truth tends to come out at some point because it follows the path of least resistance, but sometimes the lie can be so well written that it takes decades for the truth to come out.

Have you studied a master degree in every field you read casual information about? For hard science matters, sure, they should, or at very least be very connected to the field to have a very good understanding of it, critical thinking to smell out the bullshit revolutionary researches and a network of actual PhDs working on the field every day that can help them getting to the core of the topic. Apply the same concept to all the other fields, even the not hard science ones, and you may end up with a web of news and information outlets that you can believe.

Again, this is an ideal scenario, I know, but IMO there is nothing that makes it technically impossible to achieve.

Journalism is, in fact, a university degree program. My journalist friend does indeed have an MFA in Journalism. This is not an unrealistic expectation if regular pay is on the other end.

It's a superficially reasonable objection that elides the crucial detail: Modern journalism is a profession. It has standards of entry, an established mission (fact-finding, truth-seeking), and a code of ethics. We can choose to be cynical about those constraints, but the fact is that none of them apply to a random person on Youtube or Substack.

This. Say what you want about the tenets of journalism, but at least it's an ethos.

The 'citizen journalist' is at best a part time volunteer and at worst an astroturf'd contractor of a PR department.

And as the quip goes about volunteers -- willing to work for free, altruistic, and competent: pick two.

Occasionally you get delightful humans that combine all three, but they're needles in haystacks.

Like they used to?

Journalism had a pretty strong code of conduct and drive to publish the facts. They couldn’t get it right all the time, but they had a stake in doing the best they could and not losing their reputation.

The truth of things doesn’t tend to come out in time. We know the truth about many things now, but after the window in which a change could be made was obfuscated and stopped.

Facts coming out when it is convenient to the people withholding them is not a success. If they come out at all.

The old system of journalism did a surprisingly good job, until people gave up on playing by the rules and decided to play by the trappings of journalism.

Reputation matters to the center and the left of the information ecosystem today. The right has connected its media to its politics entirely, to create a propaganda machine. Part of its routine is to state that journalism and the left are biased, and don’t want to deal with the truth, and then shoehorning their narrative into it.

> The truth of things tends to come out over time, because it's the only 'narrative' that can hold up under scrutiny from any angle.

This is exactly why professional journalists, with employers who can restrict strong pressure, are important.

The truth about things which are widely observable tends to come out over time.

There are many, many things which are not widely observable.

You know the easiest way to do something secretly in government today?

Have an in-person meeting, gloss over it in online-posted minutes, then bury the evidence of it being done in the most detailed, voluminous reports.

Because finding that requires a butt in a seat through boring meetings and tracing / reconciling line items through reports.

And generally speaking, most people who aren't being paid aren't willing to do that for fun.

Journalists figure out the truth better than you - generally speaking - by having more time to do it than you do because that's their day job while yours is coding or whatever you do.

I bet they do that as well but I also think that if they have to choose between inference speed or "intelligence" they would choose the latter.

They don't have to choose, though, since they can release multiple models with different tradeoffs.

This would imply they have infinite development capacity

Not infinite, just enough to make more than one model. OpenAI and Anthropic both have four different size categories...