Changes the torque and the application of said torque for each bolt. As in “tool head has 5° of give until in place, then in ramps torque to 5nM over half a second, and holds for 1 second and then ramps to zero over .1 seconds”, and then something different for the next bolt. Then it logs that it did this for each bolt.
The tool can also be used to measure and correct the bolts as part of an inspection phase, and log the results of that inspection.
Finally, it tracks usage of the tool and can log that it needs maintenance or isn’t working correctly even if it’s just a subtle failure.

Yes, but that’s the case regardless. My message going through has always depended on someone else’s cell towers, all the random routers and switches in between, and the other person’s device.
My server likely has worse uptime, and if I’m hosting from home it probably has more hops to transit through it.

I believe their point was that even encrypted messages convey data. So if you have a record of all the encrypted messages, you can still tell who was talking, when they were talking, and approximately how much they said, even if you can’t read the messages.

If you wait until someone is gone and then loudly raid their house, you don’t need to read their messages to guess the content of what they send to people as soon as they find out. Now you know who else you want to target, despite not being able to read a single message.

This type of metadata analysis is able to reveal a lot about what’s being communicated. It’s why private communication should be ephemeral, so that only what’s directly intercepted can be scrutinized.

In this case however, Janelle Shane is actually quite well aware of how different types of AI works. She writes about them, how they work and their various limitations.

Her blog is just focused on cases of them acting oddly, or not how you would expect , or just “funny”.

If you have an unutilized asset, there’s pressure to get rid of it for the cost savings.
If you sell your asset at a loss, it looks bad for you and the company. Same for paying cancelation fees.

If you legitimately think that you’re going to need that space in the future, for example because you think that we’ll find an equilibrium between “everyone work from office” and where we are now, and that we’re trending towards an organic level of office need/desire higher than we’re at now, you might see selling now as the first step to needing to buy again later, likely for higher than you sold for. So you try to “mandate” the equilibrium that you expect so you’re not in a position to have to explain why you’re holding onto a dead and losing value property.

Executives spend a lot of time talking to people and having meetings. The job selects for people who thrive on and value face to face communication. Naturally, they overestimate how much that social aspect of the job is true for everyone else, so they estimate that the equilibrium will have a lot more office time than other people would.
To make it worse, the more power you have to influence that decision, the more likely you are to have a similar bias.

This isn’t an excuse of course, since you can overcome that bias simply by telling teams to discuss what their ideal working arrangement would be, and then running a survey. Now you have data, and you can use it to try to scale offices to what you actually want.

This is already a thing we need to deal with, security wise. An application making use of encryption doesn’t know the condition of what it views as ram, and it could very well be transferred to a durable medium due to memory pressure. Same thing with hibernation as opposed to suspension.

Depending on your application and how sensitive it is, there are different steps you can take to deal with stuff like that.

In this case the helicopter came because they blocked a major highway.

A helicopter coordinating police movements during civil unrest is pretty standard anyplace that can afford helicopters. That’s definitely not just an American thing.
Do you think France is eschewing using helicopters to coordinate police movements with their current unrest?

Is it? All I saw was a helicopter with decent optics, but nothing particularly special, and cops talking on low bandwidth radios.

Even when we get to actual behavior, we see the cops starting with the assumption that they’ll be just telling people to leave and planning routes to do so, before it changes to arresting people for blocking a freeway. They make sure people are notified that they’re under arrest early, and the make sure they have adequate transportation before they begin the arrest process.

Like, there’s plenty of scary and shitty things cops do, but this wasn’t one of them.

To me it’s important to ask “what problem is it solving”, and “how did we solve that problem in the past”, and “what does it cost”.
Crypto currency solves the problem of spending being tracked by a third party. We used to handle this by giving each other paper. The new way involves more time, and a stupendous amount of wasted electricity.
Nfts solve the problem of owning a digital asset. We used to solve this by writing down who owned it. The cost is a longer time investment, and a stupendous amount of wasted electricity.
Generative AI is solving the problem of creative content being hard to produce, and expensive. We used to solve this problem by paying people to make things for us, and not making things if you don’t have money. The cost is pissing off creatives.

The first two feel like cases where the previous solution wasn’t really bad, and so the cost isn’t worth it.

The generative AI case feels mixed, because pissing off creatives to make more profit feels shitty, but lowering barriers to entry to creativity doesn’t.

Depends on your level of security consciousness. If you’re relying on security identifiers or apis that need an “intact” system, it certainly can be a security issue if you can’t rely of those.

That being said, it’s not exactly a plausible risk for most people or apps.

Sure, I suppose. Or just don’t expand the system until there’s some measure of system in place to keep the AI cars from fucking around in emergency situations.

Some of the vehicles don’t have anyone in them.

https://missionlocal.org/2023/05/waymo-cruise-fire-department-police-san-francisco/

One of the incidents in question.

Big difference is that a human can be yelled at and told what to do, and we currently don’t have a good way for someone to do that with an autonomous vehicle.

It’s not nearly as nefarious as people seem to think. Effectively all applications that access web resources send along what they are and basic platform information.
This is part of how the application asks for content in a way that it can handle
It does a little to let you be tracked, but there are other techniques that are far more reliable for that purpose.

I don’t think they work the same way, but I think they work in ways that are close enough in function that they can be treated the same for the purposes of this conversation.

Pen and pencil are “the same”, and either of those and printed paper are “basically the same”.
The relationship between a typical modern AI system and the human mind is like that between a pencil written document and a word document: entirely dissimilar in essentially every way, except for the central issue of the discussion, namely as a means to convey the written word.

Both the human mind and a modern AI take in input data, and extract relationships and correlations from that data and store those patterns in a batched fashion with other data.
Some data is stored with a lot of weight, which is why I can quote a movie at you, and the AI can produce a watermark: they’ve been used as inputs a lot. Likewise, the AI can’t perfectly recreate those watermarks and I can’t tell you every detail from the scene: only the important bits are extracted. Less important details are too intermingled with data from other sources to be extracted with high fidelity.

Yes, that’s sorta why I picked that example. It’s a symbol that’s been used in other contexts and is almost entirely associated with a specific negative use case.

If you see a guy walking around with a swastika arm band, do you really think "oh, look at that man showing pride in his Hindu beliefs”?

Actually, I think that the opposite of a bad example. If I see you flying that flag, I’m not going to assume your an enthusiast of finish WW1 aviation.

I chose the swastika specifically because some other people used the symbol at some point and had it ruined for them. That’s a thing that happens to symbols, they get associated with shitty stuff and you stop showing the symbol, convince people to drop the objectionable meaning, or accept that people will think you endorse the shitty one.

They can have whatever they want, but you’ll have to forgive people for thinking that you align with people who display the same symbols as you.

I assume anyone flying a swastika is antisemitic, when to be fair, they might just be a fan of the Nazi stance on affordable housing and infrastructure.

If you have a problem with symbols you identify with being co-opted by people you don’t, take it up with the people you disagree with who took your symbol, not the people who also disagree with them.

The question to me is how you define what the AI is doing in a way that isn’t hilariously overbroad to the point of saying “Disney can copyright the style of having big eyes and ears”, or “computers can’t analyze images”.

Any law expanding copyright protections will be 90% used by large IP holders to prevent small creators from doing anything.

What exactly should be protected that isn’t?

I disagree that recognition implies you contain it. It’s much closer to a description than the actual thing, and a description isn’t the same as the thing. This is evidenced by you being able to look at a letter P in a font you’ve never seen before and recognize it without issue. If it was just comparison, you couldn’t do that.