Has anyone checked if she’s an AI? That’s the kind of thing an AI would screw up.

Whenever the topic comes up with leadership, I try to explain it in financial terms.

Tech debt is just like financial debt. There are times when its appropriate and necessary to take on some debt. But debt accumulates interest charges. If we just keep building up more debt without ever paying it down, it’ll eventually bankrupt the company.

The engineering team doesn’t always know when the finance team is accumulating debt or paying it back, but we trust that they are doing so appropriately.

You don’t always know when the engineering team is accumulating or paying off debt, but you need to trust that when we say we need time to to pay down tech debt, we’re serious. We’ll all be out of a job if we don’t.

They don’t usually like to hear it, but when put in those terms, they don’t have an argument against it. I’m sure if we could provide a statement showing we had $478,562.78 in tech debt at 4.75%, they’d be more understanding about paying it down.

Thank you for your service.

Asked to verify my identity with a code to my phone - standard

No, absolutely not standard. This is where red flags should go up. If your bank texts you a code when you log in, then that’s what the scammers are doing (trying to log in as you, triggering the website to send you the code to confirm that it’s you logging in (except it’s not you, it’s them), and then getting you to tell them the code so they can finish logging into your account.

What, you expect journalists to use the correct words to unambiguously convey a concept?

Nitpick: GPS satellites are not in geosynchronous orbit and are at about 12,540 miles up.

https://en.m.wikipedia.org/wiki/Global_Positioning_System

Doesn’t really negate your point, though. That’s still a lot further away than a ground based system would be.

Yup, hadn’t thought of that. Sigh.

This smells a bit like an orphan crushing machine. Is this just a private company offloading moderation duties onto already overworked public school teachers?

Just the same, smashing cars probably won’t endear the generally-docile public to the cause. I would say most people (i.e. the support force necessary for widespread change) don’t want to be associated with violence.

Which is why we should be considering the real possibility that these are false flag attacks. Want to turn the general populace against peaceful protesters? Engineer false flag attacks that make the protesters appear violent. Now you can arrest peaceful protesters and the general populace will turn and look the other way.

There have also been an uptick in incidents of arson, vandalism, and violence against Tesla showrooms that, while unrelated to the protests, have led to Musk and President Donald Trump labeling them “domestic terrorism."

It’s perfectly reasonable to think that at least some of these could be false flag opperations orchestrated by the Trump administration to give them cover to arrest innocent people and eventually declare martial law. The more that possibility is part of the mainstream conversation, the more wind it takes out of their sails towards accomplishing those goals.

There are lots of Tesla protests going on every weekend. See https://actionnetwork.org/event_campaigns/teslatakedown to find one near you.

Also see https://www.mobilize.us/indivisible/ for even more (not necessarily Tesla specific) protests going on all over the place all the time.

Participate in what you can.

It can respond to users’ verbal expressions — such as “I am feeling cold” — by automatically adjusting temperature

LOL. Who in hell wants to talk to appliances like that? If I wanted to talk to an appliances at all (not really) it would be to give very explicit instructions, not some vague complaint.

and can “chat and gossip"

Translation: “We couldn’t figure out how to get an LLM to not do this, so we’re just going to call it a feature.”

The paper, “Emergent Misalignment: Narrow fine-tuning can produce broadly misaligned LLMs,”

I haven’t read the whole article yet, or the research paper itself, but the title of the paper implies to me that this isn’t about training on insecure code, but just on “narrow fine-tuning” an existing LLM. Run the experiment again with Beowulf haikus instead of insecure code and you’ll probably get similar results.

See teslatakedown.com to find a protest near you.

Git outta here with that thar metric mumbo jumbo!