If you point out a bug to an LLM and it's says "I now see that if statement is wrong", it's because somewhere in training data is a stack overflow post matching a similar pattern with the comment "this if statement is wrong", not because the LLM actually gamed out the logic.
I know some people get out annoyed at over-simplifying LLMs as "fancy autocomplete" but this kind of problem demonstrates why that is still accurate: LLMs can string together text seen previously, even with some changes to replaceable values, but no LLM is actually thinking through logic.
0.0041
(Phone screen in casual indoor lighting, no calibration or a color suite)
Also any testbed aircraft with experimental engines/fairings mounted...
Without steering assistance that bike would be flipping over at the lightest touch as the steering would cut to the side.
(This is also why I'm excluding things like feeds and labeling from this description)
(This is a very simplified model, while the actual client and appview are separate, they both are "how you, the user retrieve posts from the network to view them")
I was proficient back in high school, but every time I try to pick it up again I get a few lessons in and give up for a few years
Why do we have to live in a time when the only way to buy most stuff is online yet every online order sends a seperate email and text message every time your package moves five feet?
Email me a single tracking link when my order ships and I'll check it if I want updates.
Lies! This isn't AI! It needs a random number generator for a slight variety of possible results for any one input state to be AI!
Even on foot out here in Western MA, major rivers of cool air along low forested stream areas followed by major heat on flat pavement.
Keep in mind that the average person's (not Twitter/bsky user) exposure to Musk is headlines for news stations about how "Elon Musk says cybertruck will be the official self-driving car of Mars colony by end of 2027" reported as serious fact
It's true that there will probably always be need for some sort of car in cities for some limited uses, but bringing a car into the city is always a compromise and must be weighed carefully for the impacts it will cause.
Wait I skimmed past this post when I saw "old Boston" because I'm not one of those folks who was around Boston from way back but - uhh - I lived there in 2017! That's not old Boston!
Someone is absolutely going to launch a startup doing AI mine detection and it'll either "work" (declare the whole area impassable) or "not work" (in very obvious ways)
You wouldn't be doing good communism unless you ensured that thinking about it was evenly distributed
My understanding of Elon comes exclusively from breathless hype pieces put out by tech journalists and those AI slop Facebook stories of how Elon personally had wise words of business knowledge for that random poor kid on the sidewalk who was all alone with no followers and also it was his birthday.
For example, Eurosky is just a PDS, it still depends on bsky for everything else, but Blacksky runs their own moderation, their own relays, their own app view, etc. You can use Blacksky regardless of what bsky thinks, all bsky can do is maybe limit their own users from seeing you.
That's where the "most people are still on bsky infrastructure" from my original thread quoted above comes in. Yes, they control a huge proportion of the overall network right now, but more and more alternatives are coming online.
In short, Bluesky PBC can control what users hosted on its PDSes do and what users using it's Relays, App Views, and moderation service (labeler) see, but if you and the audience viewing your content aren't on their infrastructure, control is up to whoever does host that infrastructure.
(And in keeping with the rest of the ecosystem, anyone can host a custom feed with whatever algorithm they want.)
And Bluesky doesn't have algorithms by default in how it shows users content. Your default feed is just a chronological feed of everything you follow. Different feeds can have algorithms that change but in the default feed bsky isn't tweaking who sees what.
Bsky can choose to label your content so users subscribed to their labeler see their choice of labels, and in an extreme circumstance they could block a whole PDS that breaks their terms from the users, but they can't choose to restrict your account directly at all.
Moderation belongs to the labeler (and PDS). Choosing to keep your data online is up to the PDS (they can remove spam accounts or ToS violations from their servers) and labels applied to an account are only active for those who subscribe to a labellling service.
(Angled lidar directly behind/beside the vehicle feels like autonomous, but that's also more sensors than most autonomous are currently using, which might indicate mapping?)
Also, that does look like it could be an autonomous test vehicle, but it might just be a lidar scanner platform for doing environment mapping? Production autonomous vehicles usually won't use racks but test or mapping vehicles might.
Not a vehicle waymo has operated and most of their current models have custom sensor fairings, not exposed roof racks and rails. Also no waymo logo. Not sure what vehicles they're running in the UK but I doubt that is it.
Screenshot of a weather app cropped closely on the temperature section. It shows a current temp of 29°F, a high of 72°F, and a low of 43°F
I feel like there's a bit of a delta between the current temperature and what it will be later today
