T156

Yes, Saint Olga of Kyiv.

ITAR has severely restricted access to nav system tech export, but there was always this open question about reckoning position by visual cues rather than GPS + dead reckoning. We all knew it was technically feasible, but we assumed that having “visual” cues was further out because AI at the time wasn’t up to the task.

There’s always going back to the old-fashioned method of having a trained pigeon inside of the device instead of a sophisticated computer vision system.

Or some old cheap/free one. A nearby university open day recently had a stack of free 8 GB USB sticks they were giving out.

At least one major manufacturer also outright pulled out of the consumer market entirely. There’s no guarantee that they’ll come back, or else they would have just kept the consumer arm.

Microsoft seems intent on pushing people to alternatives. Not only did they quietly remove the higher RAM recommendation, but now the OS is more expensive.

So what people are likely to get an underpowered computer that costs more if they have to buy a new laptop. On the tail of Apple recently releasing a relatively nice and cheap laptop to much fanfare, no less.

It sounds like we’re about to see a repeat of the Windows Vista days. Where people hate Windows because it’s slow and horrible, and it is slow and horrible because it’s running on an underspecced machine barely meant to run the previous version of windows.

It is. The Echo Dot saves the password to your account, which the Fire Stick would have pulled from.

Especially since if you’re choosing that, you usually have the device handy. At which point a direct transfer is easy to do, and that’s been a feature for a while.

It’s also good advertising. “We’ve powered this data centre with our solar panels, think of what it could power for you”.

Bad news, DDR3 prices are also going up.

Odd to think that the Windows Vista days are coming around again.

Wasn’t that one of the reasons that people didn’t like it? A lot of manufacturers basically just crammed it onto XP-spec machines, so it ran horrendously, and was generally awful to use.

Especially since when you ask it those questions, it’s still massively limited.

I once tried Apple’s replacement AI Siri thing, and asked it what the date would be next Tuesday. I got “I can’t help with that”.

Regular old Siri handled that just fine.

Even then, neither worked with “When’s the last occurrence of this event on my calendar?”. They’d either go “I’m sorry, I don’t understand”, or happily tell me when the next event was.

Neither of them seem like very difficult questions.

Pretty much. The implied plan is that they want to replace the keyword/rule-based automatic moderation system moderators have access to with LLM-based automation.

So if you had a subreddit that banned all use of the letter e, you could whack a rule up to remove according to criteria .*[eE].* which would have it remove all instances of the letter e.

The new system supposedly replaces it with a prompt-based system, where all messages run through an LLM, and you write out a prompt informing it what it should remove (“remove all messages containing the letter e”).

Is it actually doing it in-device? Given that it’s attached to OneDrive, I was rather under the impression that it was just offloading the processing to their servers.

Interesting. I would have expected cameras to be more resilient between both their internal filters and the sensor material being tougher than human eyeballs, so anything that could overwhelm that and burn out a camera sensor would be able to ruin someone’s vision almost instantly.

That’s a bad idea. You don’t want to get a stranger, or yourself in the eye.

Anything that would do meaningful damage to a camera sensor will easily do that to human eyeballs if it refracts or gets reflected.

Especially for cheaper high-power lasers, which may not properly filter out invisible wavelengths like IR.

Human brains are also considerably more efficient than an artificial neural network. Most modern ones only replicate the function of a human brain and its neurons in the most basic of aspects.

Things like encoding data itself in how the signal is transmitted are still pretty uncommon in artificial neural networks, for example.

We’d probably need somewhere close to several times the size and computing power of a human brain to even start coming close with current technology, assuming that it is possible to begin with.

It’d seem a bit suspicious, wouldn’t it?

handcuffed device

I’m surprised he wasn’t stopped. I can’t imagine the authorities would look very kindly on someone handcuffed to a thing they’re holding, encrypted database or no.

This coming off the end of the OpenAI/HuggingFace business does come across rather like Anthropic is doing a “actually, our models did it first, and better”.

Didn’t the company also recently have a spot of bother with the US government recently, where their new models were banned because of all their rabble about it being too dangerous? This hardly seems like it would help their case.

It’s a text completion model/glorified Markov chain. Of course it can’t input sanitise, it was never meant to do this to begin with.

The tool calling integrations that let it do more are basically making it add a markdown code block in JSON format into the user message, where the middleware intercepts it.

The input and “instructions” are the same thing from its perspective. There’s nothing special that differentiates the two. The user input text, so it will output text, following the most likely sequence from its training.