setsubyou

The comment I replied to said it was different from recording audio because when recording, they can tell it when to record instead of listening all the time, so that’s the only part I’m arguing with.

In this regard nothing changed:

• You can still tell it when to record, because you can turn the feature off
• If you didn’t turn off other Siri features before, it was already listening all the time

I feel like half this thread doesn’t understand what their devices are doing. Latching on to a new way in which it can record all the time, while dismissing all the existing ways it can already record all the time. How are you going to protect yourself from anything if you can only recognize this when there’s something like a transcript that you can see that makes it super obvious? Most people here are probably already carrying two or three devices around that can record them but just don’t produce a transcript.

That’s not new though, it also listens for Hey Siri all the time like all modern Apple devices have been doing for years. You can also still turn it off if it’s your own watch.

It’s a matter of framing. If you compare them to older datacenters they are atrocious because they’re bigger and consume more energy. But in the grand scheme of things all things IT pale in comparison to e.g. producing beef. Now with that framing you can’t sell AI hate though because people like beef, so it’s always just comparing large datacenters to smaller datacenters with the obvious results.

People don’t really understand this yet. There are LLMs that can do useful work running locally on 5 year old laptops (ok, only if it’s a Macbook Pro, but still). Qwen on consumer hardware isn’t close to Claude Code using cloud models, but it can do real work and this is the floor, the absolute minimum of AI we will have going forward. No matter what happens to the industry and the bubble, it will not go away. With any other outcome other than a total collapse of the bubble, AI will be more powerful than that.

I thought the (former) VP in charge of it left for OpenAI not long ago.

There also were more than two PC platforms (and we barely have two now; Apple was pretty close to going away too). I miss being the Atari ST guy in a circle of people who all had different types of computers. It didn’t last very long but it was way more fun.

Yeah on mobile I think it’s okay. Still a bit overloaded but a far cry from what Japanese web pages can be like on desktop browsers. Sometimes they’re like web design stopped in 1999 and since then all they did was add more blinky banners around the actual content.

That’s fine. All I’m saying is that the article claims that their stance is an eight year old can waive rights, while also claiming indirectly that the mother waived the rights too. Because it says that she placed parental controls on the account, which in the case of Roblox means she made an account for herself and agreed to the same terms.

I have no idea whether the terms are enforceable or not. I’m only commenting on the article sensationalizing by making it look like the company says the 8 year old agreeing to their terms is enough.

Although maybe one thing to add could be that the child could not have gotten access to chat either without the mother enabling it for her. The article makes it sound like children are immediately exposed to all sorts of things on Roblox but the reality is unverified accounts are heavily restricted, and for a child, the restrictions have to be disabled by a parent.

Is it their stance? The article says the mother “placed parental controls” on the child’s account. This means the mother had an account too and agreed to the terms as well.

I think not understanding what they’re committing is more dangerous than before

This kind of reasoning applies to every new tool. 20 years ago I was saying the same thing about work from co-workers who started programming on Java and didn’t understand what their beloved HashMap actually does under the hood.

Eventually we adapt to either the new tools or to the new dangers, the ones who don’t just become fossils.

I always thought security questions were dangerous, but I did in fact not know that quizzes that exploit them exist in the wild.

The Apple office apps are still free, but they do have an add-on subscription too now.

Apparently he’s 63. So he would have played in the streets in 1960ies Taiwan or Thailand. Both countries where most people couldn’t afford cars at the time.

Here in Germany, we can still play in the streets in 2026. We have streets designated for it.

Nowadays agents like Claude Code can run autonomously for hours just given a goal description. It doesn’t take a lot of human effort at all to set up a bunch of sessions, and these companies don’t limit how many instances you run in parallel. Agents can also spawn sub-agents that run in parallel if a task calls for parallelization. Whether all this produces good results is a different story, especially if you don’t put enough effort into the goal description. But burning tokens as such is not difficult.

Even workflows where you’re just chatting with an agent can burn a lot of tokens. When you’re chatting with an LLM, the entire history becomes part of the input each time you send something. This also applies to tool calls, so if the agent decides to read 20 files before it can work on your request that’s 20 times a file gets added to the history and 20 times that entire growing history is then sent back as input to drive the agent’s next step.

Coding is more affected by this than many other applications because even a new conversation tends to start with the agent gathering a bunch of source code files, and then the response to a task is not just a bunch of text once, but a sequence of tool calls to make edits across files, build, run tests, react to test failures, and so on, all for one actual human prompt - but in reality a back-and-forth between the LLM and the harness with a quickly growing history.

Wondering how much the AC could even do if the door is left open. But maybe the hassle is also having to turn it on again every time?

That but also if you’re not training and hosting your own model, your scanner is just subject to the same restrictions that your LLM provider applies to you on top of all the architectural problems.

1 in base 10 isn’t 1/10 and in hexadecimal it’s not 1/16.

Decimal integers in base pi are 1, 2, 3, 10.2201…, 11.2201…, 12.2201…, 20.2201… and so on.

Basically: 10.2201… = 1 * pi^1 + 0 * pi^0 + 2 * pi^-1 + 2 * pi^-2 … which approaches 4 as you add digits.

But 1 is just 1*pi^0

It’s not user replaceable but it’s not actually that hard to replace it, so this basically just takes work away from all those little repair shops. Still good for customers but probably not much of a difference for Nintendo.

None of them are profitable, they’re all selling their services at a loss