basch

That’s not free. Apples Apple ID management is atrocious. How do you password reset an ID you own?

They need to move to having the students ID under both their parents and the school, detachable from the school when a kid moves.

The devices need to be enrolled to the org and then act as thin clients so any kid can log onto any laptop, not have the laptop locked down to a specific kid.

Apple Parental controls are either controllable on the kids device or one parents device but not both at the same time and definitely not two parents at the same time. Whose bright idea was it not to allow two parents to see and manage a kids settings from their own devices, at the same time. That’s a lot of the world that doesn’t appear to anticipate two parents wanting to manage a kid from separate accounts, but Apple should know better.

When Jobs took over Apple he didn’t have a device in a quarter of the world’s pocket.

Apple absolutely needs a management layer, both for school and for family. Apple family controls are sorely broken.

300*500 kids is 150k and the difference between a school choosing chromebooks again. This is priced against the $450 Chromebook.

RAM need shave changed slightly post nvme. Normal people apps can swap just fine with a pretty seamless experience. Average people aren’t opening single files that can’t fit into 5gb of ram.

RAM is also an insanely high percentage of computer price right now. https://www.tomshardware.com/tech-industry/hp-says-memory-co...

If only a human could edit the output before posting.

once they get a phone they switch from cards to go.

Their extended popularity should be viewed closer to praying on addiction than an accomplishment of culture. The phone game, by removing the exploration of outside aspect, became just mindless collecting. If you have ever been to an event as an outside spectator, there is as much anxiety in their air as joy.

Pack opening is the equivalent of pulltabs or scratch offs. It's an asset class closer to beanie babies than baseball cards, because there is no tie to a rookie going on to have a huge career or anything but artificial scarcity.

That’s what I’m saying. Voice as the input requires a completely new ui paradigm, and chat / chatbot isn’t enough.

Interesting idea. Voice modulation software that converts your voice to how you sound in your own head.

I imagine it sort of like an eye test. Play two samples, A/B test in rapid succession until you get a match.

Microsoft Recall wasn’t popular because it was Microsoft, but it is actually what people want.

You’re the first comment as I scrolled challenging the premise. But your solution amplifies what I see as a problem.

I’d like to add, that by forcing myself to look up the answer every time I have happy accidents where I learn new ways to do things.

It’s a skill to be willing to unlearn and always presume yourself ignorant, even if you do know how to do it. It’s like confirming “is my way still best practice.”

I’ll disagree with you a little. The reason I don’t use voice is because of context switching.

With a mouse and keyboard I can switch windows.

With my voice, the computer can’t yet automatically determine if I am dictating a transcription or giving editing commands. What I really need is the interpreter listening to me to intuitively to know whether I am in the equivalent of VI command mode or insert mode.

It is the roadblock to not needing a screen at all, right now I want to visualize whether it understood me correctly because if it didn’t switch from insert to command automatically, I now have all my commands written into my paragraph. I also don’t want to listen to the computer talk back to me to confirm it listened. I want to just keep going, to keep narrating my thoughts and trust it’s doing the right things, not having to check. Having it slowly chime in to repeat that it listened derails my flow and train of thought.

TLDR The future of voice is headless vi.

Which you essentially cannot do

It is inherently randomized

That’s not true in practice thought https://sulbhajain.medium.com/why-llms-arent-truly-determini...

and in this case, less competent could mean more collateral damage, right? By doing the task competently you make less accidental misfires?

A strong argument for the other players to play ball. "Let us use your service or we will attach our kill trigger to a wildcard whim driven vendor."

I have them in the web as well.

Model training is completely different than keeping a summary of chats on the side and injecting it as context.

In my Gemini app, when k click new chat and click the filters button I have “Personalize Intelligence: Personalize chat when helpful.”

It is on every time I click new chat. Maybe you need to enable it in settings first. I can disable it to have a clean chat without personal context, but preserve the chat history unlike temporary chat.

Language models aren’t “programmed” though.

If I “convey understanding” I transfer it from one person to another. Consciousness does not transfer with understanding.

Some people argue that consciousness emerges in early childhood. I can get an infant to understand what I am saying even if they aren’t conscious.

There’s a toggle in every new Gemini chat to turn off personalization for that chat. I assume you need to make sure it’s mom globally first?

what is understanding but encoded meaning distilled into pure structure, in both cases a property of a pattern?

No need to gatekeep the word "understanding" behind subjective human experience eg qualia.

You would think, wouldn’t you?

And yet they waited until ChatGPT was a thing and threw Bard together overnight in response.

ChatGPT and Gemini has cross conversation personalization. I believe the former is off by default and the latter is on.

agree to disagree. encoding a meaning is understanding. I cited a source using the word in the same way.

A the time, it was a free language translation tool. You weren't paying for transformers in 2017.

Id really invite people to read the google blog post. https://research.google/blog/transformer-a-novel-neural-netw...

Google chose the word understanding.

Semantics. Its a encoded position that represents meaning in a way that is useful and reusable. That is "understanding." It's a mathematical representation of grasp.

It's honestly disheartening and a bit shocking how everyone has started repeating the predict the next syllable criticism.

The language model predicts the next syllable by FIRST arriving in a point in space that represents UNDERSTANDING of the input language. This was true all the way back in 2017 at the time of Attention Is All You Need. Google had a beautiful explainer page of how transformers worked, which I am struggling to find. Found it. https://research.google/blog/transformer-a-novel-neural-netw...

The example was and is simple and perfect. The word bank exists. You can tell what bank means by its proximity to words, such as river or vault. You compare bank to every word in a sentence to decide which bank it is. Rinse, repeat. A lot. You then add all the meanings together. Language models are making a frequency association of every word to every other word, and then summing it to create understanding of complex ideas, even if it doesn't understand what it is understanding and has never seen it before.

That all happens BEFORE "autocompleting the next syllable."

The magic part of LLMs is understanding the input. Being able to use that to make an educated guess of what comes next is really a lucky side effect. The fact that you can chain that together indefinitely with some random number generator thrown in and keep saying new things is pretty nifty, but a bit of a show stealer.

What really amazes me about transformers is that they completely ignored prescriptive linguistic trees and grammar rules and let the process decode the semantic structure fluidly and on the fly. (I know google uses encode/decode backwards from what I am saying here.) This lets people create crazy run on sentences that break every rule of english (or your favorite language) but instructions that are still parsable.

It is really helpful to remember that transformers origins are language translation. They are designed to take text and apply a modification to it, while keeping the meaning static. They accomplish this by first decoding meaning. The fact that they then pivoted from translation to autocomplete is a useful thing to remember when talking to them. A task a language model excels at is taking text, reducing it to meaning, and applying a template. So a good test might be "take Frankenstein, and turn it into a magic school bus episode." Frankenstein is reduced to meaning, the Magic School Bus format is the template, the meaning is output in the form of the template. This is a translation, although from English to English, represented as two completely different forms. Saying "find all the Wild Rice recipes you can, normalize their ingredients to 2 cups of broth, and create a table with ingredient ranges (min-max) for each ingredient option" is closer to a translation than it is to "autocomplete." Input -> Meaning -> Template -> Output. With my last example the template itself is also generated from its own meaning calculation.

A lot has changed since 2017, but the interpreter being the real technical achievement still holds true imho. I am more impressed with AI's ability to parse what I am saying than I am by it's output (image models not withstanding.)