“Native” sort of loses its meaning in this context - this new client is not “native” in the sense of using the existing GUI tooling for each platform (Cocoa for macOS, any of Microsoft’s own frameworks on Windows, primarily Gnome/KDE on Linux).
It is using a universal framework in GPUI, so the experience will be the same on each platform.
It is “native” in the sense that it’s not web technology on any of the platforms.
The price of memory-thing is probably not that important as compared to the AI-assisted coding thing. Making things in Rust when you have access to AI tools is dead simple, which was always the argument for making things in Electron. Porting things from one language to another also happens to be one of the sweet spots for AI coding tools, they thrive on tasks that have very clear specs - an already existing codebase with tests is just about as clear of a spec as you can get.
I’d be willing to bet $100 that AI assistance was involved in creating this new client, yes.
Similarly I’d bet the same amount that AI assistance was involved in building the former electron-based client
The streaming market leader is apparently adding an optional Al package to its Spotify Premium subscription, reportedly costing $4 to $5 per month, which could make the service’s Premium subscription up to 39 percent more expensive.
This sentence is nonsensical - if you get an optional addon, then it’s per definition not a price increase.
It’s like if I went to the grocery store, bought the same items as before and then added several more to the cart, and claimed that “my grocery bill has increased”. Apples and oranges.
Did you know Gabe Newell donated $20 million to OpenAI?
One would do well to remember that OpenAI were once upon a time the people who made highly sophisticated bots for Dota 2.
I’m not convinced their layoff rates are that much higher than other companies. It’s definitely a risk to be dealt with if intending to attempt to migrate to the U.S.
I’d guess that chat sessions use a trivial amount of tokens in relative terms.
For tech positions in tier 1 tech companies you get out ahead so far it’s not even funny. It’s pretty easy to run the numbers on how much you have left after expenses for each market
If I had to guess, for the massive salaries involved
There’s a difference between being a prepper and having some amount of societal resilience.
When national governments are the ones issuing recommendations, you can assume it’s the latter and not the former.
You can still get this level of service if you pay the astronomical prices for first class tickets on top tier airlines.
It’s worth remembering that there wasn’t really an economy class at this point, all tickets were basically reserved for the rich.
Classifying air travel as a necessity even today I think maybe stretches the definition of the word
I’d guess it’s the new tariffs making the previously very cheap purchases no longer that cheap
Success at letter counting is not a particularly good way of predicting model capabilities.
Apple typically do not do completely new ideas - they are however pretty proficient at refining existing ideas, and charging an arguably unwarranted premium for it.
I don’t think there’s even an expectation of privacy to be had on a public discord, in open channels. Closed channels are presumably not indexed.
This is false, Microsoft does not own Discord
To be clear, this is really good - so much information has moved from being accessible on the open internet in the form of public forums, to being inaccessible in the form of public discord servers. This brings back that lost information.
I struggle to see actually needing more than $2m. $1m is probably already enough
It doesn’t hurt to have the component that is optimized for drawing things to the screen be a bit more involved in drawing things to the screen.
Strictly speaking necessary? No. Does it hurt? Not at all.