First reason, LLMs are modeled from what humans have been doing, and the have been writing software that way recently so it's easier to mimick that to get straight to results. This reason might fade away in the future.
Second reason, something related to impedance (mis)match, a signal processing notion (when the interface between two media is not well-suited, it is difficult to have a signal pass through).
Going through intermediate levels makes a structured workflow where each steps follows the previous one "cheaply". On the contrary, straight generating something many layers away requires juggling with all the levels at once, hence more costly.
So "cheaply" above both means "better use of a LLM context" but also use regular tools where they are good instead of paying the high price (hardware+computation+environment) of doing it via LLM.
Interestingly, AIs are used to generate sample-level audio and some video, which may look like it contradicts the point. Still they are costly (especially video).
10 CVEs, remote code execution on platforms and individual developer's workstations.
The public announcement is three days old and I couldn't find any mention here, which is kind of surprising.
Submission title here is the HTML title. Visible title in the article "The Mother of All AI Supply Chains: Critical, Systemic Vulnerability at the Core of Anthropic’s MCP"
You're defeating your own point. CSD in practice breaks a basic feature of the desktop : knowing at first glance which of the windows will receive whatever you type on the keyboard.
For eons the standard was: the only one with the title bar showing the theme accent color. That is consistent, predictable, keeps the user in the flow.
With CSD each app does whatever inconsistent thing they can fancy. You type and oops deleted something in the wrong window.
Alas now even many default SSD setups fail at this (selected and non-selected windows look pretty much the same) and keyboard-first workflow is much hindered.
Been there, done that. Then dropped dark mode entirely. Strong filtering of blue light (use redshift or whatever your OS provides) beats dark theme IMHO, without the downsides.
I default to light mode these days. Astigmatism does not goes well with small, bright details against a dark background. Then I can turn the brightness of the display way down.
Probably a "compatible" card that would behave like the standard of the era only after being setup by a DOS driver. Linux, like DOS programs, would find it only after such a setup. Perhaps even the card setup would not survive a reboot.
> Then I personally don’t care about needing JS to view the site (it’s 2024…)
In 2024, shouldn't hacker news visitors get that receiving a passive document to display locally is a thing, and having to run unaudited code from random strangers on your machine all the time is quite another? Then reopen the conversation about browser security?
Or, instead, get that simplicity is our ally, but unneeded complexity is the root of much evil?
Visit e.g. any stackexchange site with script disabled and be enlightened.
Not even certain this will be enough. Hawking radiation encodes the information that was thrown in. Sure at present it's a perfect mess, but who knows?
Not really aligned, but somehow reminiscent of this other dystopian future of what happens when a platform makes money not on products but on ads https://www.smbc-comics.com/comic/2012-01-12
First reason, LLMs are modeled from what humans have been doing, and the have been writing software that way recently so it's easier to mimick that to get straight to results. This reason might fade away in the future.
Second reason, something related to impedance (mis)match, a signal processing notion (when the interface between two media is not well-suited, it is difficult to have a signal pass through).
Going through intermediate levels makes a structured workflow where each steps follows the previous one "cheaply". On the contrary, straight generating something many layers away requires juggling with all the levels at once, hence more costly. So "cheaply" above both means "better use of a LLM context" but also use regular tools where they are good instead of paying the high price (hardware+computation+environment) of doing it via LLM.
Interestingly, AIs are used to generate sample-level audio and some video, which may look like it contradicts the point. Still they are costly (especially video).