posted in Technology
GrapheneOS defends use of AI for coding
GrapheneOS is currently defending its use of AI coding tools on Mastodon against complaints by various accounts claiming to be users.
grapheneos.social/@GrapheneOS/117243573090596184We do not understand where you’re coming from or why you’re so incredibly angry with us. It’s not justified and does not make sense.
Replying to @General_Effort@lemmy.world
I don’t like GOS devs, they are not focused on usablity, but on security (which is completely valid, but it’s a shame that GOS is the most usable mobile OS), but holy based.
If you are disputing AI utility in coding: Recent Linux kernel versions fixed up to 3 times more CVEs in comparision to releases before AI. Also, much less of those CVEs are found in the newer versions, which means new kernel releases contain better code.
Replying to @notSys@lemmy.cafe
I think the gist here is making a distinction between AI coding and AI auditing.
To the best of my knowledge, AI coding is still significantly behind human coders in code quality and security. AI auditing, on the other hand, allows a lot of previously missed bugs to come to light, and actively helps in making code better.
Replying to @notSys@lemmy.cafe
I dont think those arguing in good faith dispute its utility, they dispute the cost-benefit assessment of those using it.
If someone, for example, made a tool that produces something of objective value every time you feed it an orphan, it has undeniable utility but many would refuse to use it on moral and ethical grounds.
That’s obviously a hyperbolic example, but the logic behind it is sound and what the detractors are using to make their assessment. LLMs do something useful, but are grievously inefficient on every resource except time, when compared with traditional methods of doing the same thing. That trade-off is unacceptable to many, especially with the backdrop of the planet’s ecosystem rapidly dying.