Upgraded claude code and it's now boldly prompting me to "switch to auto mode". I guess I will switch, but can't help wondering which is worse: * An AI agent which asks constantly for me to approve bash commands, which then makes me lazy about checking the commands even though I know there's a small chance it might decide to do something bad. * An AI agent which uses another AI agent to check the commands, but which presumably has a small chance of incorrectly letting something bad through!
AI · Takes · page 5
Further down today’s ranking of AI posts that are opinions and questions: page 5 of 5.
We need to cut down trees to build data centers so kids can ask AI to tell them about trees.
if humanity ever made real digital artificial intelligence it would probably be the greatest horror man could impose upon another entity... you see, the very frailty of our organic brains protect us from so much, we carry trauma but forget pain, we dream of horror to wake up and face the day fresh, we barely know ourselves and our history, we Hyperfocus on things, and remain blissfully unaware of unfold terrors... now imagine an intelligent mind that cannot know less than everything...
The thing that pisses me off most about generative AI is that it has no sense of humour. I mean, that's just a deal breaker for me.
The source for this claim seems to be enthusiastic about AI somehow though
Are we becoming too reliant on #AI? Are we using it to allow ourselves to be lazy? I certainly feel that way. I’ve not touched an email client or the settings of my computer in months, having Claude do it all for me. Is that a good thing? By using this enabling technology are we actually losing our core technology skills? Discuss! #blind #AssistiveTechnology
Knowing exactly how LLMs work and trying to explain to `$RELATIVE` why you can't ask ChatGPT for election guidance makes me realize why Cassandra probably was honestly just fucking relieved to be murdered in the end
Just your periodic reminder that Anthropic was founded by loons from OpenAI who thought Altman wasn't worshiping digital Jesus hard enough.
no one scrolls through code ever so it would be so easy to deploy Ai ghost code in a recomp that is either deleting the hard drive crypto mining or attempting to get all your stored info
If the AI bros jump in and start calling it SI, isn't that a tacit admission that the technology has hit its peak? If this is "super" intelligence, what's the next step? Where do we go from here?
“Don’t download the Adobe clone software off GitHub. It was vibe coded; we can’t know what it does!” My brother in Emrakul, what do you think is sitting in those src repos? Binary files? i’m all for being critical of spray n’ pray AI code, but understand what you’re critiquing, first, please!
