This is almost certainly the issue. It's very unintuitive for users, but LLMs behave much better when you clear the context often. I run /clear every third message or so with Claude Code to avoid context rot. Anthropic describes this a bit with their best practices guide [0].
This'd be a valid analogy if all compiled / interpreted languages were like INTERCAL and eg. refused to compile / execute programs that were insufficiently polite, or if the runtime wouldn't print out strings that it "felt" were too silly.
It depends from which vantage point you look at it. The person directing the company, let's imagine it was Bill Gates instructing that the code should be bug free, but its very opinionated about what a bug is at Microsoft.
> I don't know what it is, but trying to coax my goddamn tooling into doing what I want is not why I got into this field.
I can understand that, but as long as the tooling is still faster than doing it manually, that's the world we live in. Slower ways to 'craft' software are a hobby, not a profession.
(I'm glad I'm in it for building stuff, not for coding - I love the productivity gains).
This isn't likely to be a hardcoded type of classified response. I think this response is literally "you offended the model persona's sensibilities." But, yes after the first denial the models will double down.
AI still a mind virus, vibe coding hasn’t taught me anything. It just allowed me to ship microservices that I dont understand. In the past I would’ve had to learn it through tutorials and docs, and in the end I would be in a better place with my new found knowledge.
But with vibe coding, there is no knowledge transfer.
>But with vibe coding, there is no knowledge transfer.
No. I think you're wrong.
'Vibe coding' (a phrase I hate) introduced me to so many different software packages that I didn't even knew existed. Now, with no LLM subscription, I still use many of the software packages that I discovered via LLMs in my own work.
Like any other media, whether or not you use it as a springboard for learning is up to you. Literally every other store of knowledge had critics like you with that same exact opinion (boob-tube, brain-drain, couch-potatos, book-worms) -- and every single one of those forms of media ended up tremendously useful for us.
When you watch YouTube you don't have to go watch the most popular asinine shorts; you could use it to learn a language or perhaps some history or a math lesson.
Is it bad because a lazy person can be tempted into skipping the work and making the LLM do it? That's because a lazy person is involved -- not because the technology 'is a mind virus'.
P.S. 'mind virus' is the stupidest nu-speak crap term I have heard in years. In fact the whole "I don't like the thing so i'm going to give it a scary name" concept has become so tired in recent years that I have a hard time keeping up with the new dictionary -- is this what getting old is like in every era or is this special?
I happily embraced it, to each their own I guess. There are folks who mainly work on their mac/windows laptops and just ssh into their workstation, but IT gives us way more freedom (full sudo access) on Linux so I can customize a lot more which makes me a lot happier.
Capture card your video output to a Raspberry Pi, connect the Pi to a XIM over USB and have it pretend to be a keyboard and mouse, and you have an aimbot. I would guess XIMs put 0 effort into ensuring it’s a real KBM and not a dummy device.
True that wallhacks aren’t possible via peripherals, though. You might be able to get some level of info from the audio output and map knowledge, but nowhere near the same as true ESP.
That's because Edison vs. Tesla was really mostly theoretical with some funds for engineering (and marketing). The space race is really mostly engineering which requires lots of money.
reply