I'm not sure anyone will read a book nowadays.
For those of us that do, a book would be very welcome.
@Stigma Look what you are missing. Available from all good council estates, surprisingly cheaply because they have enough benefits to go to Spain and get back in time to sign on thanks to enlightened socialist governments.
@MentORPHEUS I moved out of Bangkok to an outer district. If I get up high and look back towards the city there is a clearly visible globe of smog hanging over the place.
I went back in to take part in the Songkran festival and by the end of the day my throat was sore and my eyes stung. I was coughing up the smell of Bangkok for a good few hours after.
It’s probably a combination of volume of traffic, inefficient traffic system, heat, humidity and poor vehicle regulation.
An outcome polar opposite to what lobbyists and partisans predicted, that "environmentalist wackos" would end mechanized transportation in the name of their "green agenda."
This speaks to my larger point. I’m not overly concerned with something like AI now because we’re witnessing the optimisation take place at unprecedented speeds. The “smog” of AI is a tiny moment in the history of its development, so to speak.
I’m not very impressed with the Luddite-esque attitude people take up with AI, but resistance is part of the optimisation process I guess.
Asking it culturally significant questions from a partisan position is also the worst use case imaginable, and is such a depressingly short sighted litmus test for the capabilities of these things.
I’ll be mad if we’re dragged down by lowest common denominator baying at AI for their factoids.
Read MoreGrok is this true?
@Bozza This is a great post, but it has fatalism built in.
With each cornerstone of human development there is a period of uncertainty and upheaval. Agriculture, domestication of animals, the wheel, the car, the plane, the rocket.
Anthropic create a version of Claude that is exceptional at exploiting security vulnerabilities. The translation of that is, security measures across the board just advanced exponentially out of necessity. Understanding these exploits and identifying them is a good thing. Future OS security will be leaps and bounds ahead!
I don’t think this is a one way street to dystopian futures. AI is a software abstraction of human progress, built upon the success of what works.
As for hallucinating, it is pretty real. In your use case I'd try to search eg 'is Claude hallucinating on only user supplied data'
Ah yeah you’re right on that. I do a cycle between each input session, I vet the output for inconsistencies or hallucinations.
In one case, it had applied the model I defined for one rule set to another. I caught it, because it had introduced information that didn’t share context. Well anyway, I tightened the rule set and repeated the cycle!
@Vermillion-Rx I’ve not looked into the background of the complain to be honest. I can’t really speak to that.
The efficiency is pretty fucking good, though. If I remember rightly, Claude asks for permission to access the directory of the project and/or create one. If I’m not being naive then Claude doesn’t have carte blanche access. I’m sure that would have been reported by now.

