Don't confuse me with facts! (And, thanks... I have since seen the articles about how it works based on word selection from a subset of the words that would have been randomly selected by the "heat" of the model.)
MangoCats
Does it do parametric model designs?
Yeah, I "vibe coded" a timesheet app in about $15 of Claude subscription credits. Actually did it twice: $10 for the Python version, then migrated that to Go for $5 - it does run a lot faster in Go while simultaneously consuming less power / resources. It makes a couple of nicely formatted timesheets, serves multiple clients with multiple service providers, tracks everybodys' shifts, providers can log in from their phone and accept/drop shifts, schedulers can create new open shifts for providers to consider, etc. etc. - been using it for about a month now, so far the math is flawless - as compared to the human+calculator math that was being used before which seemed to have at least 1 error per 10 shifts computed.
Everybody should have been looking at the quality of their output before LLM tools were even a thing. They never should have stopped, if they did, that's on them just like Tesla drivers napping at 80mph.
That's really about the time and effort you put into the construction of the model and what you're trying to model in the first place.
For the most part, the LLMs are really bad at building say - a cat with realistic long fur. However, they can be good coloborative tools teaching you how to use Blender to build a decent cat with realistic long fur for yourself - speeding up that learning curve quite a bit over reading "Blender for Dummies" examples and building from there...
It's like comparing a stone sculptor to a welder where the welding tools started as magnifying glasses in the direct sun a year ago worked up through acetylene torches and now you've got 400A precision TIG arc welders to screw up with...
I'm going to guess if you expressed what you want to a decent LLM, it could generate it for you to copy-paste.
Seriously, I have been screwing around with AI for coding for 12 months now. The prompts I had to give 12 months ago to make the LLM generate decent code are mostly un-necessary now, the models have learned to do those things for themselves, so I am now prompting for the less-obvious stuff. Whatever they're missing, that's what you need to be prompting.
How about: AI Generated 3D models flood the free / open source repositories, anything wrong with that?
Sounds to me like the people who designed Luna should have had Luna scaffold the appropriate deterministic tools instead of smushing all the policy decision making into 200K tokens of working space (roughly the brain capacity of six bumblebees.)
Management has been overrelying on "data dashboards" to justify their hire/fire decisions for decades. This "dashboard" just happens to be capable of producing an articulate paragraph stating why the decision matches established patterns.
If you dig a little you can get the articles explaining better, but basically: yes, you have to have the weights, and a secret key... Given those, they split the words that might be chosen at each point into two groups and only choose from one group.
Later, you can check for the watermark by splitting the same way and seeing how many times the words in the text match the group that fits with the secret key. Normal text would hit a 50% match, but unlatered AI generated text would hit 100% match, so... somewhere higher than 50 you start to suspect AI involvement, possibly edited, and by the time you get over 99% it's virtually certain that AI had some involvement - particularly if the sampling of words runs into the thousands.