LLMs are no more than pen and paper at this point. Especially for those who aren't trying to create slop. We all would want our pens to accurately reflect the strokes (well in this case thoughts) rather than adding tiny watermarks to identify that it is generated by a particular pen or a user.
Watermarking per model is just the start. The method is cheap enough to distinguish individual users.
That is an insane statement, LLMs generate swaths of text from almost nothing.
If they are adding so little value as to be as transparent as a pen and paper then why use one at all? Transcription doesn't need an LLM so that's not what you're taking about I assume.
LLMs do a whole lot more than writing your thoughts down. They write extra text. If you just want a pen and paper, use Notepad. Or better, a pen and paper
The term "powerpoint" has been genericized in the common vernacular to mean any electronic slide show, like kleenex for tissues and sellotape/scotch tape for transparent adhesive tape.
It might not have reached legal generic status but just about any working professional will understand what you mean if you refer to "making a powerpoint deck" even if you don't use the actual Microslop product.
This super interesting link showing things that were impossible even a couple of years back, sitting below the day old news about Kagi's me too browser tells everything about HN audiences. West will fall behind in robotics and mostly because of the wrong priorities of its VCs and developers.
I cannot provide the session ids but I have tried the above flag and can confirm this makes a huge amount of difference. You should treat this as bug and make this as the default behavior. Clearly the adaptive thinking is making the model plain stupid and useless. It is time you guys take this seriously and stop messing with the performance with every damn release.
This connects some many sparse dots on the map for me. Finishing it right now, thank you for commenting w the link. What a perspective, and well-written parable to communicate it.
Another anecdote/datapoint. Same experience. It seem to mask a lot of bad model issues by not talking much and overthinking stuff. The experience turns sour the more one works with it.
And yes +1 for opus. Anthropic delivered a winner after fucking up the previous opus 4.1 release.
Watermarking per model is just the start. The method is cheap enough to distinguish individual users.