LLMs can be super useful if there is an authoritative source of truth. I wrote a Langchain app that takes my Python code, asks ChatGPT to optimize it then uses symbolic analysis to perform equivalency checking. I get to write and have clear simple python code, and then I offload optimization to a bot.
I’m the same way, I have only a few apps allowed to push to my Garmin, and it’s helpful to be able to archive or delete a useless email or know there’s something worth taking my phone out for. I find myself leaving my phone in other parts of the house is more focus-friendly since I’m not getting distracted while able to keep my eyes out for work-related items.
Do you listen to risky.biz?
I installed INCH on all my browsers, it’s obviously not 100% accurate, but it is nice to get a visual cue that the article you’re reading may very well be AI generated.
ZipPy is much less robust to defeat attempts than larger model-based detectors. Earlier I asked ChatGPT to write in the voice of a highschool student and it fooled the detectors. The web-UI let’s you add LLM-generated text in the style that you’re looking at to improve the accuracy of those types of content.
I don’t think we’ll ever be able to detect it reliably enough to fail students, if they co-write with a LLM.
The Intel ones are quite a bit easier, but still not as easy as a PC. You need to disable some FW security settings to allow for a non Apple kernel to boot.