Great news everyone! I finally talk about AI hype. Someone finally mentioned LLMs one time too many, and the reckoning is upon us:
https://ludic.mataroa.blog/blog/i-will-fucking-piledrive-you-if-you-mention-ai-again/
Great news everyone! I finally talk about AI hype. Someone finally mentioned LLMs one time too many, and the reckoning is upon us:
https://ludic.mataroa.blog/blog/i-will-fucking-piledrive-you-if-you-mention-ai-again/
@ludicity This is great, and perhaps a good example of the #WriteRisky style espoused by @phillipdewet? Ironically as a way of setting oneself apart as a human writer rather than an LLM android. Can't imagine an LLM getting the tone right whilst threatening violence to mouthy mid level executives. Quick way to getting the plug pulled.
@jeztastic @phillipdewet Huh, I've never heard of this before, but I just got off the phone with one of my co-directors and we were talking about how I can pull the "build an audience" lever as many times as I want, even with fresh starts. Authenticity is at a premium. Just from the name, I'm guessing Phillip is talking about some game-theoretic/soulful stuff that is absolutely my vibe.
@ludicity @jeztastic Oh, no. I'm just a guy, standing in front of people in the business world, telling them to write better – or the AI bogeyman will get 'em.
This is the true gift of the LLMs: fear. In this case fear of sounding like a machine (which they did before, but they got away with it).
@phillipdewet @jeztastic Hah, fair. Watching people in business self-own as they talk about how much better their comms have gotten since ChatGPT came out it very painful. It's really good for writing things that don't matter but need to appear like you spent time on them, but outside of that it's like, hard pass.
@ludicity @jeztastic I am a not-so-secret fan of using LLMs as language calculators, mostly for phrase synonym search.
They really can make writing better - if used right. As in, not in the way their creators advocate at all.
@phillipdewet Yeah, they can make writing better if you already have developed the quality of judgement. People assume that I code as well as I write, but my judgement in matters of writing far outstrips my judgements around system design.
LLMs mostly bring you to some boring, corporate local maximum. Although I have to admit that I can't tell if that's a property of current technology or the system level prompts they're given by their producers.