That is correct, pepole are forgetting where models learned to code, yes from us... If you see slop, it's a mirror, this is how practically pepole code
However how to code right, you only teach a model once and you are done
Try teaching generations of programmers the same
my argument is that all code is slop. slop is simply a slur and not a real objective definition. everyone considers code they wrote good and what others wrote that they don’t understand slop
it’s not a useful term for a true discussion on the topic. I’d maybe ask you to define what you mean when you say slop
The challenge for Google post-Gemini 4 will be what they do with a good model. Anthropic & OpenAI show we are moving towards a single interface for many tasks with orchestrator agents. The Gemini 3 era was full of fragmented products for many markets. Won't work for what's next.
@mattpocockuk@kunchenguid The real question is if we have data supporting that human processes are transferable to agents giving that cost generating these tokens are built upon intuition and not evidence...?
@ryanthawks@XFreeze I use it to orchestrate other agents, but I do so on Orca.
I also taught the engineer bot to use it and just describe to grok build which models to orchestrate and always invoke it with /goal
It works like charm and I spare alot of grok bot tokens while token maxxing the rest
@rohanpaul_ai All this horse power and grok is not a leading frontier model.
I am excited to see what they will he able to scale on this infrastructure, but I guess that raw compute is not enough.