

I’ve seen this one a few times and it’s puzzled me a bit for a while now, I don’t think anyone in these companies think iterating on an LLM model alone is going to give them AGI
You only have to look at how Claude Code has taken off in the past 6 months. Sure the model is a big factor, but it’s the tooling built on top of it that makes it useful and disruptive. The model is what enabled the utility, when attached to conventionally engineered tooling.
Whenever the first AGI is created, an LLM will 100% be part of the implementation, it’s just more than a one piece puzzle





Just install the Consent-o-matic extension and you get the benefit of rejecting without the faff 90% of the time