Grok 4 feels like Artificial General Intelligence to me. It is clearly not just constructing statistically likely connections, but is drawing fairly deep insights on problems it hasn’t seen before, in ways I haven’t seen elsewhere. Here’s an example:

The biggest shortcoming I see is the adoption of confused musings from online forums as facts, coupled with the inability to derive deep insights from intermingled prose and graphics in sources. It needs contextual skepticism and more multimodal visual learning.

Also, I bet there are 100,000’s of topics where a human professional could write a definitive guidebook on topics that online forum posters frequently confuse, and that could be used to fine-tune a lot of the low quality user content driven nonsense out of current AI models.