Poetiq achieved superhuman performance on ARC-AGI-2
at ~$50/task using a mix of GPT-5.1 and Gemini 3 Pro.
Current models are powerful enough to reach AGI.
They just need the right agent scaffold.
https://poetiq.ai/posts/arcagi_announcement/
Google's new AI plays open world games and actually reasons about its environment, learns and adapts.
It uses the human user interface, not an API, so it sees what a human user would see.
In the words of the Google researchers, "this is a significant step towards AGI"
https://www.youtube.com/watch?v=f7QLiOaQqYI
The worst part? It's all true 🤦♂️
Most people are fucking lazy and stupid. Apparently, using their brains seems to cause them pain, so they will take any answer that seems to stem from an "authority".
This is not raising my esteem for humanity...