Comment by stevebmark

After using frontier models it’s hard to understand why anyone would think this is the path to AGI. Self improving models will likely have limited ability and returns. There may be breakthroughs that enable more general self improvement but the current state of frontier models isn’t that.