I'm becoming more suspicious that there is a chasm between what investors think ML can do, and what it can do. These trends never end particularly well, but by

I'm becoming more suspicious that there is a chasm between what investors think ML can do, and what it can do. These trends never end particularly well, but by the same token - it's hard to tease out the real wins from the noise. Part of this is probably the software lifecycle. Everyone wants AGI, but AGI is always further than we think.