Slow AI

The Cunning Plan Why Paying the Piper Bubbles and Pin Pricks A Certain Musky Odor The Point At the risk of constantly returning to the subject of the lack of use from large language models—and I hope to keep this one short—early in June, I started seeing a strange shift in sentiment. After years of screaming at us to “get on board or get left behind,” and after the massive backlash especially from software developers, a bunch of people—half a dozen blog posts that made it to my screen, as somebody who doesn’t seek this nonsense out at all—took it upon themselves to inform us that we’ve all gotten it (and them) wrong. In reality, they now tell us, we have all overlooked the best way to do our work with LLMs. The Cunning Plan Despite the breathless excitement (and aforementioned screaming) about networks of LLMs talking to each other—the alleged “improvements” that the companies keep trying to sell—while pulling information from the Internet and scrubbing through your files, we should…