A post urges looking beyond “sloppy empirical patchwork” in AI

The author acknowledges that time may be too short for anything else—and that patchwork might even be enough—but says “we should still try.”