Some work seems to be what we’ve done before. Building efficient infrastructure. Collecting a lot of data, then a lot more. Running ablation after ablation to understand what the model should actually look like. Debugging software and hardware, again and again. The unglamorous
AI Research's Unglamorous Core: Infrastructure, Data, Ablations
A practitioner reflects that much AI work feels familiar: building efficient infrastructure, collecting ever more data, running ablation after ablation to understand models, and endlessly debugging software and hardware.