Well, Actually: Bryson DeChambeau's Swing Gets the Full Computational Treatment
Bryson DeChambeau, that physicist masquerading as a golfer, has partnered with Google Cloud to apply deep learning to his swing. The system employs proprietary 2D and 3D models to track over 30 biomechanical points across body, club, and ball. It segments the golf shot into discrete phases: top of swing, impact, follow-through, and finish. Data is generated. Presumably someone reads it.
This illustrates the principle of temporal decomposition: breaking a continuous physical action into analyzable segments rather than treating it as an undifferentiated whole. You can apply this to any repetitive skill in your own workflow. Record yourself performing a task. Identify the phases. Look for where efficiency breaks down. The AI merely automates what a careful human observer could do, albeit more slowly and with less precision.
Bryson DeChambeau, professional golfer and noted eccentric, in collaboration with Google Cloud. The system is proprietary to this partnership; no public performance metrics beyond the technical architecture are disclosed in the source material.
Step 1: Open your phone's slow-motion camera and record yourself performing a repetitive task you want to improve: typing, a presentation gesture, a cooking technique. Step 2: Play back the video at quarter speed and manually mark three to five distinct phases of the action with timestamp notes. Step 3: Identify one phase where your movement deviates from what you consider ideal, and practice only that isolated segment for ten minutes before re-recording to compare.