Outsourcing the Doing: What an AI Headline Taught Me About Lazy Eye

I read a BBC piece this week: graduates are managing AI — but are they learning much else? And I sat with it longer than I expected. The worry isn't that the tools are bad. It's that outsourcing the doing quietly outsources the developing — the reps that build judgment, attention, coordination.

Vision has the same failure mode. If one eye does all the work, the other doesn't get stronger by waiting. It gets weaker. That's essentially the mechanism behind amblyopia: the brain picks a favourite and the weaker eye stops getting practice.

Which is why the design problem behind Amblyotube (Seven Sports, Meta Quest) interests me. It turns ordinary YouTube-style video into a vision gym by exploiting the one thing VR hardware gives you for free: two independent displays. Separate visual feeds per eye, running at the same time.

The engineering choices are what make it more than a gimmick:

• Dominant Eye Shader — adjustable blur, contrast, brightness, gamma, or opacity applied to the stronger eye. At 100% it's full digital patching, but the default intent is partial transparency so both eyes keep processing.
• Lazy Eye Sharpener (MFBF) — human-figure detection drives selective sharpening and visual accents (red outline, yellow-green highlight, adjustable Pulse Hz) only in the weaker eye's feed.
• Magenta Focus Cue — high-visibility magenta in one eye, subtle grey in the other. The brain fuses them into one 3D image. That's binocular fusion being practised, not described.
• Wedge Occlusion — blacks out complementary regions of each field to remove binocular rivalry instead of fighting it.

A built-in LogMAR check (Landolt C optotypes at a virtual 3m, clinical staircase) can even re-tune the therapy profile automatically. 30–40 minutes a day, content you already wanted to watch. No patches, no anaglyph glasses.

If you're an optometrist, vision therapist, or a builder thinking about assistive tech — worth a look.

https://www.meta.com/en-gb/experiences/amblyotube/25906906972338493/

Supplementary vision training / assistive tool for binocular coordination and depth perception. Not a diagnosis, treatment, or cure. Ages 13+. 30–40 min/day recommended, max 1 hour.