Google adds eyes to AI Mode with new visual search features (www.techradar.com)

Google has added a visual layer to AI Mode that lets you point or upload a photo and have the assistant interpret and act on it conversationally — for example “show me this style in lighter shades” or “retro 50s living room designs.” Technically, the feature uses a “visual search fan‑out” on top of AI Mode’s existing fan‑out answering strategy: an input image is decomposed into elements (objects, background, color, texture), multiple internal queries run in parallel against different retrieval models, and results are recombined to match inferred intent rather than just echo the original picture. The move is significant because it tightly fuses image understanding, conversational search, and commerce: Google ties visual results into its Shopping Graph (over 50 billion products, refreshed hourly) to surface prices, reviews, and local availability instantly. That gives Google an edge versus Pinterest Lens or Bing/Copilot visual search by combining scale, live product data, and natural-language followups. But it raises technical and UX risks — intent misreads, rank bias toward sponsored or well‑optimized imagery, and marginalization of sites lacking clean visual metadata — so the feature’s real impact will depend on retrieval relevance, ranking transparency, and how well Google suppresses noise and bias.