Overview: Teaching AI to Localize Personalized Objects Vision-language models (VLMs) like GPT-5 have made strides in recognizing general objects in complex scenes. However, researchers at MIT and the MIT-IBM Watson AI Lab identified a critical gap: these models struggle to locate personalized objects across time and varied contexts, such as a specific pet or a…