Google has been on a bit of a buying spree lately, purchasing robotics companies and others. Sometimes a large company like Google wants to buy a company for its products but often they just want the patents and people. A recent purchase of an AI company called DeepMind seems to fall into the latter category.
What caught my attention is this line in the article. (H/T Singularity Hub)
DeepMind also filed a patent for a method that could improve Google’s image search results by allowing the user to use visual descriptions rather than keywords.
Oooooh that’s pretty cool! I have been talking about for a long time about how when computers can understand what they are looking at in images and video, it will allow all that vast amount of content to become searchable. Right now we have to manually tag photos with keywords (we can tag them with faces automatically too) but this will do so much more.
I think a big way we will learn in the future will be with adaptive learning computer systems. These will constantly adapt to our interests and will dynamically create lessons as needed, any time, any place. Using technology like this image search capability will allow these AI tutors to rapidly access an ocean of image and video content and insert it into a lesson.
This is all made possible by the rapid growth in processing power of modern computers so it’s likely to become better as time goes on.