Tuesday, 9 March 2021

OpenAI discovers its visual AI can be fooled by written notes

Image of AI mistakenly recognizing an Apple as an Ipod, because the word "ipod" was written on a piece of paper and taped to the apple

OpenAI has just published a fascinating piece pointing out the strengths of its top-of-the-line vision system — as well as one big weakness: You can fool it just by slapping a text label on an item.

Write "iPod" on an apple and presto — the vision system thinks it's an iPod. — Read the rest



Share:
Powered by Blogger.