Uploaded December 2017 | Updated September 2026, 2 weeks ago
Learn more about Udacity at: udacity.com/braincraft
Can artificial intelligence tell the difference between labradoodles and fried chicken? ...probably. But they can also see things that... aren't there.
ORDER BRAINCRAFT MERCH! 🧠store.dftba.com/collections/braincraft
SUBSCRIBE to BrainCraft! 👉 ow.ly/rt5IE
More BrainCraft: Artificial Intelligence Solved This Audio Illusion. Can You? youtu.be/Ucc1Xs5UAj4
My Twitter twitter.com/nessyhill | Instagram instagram.com/nessyhill
Artificial neural networks are pretty good at image recognition – but sometimes machines misidentify objects too. And, sometimes, machines see things that... aren't there. This can have pretty big implications, which (believe or not) will have more impact that getting a labradoodle and a piece of chicken mixed up.
BrainCraft was created by Vanessa Hill (@nessyhill) and is brought to you by PBS Digital Studios. Talking psychology, neuroscience & why we act the way we do.
This episode was written by Bahar Gholipour and filmed by Dominique Taylor. Produced/edited/hosted by Vanessa. Co-hosted by Luna.
REFERENCES 📚
3D Adversarial examples: http://www.csail.mit.edu/fooling_neural_networks_with_3Dprinted_objects
labsix.org/physical-objects-that-fool-neural-nets
MIT CSAIL: youtube.com/watch?v=piYnd_wYlT8
OpenAI: youtube.com/watch?v=oeQW5qdeyy8
Dog in a hat inspired by:
google.com/intl/en-GB/about/main/machine-learning-qa
Testing AI with dogs/foods:
blog.cloudsight.ai/chihuahua-or-muffin-1bdf02ec1680
crimesagainstai.com/2016/04/19/animals-vs-food-vs-a-i
Original dog/food photos:
twitter.com/teenybiscuit/status/707727863571582978?lang=en
Learn more about Udacity at: udacity.com/braincraft
Can artificial intelligence tell the difference between labradoodles and fried chicken? ...probably. But they can also see things that... aren't there.
ORDER BRAINCRAFT MERCH! 🧠store.dftba.com/collections/braincraft
SUBSCRIBE to BrainCraft! 👉 ow.ly/rt5IE
More BrainCraft: Artificial Intelligence Solved This Audio Illusion. Can You? youtu.be/Ucc1Xs5UAj4
My Twitter twitter.com/nessyhill | Instagram instagram.com/nessyhill
Artificial neural networks are pretty good at image recognition – but sometimes machines misidentify objects too. And, sometimes, machines see things that... aren't there. This can have pretty big implications, which (believe or not) will have more impact that getting a labradoodle and a piece of chicken mixed up.
BrainCraft was created by Vanessa Hill (@nessyhill) and is brought to you by PBS Digital Studios. Talking psychology, neuroscience & why we act the way we do.
This episode was written by Bahar Gholipour and filmed by Dominique Taylor. Produced/edited/hosted by Vanessa. Co-hosted by Luna.
REFERENCES 📚
3D Adversarial examples: http://www.csail.mit.edu/fooling_neural_networks_with_3Dprinted_objects
labsix.org/physical-objects-that-fool-neural-nets
MIT CSAIL: youtube.com/watch?v=piYnd_wYlT8
OpenAI: youtube.com/watch?v=oeQW5qdeyy8
Dog in a hat inspired by:
google.com/intl/en-GB/about/main/machine-learning-qa
Testing AI with dogs/foods:
blog.cloudsight.ai/chihuahua-or-muffin-1bdf02ec1680
crimesagainstai.com/2016/04/19/animals-vs-food-vs-a-i
Original dog/food photos:
twitter.com/teenybiscuit/status/707727863571582978?lang=en
