Rebooting AI
Building Artificial Intelligence We Can Trust


Gary Marcus and Ernest Davis

Color Images

The page displays color versions of the images included or described in Rebooting AI. These are reproduced here with permission of the copyright holders.
"A plausible caption, generated automatically by AI" p. 22

From Vinyals, Oriol, Alexander Toshev, Samy Bengio, and Dumitru Erhan (2015), "Show and tell: A neural image caption generator."
"A less plausible caption, generated by the same system" p. 23

From Vinyals, Oriol, Alexander Toshev, Samy Bengio, and Dumitru Erhan (2015), "Show and tell: A neural image caption generator."
The photo of two smiling women chatting on cell phones described on p. 58 on the text cannot be included here for reasons of copyright but it can be found at Rohrbach, Anna, Lisa Anne Hendricks, Kaylee Burns, Trevor Darrell, and Kate Saenko (2018) "Object hallucination in image captioning."
"A turtle, misidentified by deep learning as a rifle," p. 58

From Athalye, Anish, Logan Engstrom, Andrew Ilyas, and Kevin Kwok (2018) "Synthesizing robust adversarial examples."
"A baseball with foam, misidentified as an espresso," p. 59

From Athalye, Anish, Logan Engstrom, Andrew Ilyas, and Kevin Kwok (2018) "Synthesizing robust adversarial examples."
The picture of the piggy bank misidentified as a tiger cat cannot be included here for copyright reasons, but it can be found at Karmon, Danny, Daniel Zoran, and Yoav Goldberg (2018) "LaVAN: Localized and Visible Adversarial Noise".
"A psychedelic toaster," p. 59

From Brown, Tom B., Dandelion Mané, Aurko Roy, Martín Abadi, and Justin Gilmer (2017), "Adversarial patch"
"A banana correctly identified by deep learning," p. 59

From Brown, Tom B., Dandelion Mané, Aurko Roy, Martín Abadi, and Justin Gilmer (2017), "Adversarial patch"
"A banana with sticker, misidentified as a toaster" p. 59

From Brown, Tom B., Dandelion Mané, Aurko Roy, Martín Abadi, and Justin Gilmer (2017), "Adversarial patch"
"An altered stop sign, misidentified as a speed limit sign" p. 60

From Evtimov, Ivan, Kevin Eykholt, Earlence Fernandes, Tadayoshi Kohno, Bo Li, Atul Prakash, Amir Rahmati, and Dawn Song (2017), "Robust physical-world attacks on machine learning models."
"A schoolbus in an unusual position, misidentified as a snowplow" p. 60

From Alcorn, Michael A., Qi Li, Zhitao Gong, Chengfei Wang, Long Mai, Wei-Shinn Ku, and Anh Nguyen (2018), "Strike (with) a pose: Neural networks are easily fooled by strange poses of familiar objects."
"Robot falling backward, trying to open a door" p. 96

From IEEE Spectrum A Compilation of Robots Falling Down at the DARPA Robotics Challenge
From the text p. 134, "... a picture of a lake containing a ripple that was shaped vaguely like a car, enough so that a vision system was fooled."

From Vondrick, Carl, Aditya Khosla, Tomasz Malisiewicz, and Antonio Torralba (2012) "Inverting and visualizing features for object detection."
Julia Child's kitchen, and details. p. 135

From Wikimedia: Julia Child's Kitchen by Matthew Bisanz
"Sawing the wrong side of a tree limb." p. 148

Copyright: Communications of the ACM.
"Ordinary objects that pose challenges for AI," p. 165

From Wikimedia, Cheese Grater and String Bag