Intelligence is the ability to learn or understand or to deal with new or trying situations. Merriam-Webster.

Me: OK Google, call Richard
Google Assistant: Richard Home or Richard Mobile?
Me: Home
Google Assistant: Home is in Portola Valley California

While driving somewhere, Michele and I will sometimes entertain ourselves by comparing the abilities of Google Assistant and Siri. It is fascinating what they can do and even more fascinating what they can’t do. Despite the above example, in almost all cases, OK Google understands better, and is more accurate and useful than Hey Siri.

I’ve read that Google is able to get better engineers because Google gives public credit to its engineers while Apple hides them behind the Apple Brand. For example, at Apple, a paper entitled An On-device Deep Neural Network for Face Detection is listed as written by the Computer Vision Machine Learning Team while, at Google, a paper entitled TFX: A TensorFlow-Based Production-Scale Machine Learning Platform – I have no idea what either title actually means, by the way – credits Akshay Naresh Modi, Chiu Yuen Koo, Chuan Yu Foo, Clemens Mewald, Denis M. Baylor, Eric Breck, Heng-Tze Cheng, Jarek Wilkiewicz, Levent Koc, Lukasz Lew, Martin A. Zinkevich, Martin Wicke, Mustafa Ispir, Neoklis Polyzotis, Noah Fiedel, Salem Elie Haykal, Steven Whang, Sudip Roy, Sukriti Ramesh, Vihan Jain, Xin Zhang, and Zakaria Haque. 

As an aside, when looking at the Google team, they are surprisingly diverse but only in strangely limited areas. The Google Team, which I picked more or less at random, is not a microcosm of America. When I go to the alphabetical Google People Directory, there is a large percentage of Indian names as well as Chinese, Pakistani, Iranian, and Arab names along with the expected European names. Even the CEO of Google, Sundar Pichaiand, was born in India. End aside. 

As another aside, Looking at that list of names above, which has so few women on it, I wonder if the fact that there are so few women in the High Tech biz is because Silicon Valley is populated by a high percentage of men from conservative cultures and they bring an unconscience bias against women. That said, I read that Google just hired two women to head all their Cloud Machine Learning technology; according to the Silicon Valley Business Journal, they are “Fei-Fei Li, who was director of AI at Stanford, and Jia Li, who was head of research at Snapchat’s parent company, Snap Inc”, so maybe, the times are a-changin’. End aside. 

Reading about AI, I realize that, when I think of Intelligence, I am really thinking of being conscious. Intellectually, I know that intelligence is not the same as consciousness, at least from the limited things I’ve read, but I find it very difficult to imagine intelligence without consciousness. It does make me feel better that I am not the only one making this mistake; certainly, HAL was conscience and Skynet, Ava in Em Machina for sure. When we drove through Georgia last fall, we asked Google Maps – with lots of double checking with Apple Maps – to plot a course diagonally across rural Georgia while staying off of highways. It did a super job but, in my imagination, at least, it is just a brute force problem, not Artificial Intelligence. Memorize every road in Georgia with its length, from intersection to intersection, crank in the speed limits of each section – speed limits which even my humble Hyundai knows – and calculate the fasted route. We did go by the world’s largest Peanut Monument and Google did take us through the very cute town of Colquitt, pop. 1,939, where I was able to buy an excellent cappuccino, but I think these were artifacts of the route and Google did not think: Oh, I’ll take him by the Peanut Monument and through Colquitt because it is the only place in rural Georgia he can get a cappuccino.  

On the other hand, when Google DeepMind taught itself to play Go and then beat Go master Lee Sedol in four out of five games, there was a lot more than brute force going on. A disclaimer here, I know nothing about Go, except that I have read that it is much more complicated than chess. Like chess, every move opens the door to many more moves, except in Go, the number of moves starts to approach the infinite. I read that there are more possible moves than all the atoms in the universe, so it is impossible for DeepMind to memorize all the moves, the program had to actually think about what move would be best. One of the things I found most amazing while reading about DeepMind vs. Sedol is that DeepMind made many unconventional moves that nobody had ever seen or thought of before and Sedol says that he learned to think out of the box more by playing DeepMind.  

Years ago, when I was an Industrial Psych major, the thinking was that the Learning Curve was not a smooth curve but a curve broken up by a series of plateaus. Sort of analogous to the evolutionary theory of Punctuated Equilibrium in biology. Change happens, we settle into the new reality, and then we have a jump to another new reality. That way of looking at learning is not as popular as it was but, after looking at the recent Google demo, it seems to me that is what is happening in Machine Learning and Artificial Intelligence (I put both down because I’m not sure of the difference). We are making a jump to a new reality.