Monthly Archives: January 2019
In July I wrote a post “NASA has Launched a sort of HAL-like Robot to help an Astronaut on the ISS” about the cool new floating robot the European Space Agency (ESA) sent to the ISS. Its name is CIMON (pronounced Sigh Mon) and it was fun to watch and speculate about how it would help astronauts in the future.
It has now been six months and there has been a potentially disconcerting change as CIMON matures. According to the Quartz publication’s Nicholas Rivero, CIMON has gotten a little testy for what is supposed to be a friendly helping assistant. Astronaut Alexander Gerst asked CIMON to play some of Gerst’s favorite music. Instead CIMON played a song called “The Man-Machine” which annoyed Gerst and he asked the robot to please stop playing.
Project Maven is an Artificial Intelligence system that assists human operators in finding objects of interest to the military. It is based on Machine Learning algorithms from very large databases of information. It is what the Department of Defense calls a “Pathfinder” meaning that is the beginning of a significant direction that the Pentagon has determined to go.
An example of a Project Maven user is the Special Operations Command (SOCOM). There are thousands of hours of video and pictures stored of information for their zones of operation. Commandos need specific information about their operations they can greatly be enhanced by the use of Artificial Intelligence. The Project Maven machine learning algorithm can be used to spot a specific kind of anomaly a huge database of information. Much of that requires visual recognition.
In September 2018 the Defense Advanced Research Projects Agency announced it will invest up to $2 billion over the next five years. That announcement was made by DARPA director Stephen Walker saying that the agency was to explore “how machines can acquire humanlike communication and reasoning capabilities.” I first became familiar with DARPA in 1980 when it was known as ARPA when I was a regional manager for Research Inc. managing the Eastern United States for its computer terminal sales to universities and businesses. I became most familiar with ARPA when I became the North Central sales representative for GTE Telenet selling their X.25 product from Bolt Beraneck and Newman which was directly active with ARPA in its development. That was the direct forerunner to the Internet.
DARPA has said that the $2 billion with fund projects over and above the 20 AI projects the Pentagon has going now. (Google, the primary contractor for Project Maven has said they are pulling out of the program. Thousands of Google employees signed the petition for Google to get out of the project and top management caved in to that demand.) According to Peter Highnam, the deputy director of DARPA, “hundreds if not thousands” of schools and companies have bid “aggressively” on these programs.
There is a good chance that Microsoft and Nvidia, and maybe Amazon, will go after this business. Google’s AI software – Tensorflow – is being used by the US military drone program. Fortunately for everyone, Google made Tensorflow open source.
As of December 2018 the DOD who oversaw Project Maven leads its new Joint Artificial Intelligence Center (JAIC) which will coordinate AI research across the Department, government labs, and private companies. The goal of this is to speed up the process of getting AI tools to the war fighters.
There is a sense of urgency to this. China does not operate with the same legal constraints that the US has, and they can mandate that companies such as Baidu, Alibaba, and Tencent must comply with the Communist Party’s wishes. We also don’t know how much technology has been transferred to the Chinese by American companies doing business in China. Russia is not a leader in Artificial Intelligence but it has a large base of students could be conscripted to train fairly fast.