With the help of Google, the U.S. military is utilizing Artificial Intelligence with a military-based drone image recognition program under Project Maven–the military’s Algorithmic Warfare Cross Functional Team (AWCFT).
Google has reportedly given Tensorflow programming kits to the Department of Defense (DOD) in order to process the drone images quickly and accurately. The drone images are flagged for human review and are for non-offensive purposes. This will help the military buy more time for getting information, and sorting everything out, while also reducing the amount of mistakes that could be made as well.
According to Google Cloud head scientists Scott Frohman, Aileen Black and Dr. Fei-Fei Li, the project has a total budget of 250 M per year. This project is also a part of an attempt to transition the DOD over to a cloud-based program.
This project, however, while relieving some, has surprised others. Some staff at Google were shocked to find out about the program while others reasoned that there should at least be a baseline of ethical guidelines drawn into machine learning. They’re also not quite clear if Google will just help out in the basic setup or in the actual development.
Project Maven launched in just 6 months time with its first data use aimed at hunting down ISIS. The vast amount of the video surveillance footages retrieved from ISIS attacks have overwhelmed the analytical team, which helped propel the project forward.
On the brighter side, gathering this amount of data minimizes administrative tasks like grouping info in spreadsheets. And since data is automated, the military is left with more time for analyzing the information, and preparing the next steps. The project will also focus the Processing, Exploitation, and Dissemination (PED) capabilities.
A total of 90 days are set to complete the first phase of the project which is to establish algorithms with their own equipment and construction or to amass algorithms from a private organization. The second phase and third phase, both of which involve hardware procuring and implementation of algorithms also have a timeline of 90 days each to complete the entire system.
Consequently, this project has turned some employees against the implementation and more than 4,000 Google staffers have signed a petition to terminate the agreement.
Despite this, Air Force Lt. Gen. John N.T. “Jack” Shanahan, who is the director for defense intelligence for warfighter support sees a potential for the project to be used in other disciplines including sensor fusion, logistics, communications and command and control that benefit DOD as a whole.
Google CEO Diane Greene did announce that Project Maven will not renew its contract with DOD next year because of the amount backlash that the company received.
Google also made a statement that their work will not be serviced for the creation of weapons.
On the other hand, Google has invented a Duplex AI through Google Assistant that is practically capable of doing anything an authorized individual commands. Google Assistant is found in almost all devices including our smartphones making it a powerful feature that could possibly be abused if the technology is not used for good.
There are also some reports of AI surpassing doctors in diagnosing skin cancer. Scientists have discovered early in January of this year that AI was a good and reliable tool in identifying skin cancer through a research conducted with 58 dermatologists and a Convolutional Neural Network (CNN) wherein the dermatologists detected 86.6 skin cancers while 95 percent were found by the CNN.
So be it through drones and the military, or robots and doctors, and whether you’re ready or not, AI tech, programs and implementation are on the rise.