A RENEWED INTEREST IN AI
Ever since the US fell behind China in the AI arms race the country has been criticized for not being innovative enough and not focusing on the things that matter. As we outline in this previous post, there are a multitude of reasons why China pulled ahead of the rest of the world. Despite the fact that this is clearly a multi faceted issue the US has neglected to focus on the development of new technologies which includes AI. The DoD has seemingly decided to try and address this.
Other than the president’s recent executive order which lacked any kind of concrete plan or details describing an AI development strategy the US has not released anything that touches this topic. Until now. Shortly after the vague executive order the Department of Defense (DoD) released a 17 page document outlining their AI strategy that supposedly began in 2018. This document focuses on 5 key steps which will shape their plan going forward.
– Delivering AI-enabled capabilities that address key missions.
– Scaling AI’s impact across DoD through a common foundation that enables decentralized development and experimentation.
– Cultivating a leading AI workforce.
– Engaging with commercial, academic, and international allies and partners.
– Leading in military ethics and AI safety.
We will analyze these points and discuss what this means for the US going forward.
DELIVERING AI-ENABLED CAPABILITIES THAT ADDRESS KEY MISSIONS
The first bullet point in the strategic approach section of this document talks about launching initiatives intended to deploy AI quickly and efficiently across a wide variety of areas. It also states that they intend to use AI to directly augment the efforts of their workers, taking over tedious tasks in order to streamline the whole process. They also mention improving the safety of operating equipment.
This first point is another way of saying that believe they must use AI to automate low end or tedious tasks. It is widely believed that this technology will augment the efforts of workers all across the globe. The DoD in particular, can’t afford to fall behind in this regard.
Also, the part that mentions using AI to improve decision making foreshadows the future use of autonomous drone strikes or missile launches. It would be in their best interest to nail down all the ethical problems beforehand, otherwise, there will be some serious problems. Obviously the paper was referring to human decision making as opposed to autonomous drones make no mistake that the DoD is well on their way to deploying AI as a standalone weapon of mass destruction.
SCALING AI’S IMPACT ACROSS THE DoD
This seems fairly straightforward. The idea is that the DoD would very much like to put in place certain platforms that provide equal access to the power of AI across the entire department. They intend to create a common foundation of shared data that includes but not limited to cloud services, shared data, and reusable tools and frameworks.
The document points out that one of the core strengths of the US military is how innovative it has been. They believe that most of their advances regarding utilizing AI technology will come from people out in the field who are deploying it as opposed to far off environments such as laboratories. Approaches like this one are being seen in many tech companies across the world so it is no surprise that the DoD talks about this being a part of their long term strategy.
CULTIVATING A LEADING AI WORKFORCE
This next point highlights the fact that AI is changing the workplace at a rapid pace and changes will be needed to deal with these changes. They state that they intend to seek out the best talent they can find in the industry and encourage rapid experimentation to try and speed up the rate of innovation.
Statements like this are nothing new. Every entity that intends on using AI powered tech has to try and seek out the most talented programmers possible. Since the field is still in its early stages there is a limited number of machine learning and deep learning practitioners which makes the recruitment step that much more important. Creating an environment in which experimentation is not only possible but encouraged is also of utmost importance to the DoD considering how far behind the US has fallen lately.
ENGAGING WITH COMMERCIAL, ACADEMIC, AND INTERNATIONAL ALLIES AND PARTNERS
No one entity has gotten all that far with AI without significant collaboration from other entities. The documents states that the DoD intends to work closely with academia in order to make appropriate advancements in an efficient time frame. They also intend to increase funding to provide their researchers with what they need for long term research into the field of Artificial Intelligence.
They also state that they intend to make contributions to the open-sourced global AI community. Most of the statements they made in this section are things that they have to say in order to put forward a certain kind of public image. This section is overall very generic and lacks substance. The details of how they actually intend to engage with allies and partners would be a nice addition but the reasons for the lack of detail are obvious. This is the US DoD after all.
LEADING IN MILITARY ETHICS AND AI SAFETY
This section states that the DoD intends to clearly communicate its vision for a future in which certain guiding principles steer the use of AI in a safe direction. Now, despite the fact that the DoD will never clearly communicate their vision doesn’t change the fact that surely someone over there has this vision. That’s whats important.
They also reiterate their desire to invest in more research into the field in order to figure out how to deploy this technology in a ethical and moral friendly way. There is also a statement in there that reads
“We will also seek opportunities to use AI to reduce unintentional harm and collateral damage via increased situational awareness and enhanced decision support.” They already mentioned decision making in the first bullet point under the strategic approach section. While they are still not yet referring to something as dangerous as autonomous missiles they are definitely hinting at the possibility.
All of this is interesting considering the fact that the US military is not exactly known for conducting its practices in the most ethical or moral friendly way. It makes one wonder why the deployment of AI would be any different. However, we will leave such statements for the diligent reader to ponder.
There is a lot more that can be said about this 17 page document and we encourage everyone to check it out for themselves using the link in the second paragraph. It is a decent read and provides some insight into how the US intends to move forward from here. Much of the Dod’s work is classified so obviously one must take documents like this with a grain of salt, but still, there is much to learn here. There is also quite a bit to look forward too as the development of AI marches on showing no signs of letting up.