These are the Pentagon’s new ethics “principles” for AI in warfare2 min read

Foster Kamer / 25 Feb 2020

On Monday, the Pentagon announced the official adoption of a series of new principles for ethical use of artificial intelligence in warfare, the Associated Press reports.

The principles were formed out of a commission with the (darkly Newspeak-y) name the Defense Innovation Board, which released its recommendations (title: “AI Principles: Recommendations on the Ethical Use of Artificial Intelligence by the Department of Defense“) to the Pentagon last October.

The board was fronted by former Google CEO Eric Schmidt, an interesting twist given (as pointed out by the AP) due to the way Google seemed to (or: pretended to) drop out of a defense department project involving A.I. in 2018 after internal protests from Google staffers (to say nothing of the way Google’s involvement was handled by the Pentagon).

“Responsible”

Per the late 2019 report, the principles are (with our paraphrasing in parenthesis):

1. Responsible (“human beings” at the D.O.D. need to exercise judgement in using A.I. in warfare)

2. Equitable (the D.O.D. “should take deliberate steps” to avoid bias in A.I. recognition)

3. Traceable (the D.O.D. should show their homework and understand how the technology works)

4. Reliable  (A.I. systems should have a “well-defined domain of use” and also, work well)

5. Governable (A.I. systems need “the ability to detect and avoid unintended harm or disruption” a.k.a. not go Skynet on us).

If this all sounds broad, harmless, ineffectual, myopic, painfully obvious, and toothless, well… “Principles”

The Next Web called the principles “hazy” and “toothless.” And Dave Gershgorn of OneZero noted that these supposed ethics are missing “‘don’t kill somebody with a robot.’”