Robots will identify targets better than humans

13 Dec

terminatorA couple of weeks ago, I wrote about how the case for laws governing armed robots is growing. The post was in response to a report issued by Human Rights Watch, which advocated for an outright ban on autonomous armed machines before they become a reality.

I argued that an outright ban was impractical and perhaps undesirable, with different countries having different needs. Some very much want to take human soldiers out of harm’s way while others might be looking at armed robots as a way of evening the odds against vastly bigger enemies.

One of the things that stood out for me in regards to the report were the claims that robots won’t be able to differentiate between an armed soldier and an innocent girl holding an ice cream cone. I suggested that the technology to do exactly that is developing quickly, so it won’t be long before it’s actually put to use.

A few commentators over at the Lawfare blog, however, took it a step further by suggesting that not only will robots be able to identify hostile targets over innocents, they’ll actually do it better than humans. As Benjamin Wittes, a senior fellow in governance studies at the Brookings Institution, wrote:

The report’s authors make a lot of assumptions about the technology that may well prove wrong. Most importantly, they reject without much examination the possibility that fully autonomous robots might, in some environments and for some situations, distinguish military targets far better and more accurately than humans can. To call for a per se ban on autonomous weapons is to insist as a matter of [international humanitarian law] on preserving a minimum level of human error in targeting. That is defensible only if one is certain that the baseline level of possible robotic error in civilian protection exceeds that baseline level of human error… I, for one, would not bet against the possibility that for some military applications, we will some day come to see mere human judgment as guaranteeing an unacceptable level of indiscriminate and disproportionate violence.

It is indeed wrong to assume that robots won’t be able to do a certain task better than humans simply because they can’t right now. The list of things that humans can do better than robots is literally shrinking every day, with cutting noodles being the latest addition. The fact is perhaps best illustrated by the cartoon below:



Posted by on December 13, 2012 in robots, war


8 responses to “Robots will identify targets better than humans

  1. Marc Venot

    December 13, 2012 at 12:59 am

    Why then the Canadian government has, at least until recently, preferred to buy for its Royal Canadian Air Force the Lockheed Martin F-35 Lightning II, instead of drones?

  2. Justin Amirkhani

    December 13, 2012 at 1:25 am

    Robots can’t even identify speech better than humans. To hell with this tomfoolery!

  3. Torontoworker

    December 13, 2012 at 9:30 am

    The authors (Brookings) of such dribble about technology being better then humans in deciding who lives and who dies, (which is what we are really saying here) use words such as ‘may’ and ‘possibility’ when describing the future accuracy of such machines or technology being ‘better’ then human reaction skills to threat determination. The facts are we would be allowing machines to kill human beings without human intervention and ‘may’ (be better) just doesn’t cut it. We all remember the sales job the public were given over the facial recognition camera’s that the police industry and police equipment hardware industry wanted the taxpayers to pay for, (and pay dearly for I might add). These systems were tested at last years Super Bowl and the results we’re pitiful. In fact 9 out of 10 test subjects with pre-programmed photo’s of their faces loaded were missed by the system due to disguises being used. Crazy people without criminal records in a database breeze right through…

    What if NAZI Germany had such robotic killing machine technology? You think that argument is over the top? Technology once developed will be transferred to other nations, some of which WILL use this technology against it’s own citizens and they won’t care one way or the other how accurate the software is – it will be about more efficient ways to put down uprisings. There are many nations today who would love this technology and who will eventually receive it once the Lock-Mart’s of the world roll it out.

    It’s all about the bucks here – lets not beat around the bush. Governments review the costs accommodated with solders such as pay and benefits, disability costs and pensions and look at how we can ‘rightsize’ the military’s human assets. This has nothing to do with ‘because it’s better’ – it’s because people can make a good buck at it. Any other argument ‘for’ this technology is an industry PR campaign rationalizing profits and shareholders returns over morals.

    • petenowak2000

      December 13, 2012 at 12:09 pm

      Oh come now, Godwin’s Law on only the third comment? Surely this is a new record!

      • Torontoworker

        December 17, 2012 at 1:21 pm

        I’m German. Next time I’ll invoke Saddam for you if that helps…

  4. Torontoworker

    December 13, 2012 at 9:36 am

    Damn auto correct. Replace accommodated with solders to ‘associated with soldiers’ in the third from last line.

    • petenowak2000

      December 17, 2012 at 2:24 pm

      I’m Polish. Touché? ; )

  5. Torontoworker

    December 19, 2012 at 10:11 am

    My wife is Polish… 🙂

%d bloggers like this: