Most Helpful Customer Reviews
87 of 95 people found the following review helpful:
4.0 out of 5 stars
JohnHawley El Paso, Texas, March 10, 2009
I work as an engineering psychologist in a U.S. Army organization that is in the forefront of R&D on military robotics and automated command and control systems. Hence, I read P.J. Sanger's Wired for War with considerable interest. I can relate to much of his discussion on an experiential basis. We routinely encounter and try to provide solutions for many of the problems Sanger discusses. As a point of interest, I was the technical lead on an Army effort looking at human performance contributors to the fratricides by the Patriot air defense missile system during the recent Gulf War mentioned on page 125. As is usually the case in a casual summary of complex events, Sanger's description of these events is superficially accurate, but there is a lot more to the story. Also, I've been told that his remark on page 197 about the radar on the DIVAD gun locking onto the exhaust fan of a port-a-potty is an urban legend. I've heard about this alleged incident, but I've never been able to find anyone in the Army air defense community who ever witnessed it personally. We work tests on that class of systems all the time, so we know the players.
Overall, I thought Sanger did a good job of describing the state of the art in robotic military systems and addressing the potential sociological and psychological impact of using these systems in current and future military operations. From my perspective, the central operational issue in using armed robotic systems in combat is balancing autonomy with effective human control (the focus of Sanger's Chapter 6.). In my view, he correctly refers to this topic as the "Issue-That-Must-Not-Be-Discussed." I was particularly struck by the difference between the attitude of those having the most on-the-ground experience with these systems (e.g. Robert Quinn's remark on page 124 that he can't even imagine how unmanned systems would "ever be able to autonomously fire their weapons.") and the almost casual attitude on this subject expressed by many of the decision makers we deal with daily. Their attitude is best summarized by the remark attributed to an unnamed former secretary of the army who responded "No" when asked if he could identify any challenges that the greater use of unmanned systems would bring to the military.
The reality associated with greater autonomy on the part of armed robotic systems is that there will likely be many more "oops moments" (Sanger's page 196) than are politically and operationally tolerable. Based on our assessment of the Patriot fratricides during the recent Gulf War, these incidents were an example of an oops moment on the part of an armed robotic system. If the past is any indicator of the future, such incidents will result in initial "surprise" and "shock" on the part of the leadership that these advanced systems behaved thusly, followed by the imposition of restrictive rules of engagements that effectively take the offending system out of the fight. Sanger is correct that we need a more realistic assessment by those in policy-making jobs of the potential problems associated with the use of armed, autonomous robotic systems in actual combat--but I'm not holding my breath waiting for this to happen.
Armed robotic systems will be fielded. They will be allowed to operate autonomously. Oops moments will occur. And unpleasant fallout and scapegoating will take place in the aftermath of such incidents. The issue of control in accord with human intent versus the illusion of control is complex and will not easily be solved. Software glitches aside, oops moments will mostly result from what Dave Woods of Ohio State University terms the "brittleness problem of automata:" An inability to satisfactorily handle unusual or ambiguous situations. I fear that the "Strong AI" necessary to satisfactorily address the brittleness problem will remain tantalizingly just over the technical horizon for some time to come.
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
32 of 35 people found the following review helpful:
5.0 out of 5 stars
A truly eye-opening book, superbly researched and written, February 2, 2009
I first heard the author talking on NPR about this topic, and both that interview and the first chapter of this book show his excitement and deep interest and understanding of this subject. For such a weighty hardback, it's remarkably hard to put down, and each section evolves intelligently from the last. I particularly enjoyed the references to modern culture, given that robotics has largely been a subject of science fiction in the last few decades rather than yielding anything practical in reality.
Well, at least so I thought - it turns out that over 12,000 robots are at war in Iraq and Afghanistan as we speak. The companies producing these machines were spurred by the very real necessities of dealing with guerrilla warfare, and avoiding the human toll associated with such difficult environments. Through a combination of human-controlled and artificially-intelligent hardware, these robots back up our soldiers and provide a super-human level of robustness and accuracy.
The author raises the complex moral questions associated with having machines killing people on the frontline, and the issues that arise when mistakes occur. There's also a fascinating discussion of stress disorders that remote pilots are suffering from - these men and women sit in offices in the US, controlling machines on the battleground far away, and return home for dinner every day after "a day's fighting".
It's also interesting to look at the design of some of the machines and their control interfaces, many of which look like Wall-E with a machine gun. Weapons companies have copied controllers from the Playstation and Xbox, taking advantage of a generation that is comfortable using these devices without extensive retraining. The distance between shooting people on Halo and making real life-or-death decisions in operating a military robot is almost absurdly non-existent.
I don't want to steal the book's thunder at all since this is one of the most gripping reads I've found in a while, and would highly recommend to everyone. While not a robotics book or a war book, it falls somewhere in the middle, and the topic is enthusiastically presented. The most chilling part is clearly that the science fiction of movies such as The Terminator is really not too far away, and we're on a cusp of a robotics revolution that will be as profound as the domination of the PC.
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
25 of 27 people found the following review helpful:
2.0 out of 5 stars
A little too sensationalist, not enough real., February 3, 2011
Singer paints a picture of vastly capable robots and software that are fielded right now. As someone working in the robotics field and trying to provide autonomous behaviors for government applications, I see how far this is from reality. As is often the case, however, reality doesn't create much buzz or sell many books.
This book feeds the popular misconception that robots are smart and getting smarter. I have a brother-in-law that was asking me about my work and how I'd done some simple AI design for computer board games for fun a long time ago. He made the comment, "I bet all that is coming in handy in your current job". I had to tell him that no, creating strategy-based behaviors for Risk has almost zero relevance to modern robotics -- we're nowhere close to a strategic level of thinking. As an industry, we're still at the level of getting a robot to move from point A to point B consistently and without running into anything. The videos on YouTube posted by researchers show some incredible things, but research is almost always 10-15 years ahead of a solid, marketable solution (toy problems in the lab are comparatively easy, real-world complexity is HARD).
The reality is this: Most mobile robots in theater right now are glorified remote control cars, operated by soldiers less than a few hundred meters away via cameras mounted on the robots. Singer talks a great deal about the Foster-Miller Talon and iRobot Packbot, because they are far and away the most common and prominent platforms in theater. However, the examples of autonomy he gives never deal with those platforms. Why? Because they have almost no autonomy for the units in the field.
Autonomy for mobile robotics is HARD. Very hard. Singer glosses over this fact by talking about the "inevitable" Singularity that is supposed to happen somewhere around 2030. Basically, the premise is that robots are not smart because they can't think fast enough to process all of the data. This is wildly inaccurate. Mobile robots are not smart because humans have not managed to impart intelligent decision-making to them. It's not like the robotics field is bemoaning the slowness of today's processors as a reason for autonomy failings. "Oh it WOULD have worked if only I had 10x the computational power..."
Also the chapter on the "Singularity" bothered me because it shows a distinct lack of understanding of the current state of the art. Computers are not getting faster at the moment. Processor speeds have flat-lined at about 3.0 GHz for the past several years because any faster and heat dissipation becomes an intractable problem. Even the speed of super-computers is capping out for the same reasons (a little more complex -- heat dissipation is easier if processors are spread out, but higher speed requires physical closeness because of the time needed for current to travel. So high-speed computers need to be compact for speed, but spacious for heat dissipation...). New computers today are coming out with more cores, so computers technically have more raw computing power (though even this has near-term limitations that prevent us from the Singularity) but computers are awful at taking advantage of parallelism. Singer points out that our brains are massively parallel, and this is why we have an edge on computers. This is true, but even is we had a computer with the same amount of parallelism, we couldn't take advantage of it. Someone has to program the thing, and no algorithms exist that mimic the behavior of the human brain (contrary to the picture that tech bloggers paint). The fact of the matter is, we don't know how the human brains works. Singer's main argument here is pointing to the trend: compare what computers can do today with what they could do 30 years ago. The flaw is in assuming that the same trend will continue. Extrapolation is notoriously bad, even though it seems to have such predictive power. The reason for such growth can be explained by us tackling the "low-hanging fruit" as far as computers go, and we're fast approaching an era of more incremental improvement as the "easy" problems are solved. Futuristic technologies such as alternative processor architectures, Quantum computing, optical computing, etc. are nowhere near workable. But even if we had a full-fledged Quantum computer right now, we really wouldn't know what to do with it. The software algorithms don't exist for it, and many of those that do don't provide dramatic improvement over what we have today.
The most valuable part of the book (and the only reason it didn't get 1 star) was the first few chapters describing a historical view of robotics as an industry.
Help other customers find the most helpful reviews
Was this review helpful to you? Yes
No
|
![[Start the discussion]](http://g-ecx.images-amazon.com/images/G/01/x-locale/communities/discussion_boards/start-new-discussion-sec._V192250339_.gif){kind=small}
