Starting with historical automata (strangely never called this) and bringing in humanoid robots, industrial robots and the whole science of robotics (plus quite a lot of artificial intelligence), the format gives us a series of chapters dealing with specific challenges such as sight, cognition and speech, each ending with a case study. The whole thing is finished off with a rather nice fiction/fact timeline on robotics through the ages, though it is rather unfortunate that the authors thought that Daleks were robots.
On the whole the coverage is good, though the level is perhaps a little superficial even for a reader with very limited knowledge of the topic. Outside the case studies there is relatively little narrative - more a collection of facts - but the book is rarely dull, helped by the glossy full colour illustrations throughout (the downside of this is the text is all on glossy paper, which makes it feel less like a real book).
There are a few small issues. The tone, as the cover suggests, is mostly positive to the extent of being sunny, and as such tends to avoid being clear about limitations. While we often hear something of what isn't possible now, it's almost always accompanied by a 'but it will get better' some time in the fuzzy future. Claims for existing robots can make them sound better than they are - we're told, for example, that Asimo could walk up stairs, which is true, but omits to point out this wasn't a universal ability, but rather the robot had to be given specific guidance for a particular flight of stairs. Also the occasional historical detail wasn't quite right. We're told SHRDLU is ‘a nonsense word Winograd made up’ - but it isn't, any more than QWERTYUIOP is a ‘nonsense word.’
Perhaps the weakest part is when dealing with self-driving cars. Much is made of them being potentially safer than human cars, and an interviewee is quoted as saying that he thinks the 'turning point will be when self-driving cars are ten times safer than human drivers.' But there is no attempt to unpack the implications. This would mean around 4,000 people a year killed in the US alone each year by self-driving cars - admittedly with 36,000 fewer expected deaths, but the 4,000 would be actual people with relatives to sue the manufacturers. There's a really interesting contrast with a section talking about self-flying planes where it says we still won't fly in planes without a human pilot 'And that's entirely justified...'
While the book could do with a bit more depth, it is attractive and covers a fair amount of ground. Perhaps best for teenagers with an interest in the topic.
Paperback: 
Using these links earns us commission at no cost to you
Using these links earns us commission at no cost to you
Review by Brian Clegg
Comments
Post a Comment