Actually, something becomes sentient at an IQ of 1. The turing tests are designed to measure sentience, however no non-sentient machine would be capable of answering a proper MENSA IQ test, as there are at least a few questions that require abstract and adaptive thinking. I wrote an article on my personal effort at AI programming (you can read it
here)
But basically, there are some other points that I have previously not expressed on this issue.
a) AI systems dont have a pre programmed urge to terminate lifeforms deemed 'inefficiant'.
b) AI != Robots
c) Robot efficiency could lead to a golden age, as farms and other food and manufacturing industries could be operated 24/7 for minimal cost. Depends really on how people react to it, if systems are setup to minimise the possibility of greed influencing it, then capatalism's natural urge to force competitive pricing, and the lower costs of manufacturing would be a massive boon to the race.
-Gwala