Visions by Michio Kaku
Computers, Biology, and miscellaneous energy physics sums up Kaku's 1997 projections for developments in science over the near term (now in 2010) and long-term (2050 and beyond).
This book reads like a survey of current science with some glimpses towards the future. Which means that Kaku is winning his bet that science's path forward should be more predictable than other prognostications. When you come across broken estimates (ITER's 1GW fusion reactor online by 2010), you think "whoops". However, when you realize that you've made it 80% through the book before then, you just smile and congratulate Kaku.