Moravec presents a synopsis of the current and projected status of robot implementation in our society. The 2000s have seen robotic toys and household appliances gaining customer acceptance. A quick explanation as to why robots are now progressing in the marketplace, after a lull in artificial intelligence (AI) during the 1970s, is that computing power has risen to above 1,000 million instructions per second (MIPS), and will continue to rise.
Robotic soccer competitions illustrate the current maximal implementation level for robotic toys. That will soon change. The author speculates that the goal of human performance levels for robots is not far away. In fact, Moravec suggests that reliable mobile utility robots will appear in hundreds of millions of industrial situations and homes in less than a decade.
On the near-term horizon are self-navigating robots, machines that will map and follow floor cleaning routes, or engage in security patrols. Perhaps some will be built to perform both functions. By 2020, robots will appear that can be programmed to handle almost any simple chore. Then, ten years later, Moravec suggests that robots will be trainable, and, within ten more years, monkeylike. Finally, within 50 years, robots will be human-like. “Properly educated, the resulting robots are likely to be intellectually formidable,” according to Moravec.
The article is direct, and quite reasonable. It follows logically from projected increases in computer power. If there are errors, they are most likely to be in the precision of time estimates, rather than in the sequence of AI application outcomes. The audience for this article would include anyone likely to be affected by impending robot implementations.