Artificial Intelligence vs. Intelligence Amplification, Part II
If we look at past analogs of AI vs. IA, we can see that IA is the way to bet.
There have been numerous attempts at producing artificial intelligence. Those attempts have been failures. The nearest thing to a success was Deep Blue as chess champion. I suspect a grandmaster assisted by a computer could probably beat an unaided computer for the foreseeable future.
In the AI scenario, technological unemployment is a real possibility as humans are replaced by robots. In the IA scenario, we still need humans to run the robots. Judging by the lack of technological unemployment in the real world, IA is more realistic.
Much of my day job involves turning scholarly papers into SGML (sometimes including MathML). The programs (some of which I wrote) will go only part of the way. The papers still need tweaking.
The theory that a few Great Minds can outdo a crowd of supposedly-lesser minds is one of the bases of central planning. The central planners supposedly know best. In the real world …
Along similar lines, attempts to keep undesirables out of the United States generally kept out people we can now recognize as assets.
In the Leviathan version of AI, democracies have generally been more successful than dictatorships. Similarly, attempts to regulate corporations (whether pro- or anti-corporate), have generally pushed business across the state or national boundary.