Since atomic and nuclear weapons were developed in the mid-20th century, containing their spread and preventing their use has been a major necessity for national security and civil society. For better or worse, equally powerful and dangerous technologies are likely to be developed in this century, including advanced artificial intelligence.
Repressive governments are already using facial recognition to track their citizens. Microsoft President Brad Smith recently warned that if legal means to restrain the misuse of artificial intelligence are not created in the next few years, governments will be able to watch people daily as in an Orwellian dystopia.
An even more dangerous possibility is that in a few more decades, we will create artificial-intelligence applications capable of self-improvement over time. This could mean these applications eventually become too sophisticated and powerful for anyone, good or bad, to understand or control.
Fortunately there is still time to find ways to keep artificial intelligence under control so it helps instead of hurts human life and freedom.
I hope eventually a treaty to control uses of artificial intelligence — similar to the Reagan-era treaties to control nuclear missiles — might be achieved,
JOHN J. DZIAK
State College, Pa.