Flashback: When AI paid chump change
Now that AI's brightest minds earn six- and seven-figure salaries, it's worth remembering that the field didn't always command the big bucks. I was reminded of this last week on stumbling over reactions to a famous 18-year-old Wired essay by computer pioneer Bill Joy.
What's going on: As in all other work, AI salaries are a function of supply and demand — there are relatively few genuine AI researchers, and extremely high demand for them. Ph.D. salaries could return to Earth if the fever diminishes — if, for example, investors sour on AI.
The 2000 essay by Joy, then chief scientist at Sun Microsystems, was called "Why the future doesn't need us." In it, he argued that new forms of technology were potentially dangerous and could render humans obsolete.
It sparked an ardent debate. One impassioned reaction was from Richard Wallace, a chatbot expert who created ALICE, the inspiration for Spike Jonze's Her, and who now works at Pandorabots. Wallace wrote:
"I found it slightly disingenuous for Joy to announce that robotics is the next big threat to mankind, when most people working in robotics and AI are barely scratching out a living. We would all like to found successful companies like Sun and become wealthy philosophers. But the last thing we need right now is more government regulations or the kind of negative publicity that gives pause to our investors. Our small startups are hardly as threatening as nuclear proliferation."
I asked Andrew Moore, dean of computer science at Carnegie Mellon University and a former VP of engineering at Google, for his thoughts on the 18-year-old comment.
- AI researchers are paid well, Moore said — "I do not feel sorry for scientists and engineers in the world of robotics and AI" — but pay disparity is a real problem. He said:
"If a recent grad can get an overall compensation package of 180K for working on AI for web search engines and a comp package of much less than half of that for (say) working for USDA on AI for preventing food-borne outbreaks, then we are going to see public safety and welfare losing out in AI adoption."
Moore agreed with Wallace about regulation: While self-driving cars should come under close oversight, he said, "regulating against potentially hitting the singularity would be like regulating against worm hole space travel."
Go deeper: Why AI researchers are so highly paid (NYT)