Irving Wladawsky-Berger

The Three Laws of Robotics

Earlier this year, MIT professor emeritus Rodney Brooks gave the closing keynote at the 2024 MIT Sloan CIO Symposium. Professor Brooks was director of the MIT AI Lab from 1997 to 2003, and was the founding director of MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) from 2003 until 2007. A robotics entrepreneur, he’s founded of a number of companies, including iRobot, Rethink Robotics, and Robust.AI.

Given that AI was the overall theme of this year’s MIT Symposium, Brooks focused his keynote on “What Works and Doesn't Work with AI.” He reminded the audience that “AI has been on the verge of changing everything” ever since the field was founded in the 1950s, and explained why so many AI predictions have turned out so wrong.

I wrote about his excellent presentation in “Artificial Intelligence: Realistic Expectations vs. Irrational Exuberance,” and concluded the blog with what Brooks called “My Three Laws of Artificial Intelligence”:

• When an AI system performs a task, human observers immediately estimate its general competence in areas that seem related. Usually that estimate is wildly overinflated.
• Most successful AI deployments have a human somewhere in the loop (perhaps the person they are helping) and their intelligence smooths the edges.
• Without carefully boxing in how an AI system is deployed there is always a long tail of special cases that take decades to discover and fix. Paradoxically all those fixes are AI-complete themselves.

At the end of July, Brooks posted two new essays in his website. One was “Rodney Brooks’ Three Laws of Artificial Intelligence,” where he elaborated on his Three Laws of AI. I’d like to now discuss his second essay, “Rodney Brooks’ Three Laws of Robotics.”

Is the AI Revolution Already Losing Steam?

“It sure seems like the AI hype train is just leaving the station, and we should all hop aboard,” said “The AI Revolution Is Already Losing Steam,” a recent WSJ article by technology columnist Christopher Mims. “But significant disappointment may be on the horizon, both in terms of what AI can do, and the returns it will generate for investors. The rate of improvement for AIs is slowing, and there appear to be fewer applications than originally imagined for even the most capable of them. It is wildly expensive to build and run AI. New, competing AI models are popping up constantly, but it takes a long time for them to have a meaningful impact on how most people actually work.”

As we’ve learned since the advent of the Industrial Revolution, there’s generally been a significant time lag between the broad acceptance of a major new transformative technology and its ensuing impact on industries and economies. After transitioning from the R&D labs to the marketplace, it takes considerable time, — often decades, — for the products, services, and business models based on the new technologies to be widely embraced across economies and societies, and for their full benefits to be realized.

The reason, explained economists  Erik Brynjolfsson, Daniel Rock, and Chad Syverson in “The Productivity J-Curve,” a 2018 NBER working paper, is that while historically transformative technologies have the potential to radically change the economic environment, “realizing that potential requires larger intangible and often unmeasured investments and a fundamental rethinking of the organization of production itself.”

Their paper identifies two phases, investment and harvesting, in the life cycle of a historically transformative technology. In general, these technologies require massive complementary investments, such as business process redesign, co-invention of new products and business models, and the re-skilling of the workforce. Moreover, the more transformative the technologies, the longer it takes for them to reach the harvesting phase when they are widely embraced by companies and industries across the economy. Such an evolution has been has been pithily captured in what’s become known as Amara’s Law: We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.