3 Ways to Reset AI Expectations
๐ Abstract
The article discusses the current state of AI, highlighting both the hype and the reality around its development and deployment. It presents the "Three Laws of Artificial Intelligence" proposed by MIT professor emeritus Rodney Brooks, which aim to provide a more balanced perspective on the capabilities and limitations of AI systems. The article also explores the reasons why many AI predictions have been inaccurate, outlining Brooks' "Seven Deadly Sins of Predicting the Future of AI."
๐ Q&A
[01] AI: Hype vs. Reality
1. What are the two extreme messages often heard about AI?
- AI is either seen as a panacea for all problems or a dangerous threat that requires caution.
2. What did Rodney Brooks, the MIT professor emeritus, present in his closing keynote?
- Brooks presented his "Three Laws of Artificial Intelligence" to put AI in perspective:
- Most successful AI deployments have a human somewhere in the loop, 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.
3. What were the early beliefs and expectations about AI development?
- In the 1950s, the founders of the AI field believed that human intelligence could be precisely expressed as software and executed in powerful computers.
- In the 1980s, leading AI researchers were convinced that human-like cognitive capabilities could be developed within a generation.
4. What happened after these early ambitious AI approaches?
- It became clear that the difficulties of developing machines with human-like intelligence had been grossly underestimated, as cognitive capabilities like language, thinking, and reasoning cannot be easily expressed in software.
- After years of unfulfilled promises and hype, these ambitious AI approaches were abandoned, leading to a "so-called AI winter of reduced interest and funding."
[02] The Rebirth and Milestones of AI
1. How was AI reborn in the 1990s?
- Instead of trying to program human-like intelligence, the field embraced a statistical approach based on analyzing patterns in vast amounts of data with sophisticated algorithms and high-performance supercomputers.
- This information-based approach produced something akin to intelligence, and the statistical approaches scaled very well as more data, sophisticated algorithms, and powerful supercomputers became available.
2. What were some of the important milestones achieved by AI in the following decades?
- Deep Blue's win over chess grandmaster Garry Kasparov in 1997
- Watson's win on the Jeopardy! Challenge against the two best human Jeopardy! players in 2011
- AlphaGo's unexpected win over Lee Sedol, one of the world's top Go players, in 2016
- Successful completion of the 2007 DARPA Grand Challenge for self-driving vehicles and the 2012 DARPA Robotics Challenge for the use of robots in disaster or emergency-response scenarios
3. How did AI appear after these milestones?
- After these and other milestones, AI appeared to be "on the verge of changing everything," as stated in the article.
[03] The Seven Deadly Sins of Predicting the Future of AI
1. What are the "Seven Deadly Sins of Predicting the Future of AI" according to Rodney Brooks?
- Exponentials: Overestimating the continued exponential growth of technology
- Performance versus Competence: Generalizing a system's performance on a specific task to its overall competence
- Speed of Deployment: Underestimating the time it takes for new AI systems to be widely deployed
- Hollywood Scenarios: Ignoring the gradual evolution of technology and the changes in the world that will occur before the development of super-intelligent AI
- Suitcase Words: Using terms with multiple, confusing meanings that mislead people about the actual capabilities of AI systems
2. Why have so many AI predictions been wrong, according to Brooks?
- The reasons include the "Seven Deadly Sins" that lead to irrational exuberance and unrealistic expectations about the pace and capabilities of AI development.
3. What was Brooks' purpose in making his annual Predictions Scorecard?
- Brooks made his predictions with attached dates to slow down the expectations and inject some reality into what he saw as irrational exuberance around AI.
[04] The Bottom Line
1. What is the overall conclusion about the state of AI?
- The article concludes that we've come a long way with AI, but there is still a long way to go. The truth about AI likely lies somewhere between the two extreme messages of it being a panacea or a dangerous threat.