- MIT’s Brooks warns against overhyping generative AI.
- Purpose-built tech trumps broad AI applications.
- Challenges persist in AI development, despite potential in eldercare robots.
AI’s not-so-secret admirer
MIT’s Rodney Brooks, a robotics pioneer, believes we’re overestimating generative AI’s capabilities.
The co-founder of iRobot and Robust.ai acknowledges the technology’s impressiveness but warns against assuming it can match human abilities across the board.
Brooks emphasizes the importance of evaluating AI systems carefully, noting that people often generalize AI capabilities too broadly based on limited demonstrations.
Language models: not always the answer
Brooks argues that applying generative AI to every problem isn’t practical. He cites his warehouse robotics company, Robust.ai, where using language models for robot navigation would be inefficient compared to direct data streams from warehouse management software.
The robotics expert stresses the need for purpose-built technology that’s easy to understand and deploy at scale, always considering the business case and return on investment.
The long tail of AI challenges
Despite technological advancements, Brooks points out that AI will always face hard-to-solve outlier cases.
He debunks the myth of perpetual exponential growth in AI capabilities, using the iPod’s storage evolution as an analogy.
Looking ahead, Brooks sees potential for language models in domestic robots, particularly for eldercare, but cautions that significant challenges in control theory and optimization remain to be solved.
