Rodney Brooks AI Scorecard 2024
Rodney Brooks updates his AI Scorecard Jan 2024
Three things LLMs have made us re-think
The “Turing Test” is no longer relevant
The Chinese Room still doesn’t understand Chinese. Brooks changed his mind about what he thought in 2002, that the question was moot because it was too technically challenging, but that if it happened it would mean that yes the system as a whole understands Chinese. Now he thinks it’s more complicated.
Chomsky’s Universal Grammar: LLMs seem to make this irrelevant – a large enough data set seems to understand grammar by itself . No brain regions have been identified that seem to correspond to a UG. But remember that humans are orders-of-magnitude more efficient (need much less training), so maybe there’s more to this than it looks.
Transportation
He stands by his 2018 predictions
Driverless cars are here now, sort of, but a serious driverless taxi won’t happen before 2032, and even then it’ll be for pre-determined points.
Electric cars hit 30% of US Sales: 2027
Flying cars will not happen anytime soon.
Robots
There are plenty of naive entrepreneurs saying that work will be changed by humanoid robots within a handful of years. They are wrong. My lack of predictions about humanoid robots were based on my expectations that they will not play any significant role for at least another 25 years.
AI
A robot that seems as intelligent, as attentive, and as faithful, as a dog: 2048 because this is much harder than people think.