InvestAI etc. - 5/28/2024

An open thread updated weekly.

Topics discussed this week:

Robotics pioneer Rodney Brooks

Challenges for AI with full self-driving cars

Precedent of 3G mobile technology

Daron Acemoglu on AI hype

Robotics pioneer Rodney Brooks

>

SAMI >

Richard, in the past you mentioned Rodney Brooks and his views about full self-driving (FSD) cars. This is an important case study for AI because FSD requires successful AI implementation in order to be fully functional. Can you expand on Brooks’s expertise and views?

>

RICHARD >

Sure. Robotics pioneer Rodney Brooks has followed the AI space for decades (fun fact: he worked in a lab next to mine, back at Stanford in the 1980s). As a former head of the legendary MIT AI Lab, Brooks has been involved with many robotics companies, including as co-founder of the one that makes the Roomba vacuum cleaner robots. Nobody has thought about self-driving cars more than he has.

Since 2017 he has been on the record with specific predictions, all of which have so far been remarkable prescient: 17 out of 17 of his forecasts from then were correct. See Brooks’s 2024 predictions scorecard.

Among current predictions, he is betting against electric cars (sorry California, gas will dominate for a long time) and flying cars (very unlikely, even in 2050). He thinks driverless taxi services (aka robotaxis) will not happen before 2032, and even then only in restricted areas of major cities. It will be 2045 before a majority of US cities feel comfortable enough with driverless cars to restrict human drivers and parking.

>

SAMI >

His timeline for FSD is a lot longer than what we are hearing from companies like Tesla or Waymo. They expect full autonomy much sooner than 2032 or 2045. This is worth emphasizing in terms of what is being communicated to the stock market and to investors.

In the case of Tesla, unit sales growth has abated and the main narrative has moved to FSD and robotaxis. Leading investors such as ARK Invest’s Cathie Wood and Baron Capital’s Ron Baron are banking on FSD within this business cycle. And Elon Musk himself has stated: “If you don’t believe that we will solve autonomy [FSD], you should not invest in the stock.” [April 23rd, 2024, Q1 conference call]

Regarding AI, the prevailing narrative from Tesla is that its automated driving technology is far ahead the competition because its already-installed versions of FSD have been collecting millions of pieces of data. This huge trove of information on driver behavior and road conditions is deemed to be an essential element for rolling out AI-supported FSD in the future. Is this real and as important as it sounds?

>