Our AI writing assistant, WriteUp, can assist you in easily writing any text. Click here to experience its capabilities.

Face it, self-driving cars still haven’t earned their stripes

View Original View Raw

Summary

Self-driving cars have made a lot of progress in the past few years, but they still have a problem with edge cases, which are out-of-the-ordinary circumstances that often confound machine learning algorithms. Despite this, they are still being tested on public roads in California at a cost of over $100 billion, but the cars still need human help to deal with edge cases. Recently, a few driverless cars have gotten stuck in concrete because of edge cases, demonstrating that the problem is far from solved. To address this, humans must use reasoning to solve edge cases instead of relying on data.

Q&As

What is the major problem with driverless cars?
The major problem with driverless cars is edge cases, out-of-the-ordinary circumstances that often confound machine learning algorithms.

How much money has been invested in the development of autonomous cars?
Over $100 billion has been invested in the development of autonomous cars.

What was Gary Marcus's opinion on driverless cars in a 2016 interview?
In a 2016 interview, Gary Marcus expressed that driverless cars were not extendable to the real world and that they struggle with recognizing uncommon scenarios.

What happened shortly after a California Public Utilities Commission approved Cruise and Waymo for operation 24/7?
Shortly after the California Public Utilities Commission approved Cruise and Waymo for operation 24/7, chaos ensued.

What is Gary Marcus's opinion on mission-critical uses of machine learning?
Gary Marcus's opinion on mission-critical uses of machine learning is that edge cases are everywhere and that anyone who thinks any of this is going to be easy is fooling themselves.

AI Comments

👍 This article provides a thorough and comprehensive analysis of the current state of technology regarding self-driving cars and the edge case problems that have yet to be solved. It is a great source of insight into the complexities of this field.

👎 This article contains a lot of technical jargon that may not be accessible to readers who are not knowledgeable in the field of self-driving cars. It may be difficult for some people to understand the nuances of the article.

AI Discussion

Me: It's about self-driving cars and the fact that they still haven't earned their stripes. Edge cases remain a serious, unsolved problem, even after over a hundred billion dollars in investment. Basically, the article is saying that while self-driving cars have made some progress, they still struggle with edge cases and real-world scenarios that they haven't been trained on. The article also talks about how the California Public Utilities Commission recently approved self-driving cars for 24/7 operation in San Francisco, which may be a bit premature.

Friend: Wow, that's a lot of money for something that's not even fully baked yet! What do you think the implications of this are?

Me: I think the implications are that we need to be cautious and thoughtful when investing in new technologies, especially when it comes to mission-critical applications like self-driving cars. We need to ensure that we have the right technology to address edge cases in the real world and that we understand the potential risks before we roll out new technologies. It's also important to remember that humans still have an important role to play in many applications, even if AI is involved.

Action items

Research the current state of self-driving cars and the challenges they face.
Follow the development of new technologies and methodologies that address edge cases in the real world.
Listen to Episode 2 of the podcast Humans versus Machines to learn more about the risks of AI.

Technical terms

Edge Cases: Edge cases are unusual or unexpected scenarios that can cause a system to fail or behave in an unexpected way.
Brute Force: Brute force is a type of algorithm that uses trial and error to solve a problem.
Deep Blue: Deep Blue was a chess-playing computer developed by IBM in the 1990s.
LLMs: LLMs stands for "Logic-based Learning Machines," which are computer programs that use logic to learn from data.
Mission Control: Mission control is a term used to refer to the centralized control center for a mission, such as a space mission or a driverless car mission.
CPUC: CPUC stands for the California Public Utilities Commission, which is the state agency responsible for regulating privately owned electric, natural gas, telecommunications, water, railroad, rail transit, and passenger transportation companies.
X: X is a social media platform.
Humans versus Machines: Humans versus Machines is a podcast hosted by Gary Marcus.