Regarding the version of Tesla and subsequent improvement issues

On March 18th, Tesla comprehensively released the latest FSD Beta software V12.3. From the feedback of car owners across the internet, it is evident that this version has made significant improvements compared to the previous one.

Ashok Elluswamy, the Director of Tesla's autonomous driving software, mentioned that FSD Beta V12 has now surpassed the V11 stack, with many car owners joining this wave of deployment and posting that they have updated to the latest V12.3 version.

Elluswamy also stated:

"This is the first FSD version to use neural network 'end-to-end', directly driving and controlling the vehicle from raw camera videos and vehicle kinematics."

"The next generation 12.x version will bring unprecedented progress."

A tech blogger made an analogy: If Tesla FSD V12.3 is similar to ChatGPT-3 or 3.5, then FSD V12.4 may be similar to GPT-4.

Musk has previously mentioned that handling a large amount of video training data requires significant computing power, posing a huge challenge for Tesla. However, Musk recently posted that there has been a breakthrough in the issue of limited computing power—indicating that the nearly $2 billion investment made earlier is starting to pay off.

While the progress of Tesla FSD 12.3 is promising, if Tesla's training computation can ultimately be controlled within its capabilities, the next issue will be: how quickly can training computation identify and solve problems, and how many unresolved issues will there be? By 2024, can Tesla accelerate software releases and improvements, bringing higher reliability and functionality to FSD users worldwide?