I read the entire article, extremely interesting. I think he's got the correct idea to simplify automated driving by NOT writing a million line of code to account for every situation.
Yes, exactly. I've always said rote is worse than calculation, equation, knowing the goals. But he might have been throwing that out there as a red herring. I can't imagine any successful AI being millions of lines of code. That would be horrible. It has to be short, precisely woven.
Instead, his is a learned computer, which is where computers will have to go to advance. Plus, if his algo's are good, it will be far less labor intensive to code, etc. I also thought it was interesting to read Elon's apparent note about Tesla "getting rid of Mobile Eye,".
There's something corrupt about thinking that an area of California full of redwood trees would be worse than a war-torn religious mecca at programming something. Of course Musk will have had come to that realization. I have no idea of the history behind it. However, it's too bad he shot his wad in the war-torn area: it's OK to test stuff there, as long as the AI doesn't get religious. Um, ok, I take that back. Also, California has been getting communist religious -- that's dangerous too. Of all people, I like the fact that Musk is doing it, since he can tap into multiple morals, areas, and disciplines.
This garage guy -- he's just obviously better at it. I'm glad he's played the ground too. Musk's approach always seemed plodding and slow to me. I've seen that Musk can be bad at making deals: not really sure why he's giving Hotz the middle finger. Musk is overly corporate now. That's something you need with big factories to some extent. Of course metrics are metrics and everyone can measure like that so maybe they don't need to get better at it --- just let it roll on to its conclusion by itself.
- - - Updated - - -
I find the article tone juvenile, bordering on childish thinking, if not on ignorance and a lack of understanding of the complexities involved in the discussed topic.
...
Hopefully, he'll also learn how much he does not know. He might also learn to be more humble and to be grateful to these developers that persevere with mundane bug fixing.
About the knowing: a lot of us are taught we don't know, and when we clearly do know far better than we're told we know, our great realization is our own belief in ourselves. That is extremely powerful: that all the nay-sayers are just WRONG. That is something that we have to go through to to be able to actually work on ANYTHING, otherwise we get stuck, stagnate and die. Furthermore, the "I know everything there is to know" is an admission that you are at the cutting edge and must yourself move it forward (along with anybody around you), not that all knowledge of all things has been downloaded into your brain. HOWEVER: in any field, there is a point at which the equations are well written, and often one can look at the old equations of yesteryear and realize that one's equations are far superior. Sometimes you just like to sit down and appreciate your accomplishments and say "my equations are right, I know them, they are good enough to cover everything in this field", essentially erasing the field and subsuming it in some new goal. Putting knowledge to bed is a great thing indeed.
All I'm trying to explain to you is that there is a language of inventors and a language of laggards that is different although it often uses similar words and sentences (unfortunately). There are also commonalities.
About the whole Google/Facebook bugs/ads thing: I have never worked there, so I'll be the one who raises my hand and says I don't know. I've known some people who had to quit though. Ads are stupid. No one should ever use ads. Micropayments are better. The whole reason there was resistance to micropayments is that corporate types of the time wanted subscription, worship, religion, contract, lock-in, and raping charges way over competitive value, but these days there are enough people who know what the Internet is that micropayments would work fine. We don't need ads, coercion, lies, theft.