Musk: FSD And AGI NEXT YEAR--Here's How!
ฝัง
- เผยแพร่เมื่อ 12 มี.ค. 2024
- Elon Musk does Ray Kurzweil one better and says he expects AI to be smarter than any human by 2025--and Andrej Karpathy weighs in on how automated coding relates to Tesla's Full Self Driving or FSD. And amazingly enough all of these things, plus compute power per dollar, relate to each other!
**If you are looking to purchase a new Tesla Car, Solar roof, Solar tiles or PowerWall, just click this link to get up to $500 off! www.tesla.com/referral/john11286. Thank you!
Join this channel to get access to perks:
/ @drknowitallknows
**To become part of our Patreon team, help support the channel, and get awesome perks, check out our Patreon site here: / drknowitallknows . Thanks for your support!
Get The Elon Musk Mission (I've got two chapters in it) here:
Paperback: amzn.to/3TQXV9g
Kindle: amzn.to/3U7f7Hr!
**Want some awesome Dr. Know-it-all merch, including the AI STUDENT DRIVER Bumper Sticker? Check out our awesome Merch store: drknowitall.itemorder.com/sale
For a limited time, use the code "Knows2021" to get 20% off your entire order!
**Check out Artimatic: www.artimatic.io
**You can help support this channel with one click! We have an Amazon Affiliate link in several countries. If you click the link for your country, anything you buy from Amazon in the next several hours gives us a small commission, and costs you nothing. Thank you!
* USA: amzn.to/39n5mPH
* Germany: amzn.to/2XbdxJi
* United Kingdom: amzn.to/3hGlzTR
* France: amzn.to/2KRAwXh
* Spain: amzn.to/3hJYYFV
**What do we use to shoot our videos?
-Sony alpha a7 III: amzn.to/3czV2XJ
--and lens: amzn.to/3aujOqE
-Feelworld portable field monitor: amzn.to/38yf2ah
-Neewer compact desk tripod: amzn.to/3l8yrUk
-Glidegear teleprompter: amzn.to/3rJeFkP
-Neewer dimmable LED lights: amzn.to/3qAg3oF
-Rode Wireless Go II Lavalier microphones: amzn.to/3eC9jUZ
-Rode NT USB+ Studio Microphone: amzn.to/3U65Q3w
-Focusrite Scarlette 2i2 audio interface: amzn.to/3l8vqDu
-Studio soundproofing tiles: amzn.to/3rFUtQU
-Sony MDR-7506 Professional Headphones: amzn.to/2OoDdBd
-Apple M1 Max Studio: amzn.to/3GfxPYY
-Apple M1 MacBook Pro: amzn.to/3wPYV1D
-Docking Station for MacBook: amzn.to/3yIhc1S
-Philips Brilliance 4K Docking Monitor: amzn.to/3xwSKAb
-Sabrent 8TB SSD drive: amzn.to/3rhSxQM
-DJI Mavic Mini Drone: amzn.to/2OnHCEw
-GoPro Hero 9 Black action camera: amzn.to/3vgVMrH
-GoPro Max 360 camera: amzn.to/3nORGYk
-Tesla phone mount: amzn.to/3U92fl9
-Suction car mount for camera: amzn.to/3tcUfRK
-Extender Rod for car mount camera: amzn.to/3wHQXsw
**Here are a few products we've found really fun and/or useful:
-NeoCharge Dryer/EV charger splitter: amzn.to/39UcKWx
-Lift pucks for your Tesla: amzn.to/3vJF3iB
-Emergency tire fill and repair kit: amzn.to/3vMkL8d
-CO2 Monitor: amzn.to/3PsQRh2
-Camping mattress for your Tesla model S/3/X/Y: amzn.to/3m7ffef
**Music by Zenlee. Check out his amazing music on instagram -@zenlee_music
or TH-cam - / @zenlee_music
Tesla Stock: TSLA
**EVANNEX
Check out the Evannex web site: evannex.com/
If you use my discount code, KnowsEVs, you get $10 off any order over $100!
**For business inquiries, please email me here: DrKnowItAllKnows@gmail.com
Twitter: / drknowitall16
Also on Twitter: @Tesla_UnPR: / tesla_un
Instagram: @drknowitallknows
**Want some outdoorsy videos? Check out Whole Nuts and Donuts: / @wholenutsanddonuts5741
Sources:
/ 1767738797276451090
/ 1767651809181548583 - วิทยาศาสตร์และเทคโนโลยี
Reminds me of the short story by Isaac Asimov back in the Golden days, about scientists building a large computer to explore the question if there is a god. When finally the computer was switched on, the computer answered this question:”Now there is”. 🤔
I've been thinking about this. How long until swaths of the public start worshipping AI as a god? I'm sure some already do. It'll be interesting to see the new religions that come out of this
@@perrybb2 How small their idea of what God is.....
@@Smitty65721people like measurable improvements in there life. Which is not how god works. I can understand why people would prefer to worship a false idol that gives immediate positive reinforcement. Though I agree that it is ridiculous to worship a man made machine.
@@TerraGlide NOT ridiculous if that entity becomes the giver and sustainer of life
Nice synopsis.
I was a solo software developer for 30 yrs. I would sit down with the decision makers and let them talk about what they saw as their problems and needs while I asked clarifying questions. I would then describe a piece of software that I could build that I thought would best address their needs. I would then build what we had described, putting it in their hands as I went and incorporating their feedback. This is what is needed and what it appears LLMs can do. Damn I am glad I am retired.
My story is almost identical to yours. I'm semi-retired now, and playing with the new toys is fascinating for me, it should probably be terrifying for any 30-year-old programmer
Next year? Every year next year?
This sort of projection is why I think v12 has a shot at training up at such a rapid rate that it'll quickly become near perfect... just get enough good video and then all extraneous avoidance scenarios in a rich way, and you're a long way down the road toward far, far better than human-best-efforts...
It’s interesting to me that the people who talk about FSD discuss it as a standalone, not as a part of a transportation system. The real value will come, not when individual cars can drive themselves like people. Instead, the real change will come when cars can talk to each other and the road. Take rush-hour delay as an example. It exists only because neither human beings nor an AI acting like a human can be sure what’s going on with the cars around them. If they could, cars could act like F1 or nascar drivers and drive a few inches apart. Bingo, no congestion. Or, we could have truly optimized route planning that could save time, fuel and aggravation. I’m not sure why this aspect isn’t addressed more frequently.
FSD enables the robotaxi which could dramatically reduce the number of cars society needs.
@@coolnameproductions2180 factor in what was once termed Jevons' paradox... The net result would be more travel and more transportation (deliberately not cars)
Talking of the thought process of software development, IMHO general engineering is basically the same overall thinking process. In college I was going to go into Mechanical Engineering, pretty much the most general version of engineering I believe but computers were in the infancy (IBM 360 was the new big dog on the block displacing IBM 1401 and accounting machines), so software was my life, but understanding the engineering process helped me greatly in attacking computer software problems. Hopefully school is where we learn HOW TO THINK, and we keep honing the skill thereafter. Computers just help us think better if we apply them appropriately to the task we have at hand.
Wonder if Musk will incorporate neural link data when training Optimus.
7:58 - I would add to that CEO explainer some thing far more common every day, explaining technical realities to the analyst or liaison, who writes the pseudo code or specs for the engineers.
I defiantly agree with you about the idea of an agent for a go between. I am a great example; I try to use Chat GTP, and image creator models etc. and I really can't get much out of them. If I had an agent explaining to me the prompts. etc. they would be much more useful. As it stands today it is only a bit of fun for me to try but no real good use for them.
It seems like the vital skills in the future are going to be in clearly communicating what you want in a way that the robot overlords can understand.
As used to be in good hotels, perhaps the AI will understand before we know ourselves what we require and make it so ahead of time
This really reminds me of Charles Stross's Accelerando book.
You would imagine that someone styling themselves Dr. would know what 'exponential' means.
I have also experienced the communication barrier when it comes to clients expressing what they want visually. they don't speak the language
The two stages of design you are deexciting is conceptual design vs detailed design.
The conceptual designer does a requirements analysis the defines the over all approach of the design (e.g. whether to use Kalman filter) and the detailed designer implements that design and a component level of testing (e.g. builds and tests the Kalman filter).
Also the conceptual designer might text the final system level design to see if it meets the requirements defined in teh requirements analysis.
Typically the conceptual designers needs a broad general knowledge while the detailed designer needs a deep domain knowledge of the component it is designing though there needs to be a large overlap between the two for a smooth transition between the two with few hiccups.
Off topic but on my mind.
Visual Language Model
VLM V LLM
A term now being used by Open Ai in the context of figure HumBot speaking and working demo. Tesla has been exercising VLM forever . Interesting new Start Point of common understanding.
Anyone remember the old movie Forbidden Planet? Earth is now that planet.
Seems like you know it all but when the stick price goes up
Once you mention Tesla level of self drive I lost internet in anything you might say 😢
Another possible avenue is bypassing high level languages and have the AI write nicely optimised machine code, like people did in the good old days. That way we could end up with code which runs many times faster on the same hardware than say Python (or even C) would.
Stock price
last 1% is always the hardest.
Here's my take on AI taking over software engineering (disclaimer: I've been a software engineer for 40+ years) - for brand new systems and brand new codebases, I can see the AI tools eventually doing a lot of great work (although I have some doubts about extremely complex systems - and I've worked on dozens of them over the years) ... but, for the huge base of existing software, there is absolutely no way that AI will ever be able to untangle the massive system messes that exist today - and be able to enhance those codebases with new features or more importantly, figure out why something is broken and knowing how to properly debug and fix it ... that will never, ever happen - and having worked on dozens of those kinds of systems (including the current one I'm helping to enhance and maintain), I say that with tremendous confidence ... but when all the existing codebases have been fully replaced with new systems that were created with AI ... well, more power to them ... but in the next 5 or 10 years? Yeah, I don't think so ... maybe about 50 to 100 years down the road, when today's existing codebases are no longer being used ... but there are banks and other institutions (like most governments) that are *still* using codebases that were developed back in the 1960s ... that's my 2 cents, John ... whaddyathink? 😱
I would love to hear your analysis of Lex Fridman’s interview with Yann LeCun and the principal model limitations he was talking about. How much detail do we know about how Tesla is addressing these limitations?
Of course Elon will not understand, but he will pretend and make people believe he is archiving something great.
So how Tesla compact/ Model 2 production ramping up earliest in 2027 fits to this picture? Is that analyst in Barron's setting "not Magnificent" TSLA target price to $125/share wrong?
Hi John, just to add that FSD v12 is a good example of why code will disappear... The C++ code was inferior BECAUSE it was humans attempting to catch all the edge cases and then building a rules based system that covered all the bases! The solution just let the neural networks do their thing LoL
Debugging c++ was/is difficult and time consuming. AI learning is continuous and able to "code" 24/7.
99.9% of the trips? of days? of seconds?
I think you are right on. We should have FSD in 2025, and waiting on the bureaucrats.
Ray predicted 2026 for 100 percent renewable energy in his 2004 book now hes saying 2034 which if correct i will still give credit because pretty close for huge prediction but the time is getting pushed back as we approach.
You'll know when AI becomes aware, when it makes up a list of Aspies and only wants to talk to them.
I find it ironic that when we combined the horse and cart whilst replacing the horse with combustion engines we called the resulting thing an automobile... Clearly now we are entering the real AUTOmobile age, however briefly, as the autonomy actually presages the separation of the horse from the cart once more... You see the reason the combustion engine was grafted unto the cart was due to it's non-autonomous nature (it needed a human driver to direct it LoL) Now that we are moving back to the horse that can move autonomously it makes no sense to join the two and by doing so reducing the horses capacity to move other carts when not required for the first cart...
Not end of this year? Bearish 😮
FSD coming before xmas 2025 🤞
you are predicting it for april first? XD
👍
A bit scary if bad people with evil intentions start creating software to do harmful things! John .. maybe you can come up with how this could damage humanity rather than advance it because for the moment your enthusiasm seems to say only good can come from this. 😮
Hi John, using AI to code will be a transitory phase and probably relatively short lived! AI will be able to perform the tasks for which coding is currently undertaken raw so why would it need to develop code using tools designed to allow humans to build inferior fixed rule based systems. Coding will be irrelevant except to us humans for our own learning / cognitive advancement like sport develops our physical abilities. Humans will be inferior for any task that requires extreme cognitive focus except maybe those humans that are neuro diverse enough to specialise in that way. FSD v12 is a good example of this in action, the C++ code was inferior to just letting the neural networks do their thing!
Any collaboration with starlink that can accelerate fsd development? Cars can talk to each other in real time.
Come on John! I love this stuff too, but we all know when Elon says something will happen next year, it almost certainly won’t. Even still, it’s just a question of when, not if.
Not quite. Full self driving where there is no access by pedestrians or regular human driven vehicles, absolutely. But amongst them, it may never be possible.
Im short tsla
Go with Musk's prediction...first hand knowledge.....his Tesla Giga Factory is already doing it.
AI will never be able to do everything in general that humans can do. (Anything in particular, sure. And all at the same time, of course.)
NEVER? .. really, so the progress we have seen in the last 12 months in your mind means not in 100 years of such change really?
@@steve.k4735 You misunderstood my comment.
When will Tesla monetize its ai?
It will have to first develop AI, which they and no one else have to date. Large language models are not intelligent or conscious, or anywhere close to it.
Sounds like the end of the world to me 😮
Is that so bad?
Sounds like technological advancement that will make everything much cheaper and improve our standards of living while not having to slave away at a job for it
So by next year not only will AI be as intelligent as Albert Einstein, but this also means that since this AI can be duplicated we can create 1000’s of Einsteins in parallel working 24/7 which is like having 1000’s of Einsteins that never sleep working on 1000’s of different areas .. cancer, cosmology, children psychology, nuclear fusion, … AMAZING. Will it be also as wise as the wisest man on earth by next year !!???
So sick of these BS FSD timelines! One year away for YEARS….
AI will be held up by the garbage in garbage out problem. We have the computational power, we need to create the educational process. It is clear from Google's fiasco that pure computational power without education gushes out gargantuan quantities of unreliable information. It can still make really nice wall art at Walmart. Yes, we need to incrementally review AI performance and improve it through AI generated programming. Education / Learning involves feedback. Tesla is way ahead in this understanding.
Hasn’t this guy “predicted” FSD every year since 2015?
To be fair to Elon, it wasn't until 2016 that he said it was fully autonomous, right now.
Does it even matter?
Exponential growth of FSD?
There is a very high probability that EVERY Tesla owner is now "underwater".
AI defines "Underwater" this way:
Trading In for a New Car: When you want to trade your car in for a new one, the negative equity (the difference between the car’s value and what you owe) must be addressed. It can either be paid off upfront or rolled into the loan for the new car, potentially prolonging the cycle of being underwater.
My 2022 Tesla Model Y LR fully loaded had an MSRP of $80,000 in May of 2022
Today that same exact car has a MSRP of $60,000
Elon gutted the price $20,000 in 1 year and 10 months
We paid cash and plan to keep the Tesla many years so we have been spared drowning
but Elon has put EVERY Tesla owner who financed their Tesla "underwater"
they owe more to the bank than Elon's gutting of the resale value of their car
So does it even matter aoubt exponential growth concerning anything about a Tesla?
I think not...I doubt the stock market will care either...just a guess
🙋♂️ AN HONEST QUESTION…WHY WOULD ANYONE WHO BUYS A TESLA WANT TO SELL IT… Traditionally when you drive a car off the wall, it loses 25% of its value give or take😞
@@budgetaudiophilelife-long5461 How about the death of the owner
divorce of the owner
moving to another country
legal reasons
company car was awarded to you so no need for another car
the list goes on and on
ALL VALID POINTS !!
Always selling fantasy instead of the present. There is zero chance of level 5 FSD by 2025. I'll bet you $1000 but of course you'd be a total fool to take me up on that bet.
Full self driving next year, wow! I can't believe it's already going to be 2017 soon!
There probably won’t be a need to learn a programming language. Learning psudocode would be more than enough.
I have zero faith in any prediction Elon makes... he has been proven over and over..wrong, and I'm a Tesla fan. Off topic, I'm also getting concerned about his politics, but that's for another day.
You are not alone.
It sounds like you’re a Democrat? That’s just sad.
When I hear AI I get furious. Smart AI is ridiculous, just another way to remove God from your life. If you think it’s ridiculous now, just wait until AI takes over.
Tesla stock is a mess. It has been my biggest position. Not anymore. I've sold half of it two weeks ago, will start buying back at 130. I have been a big Tesla fan and drive a M3. But the leadership by Mr. Alt Right Musk is a mess. Greetings from Berlin
Sell all your shares, you are not one of us.
@@garethrobinson2275 I don't need a cult like you do to feel confident, and spare me your cheesy Pluralis Majestatis.
@@garethrobinson2275 tit
4 years in and i'd love to be able to sell but i'm not prepared to take the hit, i'll probably add at the next low and trade the 150-250 yoyo back and forth whenever elon crashes the stock
He’s against mass illegal immigration and for free speech. If that is alt right to you, you’re pretty far left. Maybe stop letting the legacy media tell you what to think.
Human stupidity is infinite. But so is human genius. And now man has created something that will surpass it.