AI Just Took a Quantum Leap
ฝัง
- เผยแพร่เมื่อ 21 ก.พ. 2024
- What does one CREEPY Sora video have to tell us about Compute power vs understanding of the world? Turns out, a LOT--and Tesla's Full Self Driving has been doing it for a year now!
**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!
Go to drinkag1.com/drknowitall to get started on your first purchase and receive a FREE 1-year supply of Vitamin D3+K2 and 5 travel packs.
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
OpenAI Technical Report (with video): openai.com/research/video-gen...
Ashok Discusses Tesla Generative AI at CVPR: • ONE MODEL to Rule Them...
Elon Discusses Diffusion vs Transformers: • Musk: Diffusion to Rep...
All about Transformers: • Explaining TRANSFORMER... - วิทยาศาสตร์และเทคโนโลยี
As has been mentioned before, AI was supposed to take away all the dirty, dangerous, and tedious jobs. Now we see it's taking all the creative and fun ones.
Lol everyone is gonna have to get a "real" job while the robots make music and movies 😂 see you after lunchbreak Kanye
Many dangerous jobs have multitudes of cameras observing employees. That data can be interpreted by humanoid bots to carry out the same tasks within the same parameters. It'll happen just as fast for manual work and very soon
@@davidarbuckle5470 Excellent, then Utopia is coming soon!
@@garethrobinson2275what makes you think they would want to keep us around as useless eaters?
@Instant_Nerf Data, they eat data, and the universe is less interesting without us.
"a very scary thing" ... I think it's fantastic!
It fills me with awe and optimism.
The amazing future we've been promised for so long is about to come true.
I kind of scares me. Soon, we will have no idea what is real and what is fake. The potential for bad actors is limitless.
@@thefish5861I actually consider that a positive. It’ll force humans to become critical thinkers instead of just assuming whatever they see or read is real. Evolutionary step, in my opinion.
@@ZenTheMC I don't see how its going to force anyone into critical thinking. If people are already gullible enough to fall for the most basic of deep fakes and ai generated content thinking it is as real as anything, then how is the technology getting way better going to force people to get it? If anything it is going to make it vary easy for bad actors to take advantage of the population. The population has already been fooled and taken advantage of as long as humans have been together in mass, how does this technology make you more optimistic in this regard?
@@ZenTheMC You gonna waste a huge amount of your life if you have to fact check every little thing that comes your way. I think that in the end if too much false information ends up in the news , the internet etc some people will go back to books, papers and other channels of information that are not internet related and the vast majority will end up as a dangerous brainwashed mob
@@ZenTheMC Or is goes the route laid out in the movie Idiocracy
Great Update
Remember, a quantum leap is the smallest leap you can take..
The future of AI and Robotics is on a Quantum Leap, just as you indicated.... Tomorrow is already Yesterday.
Honestly if i see two day old news about ai i consider it outdated
Elon said Dojo was optimised for video and now it becomes clear why video training is important not just for FSD but for AGI also.
Which makes sense because how can you understand the world without seeing it.
Does this technology mean that politicians are going to have to start doing public speeches and eschewing being on video?
Maybe people will not trust anything they didn't witness first hand going forward.
Just like it used to be just 100 years ago and before.
People have been trusting politicians? I knew we were going wrong somewhere.
Crazy thing is by the time the election is over we will have the tech to make indistinguishable video/audio clones of people. The prez could end up dying but their team could continue making videos without the public knowing
@@garethrobinson2275 EUREKA! 😆
when showing a slide please remove your picture from the corner as it can obscure important details.
Regulators are going to require these self driving vehicles to not only be able to drive better than a human, they are going to require them to know the traffic laws and follow them. This is not something you get from driving videos.
What happens if we give it two or more “eyes”, and maybe in various wavelengths, and add sound, temperature, wind, humidity, smell (chromatography signatures?), and more sensors? Maybe pressure (feeling stuff)?
Could you add the link of the book in the description?
So how many quantum leaps are still required to make FSD work?
1 per day for 3,650 days
Will you be summarizing the token information from Andraj K?
Adding Mamba?
Hi John, I agree with your take on this and that we are really starting to see a pivotal moment for computers understanding and modelling of the real world. It really is hard to imagine. If we see in the next few years embodied robots with super human capability walking among us, I wonder if decentralized content moderation like Community Notes will gradually shift to cover the physical world in terms of what the public wants humanoid robots to be able to do. I think a model where humans collectively come to consensus on the way we want robots to behave in our world (beyond base level laws) would go a long way to a prosperous future and remove the risk of a few powerful individuals having to much control.
I'm only half way through the Karpathy Tokenization video as well ;)
The leash also jumps into her hand, magically.
I have no idea what the heck you are talking about but it seems pretty amazing 😅🤯
Exactly. Most of this is crap.
What he is talking about is OpenAI's 'Sora' - which can create a video after being given only a text description of what is happening. By adding more 'compute' to the task - you can see the visible improvement that it is able to output. He is also talking about how Tesla's self driving system builds a 3D model of the real world and makes accurate predictions on what is about to happen before it actually happens.
love your videos, you make me feel like i am in future
I was one of the lucky people to get version the limited early version of FSD Beta V12.2.1😊
The behavior of prioritizing FSD actions over mapping navigation path is definitely superior and more human like than the previous version 11... which would cause FSD to error fail within sight of arriving at my home address.
Ideally I hope future versions will be capable of doing multiple actions in sequence to park my car in my preferred garage space of a multi car garage to successfully negotiating the electronically linked garage door opener to both open and close as needed. This FSD version currently ends travel within sight of my house about a couple of houses away from my driveway. A new side effect is that having autopilot take a different route to my home causes the car to not complete the driving path by stopping on the street below and one street away from my house. It seems to be confused about the current distance from my house one street away and brings the car to a complete stop, no knowing what to do but also not terminating the drive. I had to take manual control to finish going around three sides of the block to get home 😢 I am sure part of this issue is because the map location for my home address is pointing to a place between my residence and the next house which would be impossible to drive to without jumping street curbs and off roading through the front yards.
I'm still hoping to see an option to associate a specific location for the car to be parked at and linked to the home address with another one linked to the work address.
It is interesting that there have not been outstanding art or literature for many years. The works that are now passed off as "Art" is often from people who have no classical training but only turning out poor quality for commercial markets. The jury if out on the possible societal effects of AI. It is reminiscent of the Luddites trying to stop mechanical looms. Creative people will always have a place in society. We could be moving to a world of great abundance that has eluded humanity. There will be society changes, but could be positive for everyone.
"Creative people will always have a place in (a healthy) society."
Terrific job distilling this breakthrough. The world will soon wake up to new, exciting and yes, somewhat terrifying possibilities. It will be up to us, evolved apes, to steer it in the right direction.
Looks like AI generated video for the Dandy Warhols w/Slash - "I'd like to help you with your problem".
I remember Yann LeCun specifically saying text to video would never get rid of smudging lol. Guess we’re all getting surprised.
In other words, a patch is worth a thousand lines of text.
It won’t be long before we can give it Gone With the Wind and get a movie from it.
When TSLA ATH 😢
This is all truly fascinating stuff; I am mostly interested, however, in how AI can improve medicine and healthcare, more effectively distribute resources, particularly to the Have-Nots of the world, and perhaps most importantly, allow for a direct democracy in a way that gets these politicians off our backs. While what we really need is a way to cut through all the ridiculous stuff that isn’t helping people, we seem to be creating mostly just new ways of entertaining ourselves and deceiving each other. On a more positive note, Autonomous vehicles and artificially intelligent humanoid robots are significant examples of what great things can come from AI. Great channel Dr., please keep up the great work. I know you are in this stuff for the fun of it and to move society forward into a better tomorrow.
Should I still buy $NVDA?!
I just watched irobot again last night and it occurs in 2035. Seems eerily accurate.
What happens when we deliberately force a real physics version of the world into the training model? Make it a true real world model of the world. That will be amazing.
So do you think Apple is using people's camera videos from icloud storage to train their AI?
man....i miss the old dr know it all. he used to post things other than tesla and AI...like his classic ''do we deserve democracy'' or ''how many bones in a dogs body''
and who can forget the best one of all:"''what is hdmi?''
it just sucks to see a good creator succumb to this stupid algoritm clout chasing behavior.
i miss you man
Man, you said it! I really missed those insightful videos, I loved getting his perspective on things.
Besides entertainment, what uses can this so-called simulator be if under the hood is a black box? I genuinely wanna know
Copy/Paste this into your Ai chatbot of choice...
"Consider this hypothetical. there is a robot that can imagine. It can manipulate its environment the same way a human can and can speak any language. The Imagination in this robot manifests as a vivid 3d reconstruction of what it is thinking about. End hypothetical. Please tell me some benefits of a robot like this."
Unless the bots have the right sensors, such as FLIR, as Munro advocates for, the nth degree of attainment will not be achieved...but, maybe the nth degree is not needed.
Alex -- im gonna take the fluffy puppy in the middle for 200 please
Very interesting. Absolutely on target with the biological references. To be embodied in the world is to be dynamic in time. To be reasonably aware of the world is to be able to imagine/predict what seems likely to happen next. Essentially this is AI becoming sensitive to existential phenomenology. Of course, even the shortest predictions can be wrong, but (biologic) agents have to plan and act on their best estimates anyway, even in face of the possibility of being in error. The better a bug can do that, it can avoid predators. For a car, in a driver's world, the better it can do that - with 360 camera vision, much faster "reflexes", etc., the safer we all are (pedestrians, passengers, other drivers, etc.). IMO, this is a big reason NOT to go cheap on sensors of all sorts for the Optimus Robot.
USS for Optimus.
@@OrbitTheSunIf that means ultrasound, that does have some uses: e.g., (1) high powered, great for cleaning; (2) could be for barrier detection and direction finding in the dark; (3) could be for looking for broken bones, aneurysms, and babies. Lots of other things probably. Bluetooth would be better for device communication.
Star Trek Next Generation Hologram Deck is next. 3d 360 world created by AI!
It’s sky net and the Matrix movie becoming real.
Thx for taking the time to walk us through this. I think it is cool but my son the artist does not.
Oxford defines a Quantum Leap as "a huge, often sudden, increase or advance in something.". I dont know what these commenters are going on about. Great video!
If I could pause the entertainment aspects of AI and shunt all of that development into practical benefits like legal, political, materials science, and medical, I would. I’m sick of demos trying to entertain me.
When can I vote for an AI to run my state? Human politicians are far too hackable.
Sora e top
Poor college students. What can they possibly say they are going to major in?
im the poor guy. need career guidance yall.
@@CausallyExplained It is disheartening. I never thought I'd see it in my lifetime. I wished for it. Now it's here. Although I'm a retired boomer, I worry about my four granddaughters 3 to 16 years old. Universal income might sound great to some. But It robs humans of their creative spirit. Stat safe amigo.
OK...A few questions.... If Tesla FSD AI is learning from videos of human driving can it get better than a human who always pays attention at the wheel? I suspect it can but what do you train it on after you have already exposed it to the best of human drivers?
Will Tesla's training computer have to create a simulation video for every one-off weird driving situations using this AI video tech? How do you know you have covered every conceivable permutation.... other than actually encountering it in the real world? (If there is such a thing).
Does this replace Tesla's use of UnReal engine for simulations?
It learns like we do. They show it videos to give it experience, but it operates like we do with that experience, in the sense that it makes decisions based on what has been learned before.
@@caryg4638 Right, but does that infer that it can't get better than the data it's being fed?.... which is just human level? I.E. will it never get superhuman at driving?
@@nickmcconnell1291 it most certainly will get better, because better is only a product of accuracy and non-accidents. If it can do the correct thing in all scenarios, having learned, and if it cannot make mistakes/errors in all scenarios, then it is most certainly better. What you mean about superhuman ability doesn’t really apply to on-road things (maybe race tracks where inches matter), except to infer that it will be able to predict things seconds in advance based on traffic patterns, at which point I would imagine, with the correct amount of processing and data, it would, because that is the foundation of llm already, predicting.
You will never get to zero accidents or wrecks, stuff happens but that doesn’t mean that it’s not safe. The best drivers if they drive enough in high traffic areas will have a accident sooner or later.
@@caryg4638 I get your point. However I imagine that their are driving situation scenarios that the best human could not react appropriately too or fast enough too to avoid a collision. I would hope that somehow an AI could get to where it could.
For example:
A semi truck parked on a hill looses its parking brake and rolls backwards towards your car at speed. Your car is proceeding up that same hill. At the same time there are pedestrians on the sidewalks around you and traffic behind you.
Thanks John.
My one concern is that listening to your presentations you continue to equate FSD with the Tesla AI system that helps design and train FSD. As programmers and engineers, we know they are not the same. We need to teach the public about the major differences between these AI systems. From the beginning of this leap forward in AI, the public needs to understand the difference between a fixed AI and a dynamic AI system running on a supercomputer.
FSD and the internal Optimus AIs are examples of these fixed AIs. They are unable to learn or to reason. They are limited to the skills imparted by training or programming.
Tesla's training AI system and Open AI's system run on supercomputers and have far greater abilities and potentials than today's mobile AIs. These large scale AIs are able to construct digital realities that generally conform to real world physics from brief prompts. This is only one of their many skills. We know that Tesla's AI training system is helping to design AI systems by suggesting AI models and functions for each stage in the new AI. We know that these AI systems can project the future and respond to those projections with realistic behaviors. These are not skills of the FSD or Optimus mobile AIs.
Even these fixed AIs like Optimus or FSD 12 must be able to predict the future for a few seconds. They are trained to do this.
Hey I've been watching for you to finally get something right here on this channel - AI has taken a Quantium Leap! For sure since a quantum leap is the smallest possible leap - you got it right this time.
Head in the sand.
When sora released to public its the end of the world as we know it
...but I feel fine.
Training and inference have nothing connecting them.
For all but not Tesla lol
There is still something fishy about the dog's tail and right foreleg. And the woman's forearm. Hm.
The leap will be when it figures out how to benefit from being glorified and worshipped by humans.
This is all interesting, but I can't help but think, that so much human innovation is focused and virtual reality. That does not truly affect our physical world, nearly as much of the locomotive and flight did well over 100 years.... Yawn
24 min of click bait
Whichever combination of voice commands I use in my Model 3, 99% of the time I get “command not understood”. Useless gimmickry!
You do know quantum means very, very, very small?
Open mouths for views,😮
Your content is pretty good, your thumbnails are off-putting. Even Mr.Beast stopped using open mouth thumbnails. Do better please.