Due to popular demand, Tesla AI team release roadmap:
September 2024
- v12.5.2 with ~3x improved miles between necessary interventions
- v12.5.2 on AI3 computer (unified models for AI3 and AI4)
- Actually Smart Summon
- Cybertruck Autopark 📐
- Eye-tracking with sunglasses 🕶️
-
@Tesla
AI team is at
@CVPR
in Vancouver this week! If you are also here, stop by and check out what we have been working on for Autopilot, Optimus, and dojo!
#CVPR2023
We have put forward two especially important proposals for our Annual Meeting of Stockholders—and we need your vote.
Protect your rights as stockholders & protect the value of your investment by voting FOR the ratification of the 2018 CEO Performance Award & FOR Reincorporating
it’s amazing what a small group of extremely talented and dedicated engineers can achieve in a short amount of time, join us if you want to be part of it.
When parking, you can now see a high fidelity 3D representation of the world around your vehicle, including proximity & shape of nearby objects, barriers, vehicles & painted road markings
By using a dedicated neural network to model obstacles & paint lines, we can accurately
Career Update: Today, I bid farewell to Waymo, marking the end of a chapter in my career.
Joining Waymo in late 2019, I entered a world where Level 4 (L4) robotaxi services were a concept rather than a reality. Now, in 2024, Waymo operates a rider-only robotaxi service in four
The 2023 Holiday Update rolls out next week
Here’s what’s coming...
Custom Lock Sounds
Replace the horn lock sound of your vehicle with another sound—like a screaming goat 🐐
LAN Party on Wheels
Play your favorite games on the rear touchscreen 🎮
Rear Screen Bluetooth
Record high deliveries and productions, beating wall st estimates, 36% YoY growth, all happening during a recession, but a failure according to
@Reuters
Hardware is hard.
That’s why Elon is by far the greatest founder of all time.
Remember — countless startups die just while trying to put stationary beige boxes on desktops. Very smart people get crushed by supply chain disruptions, or China tariffs, or lockdowns, or shipping
Tesla full self driving has made significant strides in the last year, and I am feeling very confident about this bet. I have made multiple two hour drives without touching the wheel.
I wouldn’t have wanted the task of leading a self driving effort for a couple reasons:
There
Tesla FSD 12 feels like a ChatGPT 3.5 moment. It’s not perfect, but it’s super impressive and you can tell it’s something very different. Can’t wait for the ChatGPT 4 equivalent.
great summery on the perspectives of scaling foundation models for robotics from both sides
my personal belief:
@tesla
will solve self-driving with the end-to-end learning approach, and it will provide a concrete data point and encouragement for the entire robotics community
There was a lot of good and interesting debate on "is scaling all we need to solve robotics?" at
#CoRL23
. I spent some time writing up a blog post about all the points I heard on both sides:
Below is inspired by
@Tesla
and
@philduan
who presented this at Telsa AI day, and the great FSD who team. Thank you for leading and showing the way. 🙏🏽
Had a great NBA talk with
@BigWos
while
@Tesla
FSD drove us around…
Timestamps
00:35 Bronny’s future
1:50 Lakers talk
2:55 If LeBron leaves…Philly?
4:50 Victor vs. Rudy for DPOY
6:40 FSD parks for tacos
7:30 Faith in the Bucks
8:35 Back to the office
“with a large enough network, a large enough labelled training set, and a large enough budget for GPUs, success is guaranteed.” -
@ilyasut
happy new year
may you be on the right side of the scaling law
great insight from
@ylecun
:
“the potential solution to [video understanding] is not generative. the models work the best do not generate images, they do not reconstruct, they do not predict.
what they do is they predict but in a space of abstract representation. what’s
super impressive work, making data collection much easier is going to be crucial for advancing robotics to the next level, kudos to
@chichengcc
and team
Can we collect robot data without any robots?
Introducing Universal Manipulation Interface (UMI)
An open-source $400 system from
@Stanford
designed to democratize robot data collection
0 teleop -> autonomously wash dishes (precise), toss (dynamic), and fold clothes (bimanual)
Heroes look like
@elonmusk
. They risk their own capital and that of those who voluntarily contributed it, in ambitious projects, without violence, toward grand human achievement.
Villains look like
@SenWarren
. They steal other peoples' capital to force their opinions on those
Inspecting large amount of data manually and visualizing the exact data fed into the network (after all the filtering, post-processing, etc.) is one of the best ML practices I’ve learned from
@karpathy
.
One pattern I noticed is that great AI researchers are willing to manually inspect lots of data. And more than that, they build infrastructure that allows them to manually inspect data quickly. Though not glamorous, manually examining data gives valuable intuitions about the
1/ Weekly 1:1s with direct reports are a staple of Silicon Valley management. The idea is to check in, see how they're doing, and provide feedback. I did this for 10+ years at Facebook & Dropbox. Frankly, I hated it and found it useless. But it's what "good" managers did.
“life needs to be more than just solving problems every day, you need to wake up and be excited about the future”
“our existence cannot just be about solving one miserable problem after another, there need to be reasons to live”
true for life, true for work
Top-tier product managers = Top-tier storytellers.
It's not just about solving a problem for the customer; it's about creating a compelling narrative for your team and your entire company to rally beneath. Your job isn't just execution, it's inspiration.
Particularly I want to highlight this new role we have for building foundation models for both Autopilot and Optimus. We are looking for exceptional AI researchers to join this effort. Please feel free to DM me for more details.
Only two options for competitors to keep up with Tesla solving self-driving using AI:
1) Make a vision-only system, invest billions of dollars in hardware and compute, and gather metric-shit-tons of real-world video with your own fleet.
2) License Tesla’s solution.
$TSLA
I spent a 5-month internship in Tesla's Autopilot team and it was such an amazing experience. I worked with super dedicated and talented people while solving one of the hardest robotics and AI challenges. The team is
@CVPR
and is hiring (
@philduan
)!
Incentive plans = cheat sheets for companies’ future performance.
🌟Success Stories:
Top performers like $TSLA and $PANW have stock performance as a key component in their incentive plans.
$TSLA's remarkable plan allowed Elon to earn options for achieving 12 targets linked to
it’s pretty crazy how linkedin decided to destroy their own product
below screenshots are some of the example posts dominating my feed:
1) reasons i should work remotely
2) linkedin is shutting down their office for next week
3) t-mobile had an amazing year and some ui tutorial
You should definitely chat with Phil & folks from Tesla if you are excited about large scale vision, robotics and more! They got a ton of data and compute to test your newest algorithm 👀
(Also wonderful people! Had a great time there last summer)