Former Tesla FSD Leader, Karpathy, Offers Glimpse into the Future of Artificial General Intelligence

Karpathy on AGI

Not a Tesla App

Andrej Karpathy, former Tesla Autopilot Director, offers a fascinating perspective on Artificial General Intelligence (AGI) by drawing parallels with the evolution of self-driving technology. As the world grapples with the implications of AGI, Karpathy suggests that the trajectory of autonomous vehicles provides valuable insights into what AGI might entail and its impact on society.

The blog post has since been deleted, but you can still check it out here; Karpathy defines AGI as an autonomous system exceeding human capabilities in most economically valuable tasks. This definition hinges on two criteria: complete autonomy and broad economic applicability. According to Karpathy, the journey of self-driving technology exemplifies the societal dynamics of increasing automation and, by extension, AGI's potential evolution.

The Gradual Rise of Automation

Self-driving technology stands out due to its high visibility, significant economic footprint, large human workforce, and the complex challenge of automating driving. Unlike other sectors that have seen automation, the path to self-driving is a prime example of AGI's characteristics: accessibility, economic importance, workforce impact, and technical challenge.

Karpathy outlines the gradual development of driving automation. Initially, vehicles featured Level 2 driver assistance, where AI collaborates with humans in navigation, handling many low-level driving aspects while allowing human intervention. This partial automation is analogous to AI tools in various industries, like GitHub Copilot in programming, highlighting the incremental nature of AI advancement.

The leap to full automation, as seen in Waymo's driverless cars, marks a significant milestone. In cities like San Francisco, Waymo offers autonomous rides in a small, geo-fenced area, however, it helps showcase a future where AI will surpass human driving abilities. The transition to full autonomy will depend on public awareness, trust, preferences, and supply constraints in creating a large automated fleet.

Global Expansion: Challenges and Opportunities

The globalization of full automation, Karpathy notes, is a gradual, resource-intensive process. Waymo's current limitations to specific cities illustrate the challenges of expanding automated services, including adapting to local conditions and regulations. This expansion mirrors the broader trajectory of AGI deployment across various sectors, where scalability is both achievable and gradual.

Society's reaction to self-driving technology parallels AGI's potential impact. Despite significant advancements, public awareness and acceptance vary. Some view autonomous vehicles with curiosity and skepticism, while others adapt quickly. This range of responses suggests how society might adapt to AGI in various industries.

Economically, self-driving technology has both eliminated and created jobs. While driver roles are phased out, new positions in data labeling, remote support, fleet maintenance, and sensor technology emerge. This transformation reflects the broader economic implications of AGI, where work is not merely eliminated but refactored and reshaped.

The competitive landscape in self-driving technology, with companies like Waymo, Tesla, and others, mirrors the expected consolidation in AGI-related industries. As with self-driving, only a few companies may dominate the AGI space after an initial burst of growth and competition.

Karpathy envisions AGI as a gradual, society-involved evolution rather than a sudden, uncontrollable leap. Just as self-driving technology is transforming transportation, making it safer and more efficient, AGI promises to reshape various sectors.

Follow

Follow

Follow

What’s Coming Next in Tesla FSD V14

Not a Tesla App

With FSD V13.2.6 continuing to make its way to AI4 vehicles, Tesla has been on a streak with minor FSD improvements since the launch of FSD V13 just a little over two months ago. 

FSD V13 brought a new slate of features, including Start FSD from Park, Reverse, and Park at Destination. It also introduced full-resolution video input using the AI4 cameras at 36hz and made use of the new Cortex supercomputer to get faster and more accurate decision-making.

So, what’s next with FSD V14? Tesla gave us a sneak peek at what’s next for FSD.

FSD V14

The standout feature of FSD V14 will be auto-regressive transformers. While that’s a complex term for those unfamiliar with AI or machine learning, we’ll break it down.

Auto-Regressive

An auto-regressive transformer processes sequential data in time, using that information to predict future elements based on previous ones. Imagine completing a sentence: You use the words already written to guess what comes next. This process isn't just about filling in the blank; it's about understanding the flow of the sentence and anticipating the speaker's intent.

FSD could analyze a sequence of camera images to identify pedestrians and predict their likely path based on their current movement and surrounding context. The system's auto-regressive nature allows it to learn from past sequences and improve its predictions over time, adapting to different driving scenarios.

Today, FSD reacts to what it sees, but soon it’ll be able to anticipate what will help, much like humans.

Transformers

The second part of that term is transformer, which is a component used to understand the relationships of elements inside a time sequence. It identifies which parts of the input are most crucial for making accurate predictions, allowing the system to prioritize information much like a human would. Think of it as weighing different pieces of evidence to arrive at a conclusion. For example, a transformer might recognize that a blinking turn signal is more important than the color of the car when predicting a lane change.

Putting It Together

Putting all that together, Tesla’s use of auto-regressive transformers means they’ll be working on how FSD can predict the plans and paths of the world around it. This will improve FSD’s already powerful perception and allow it to predict how other vehicles and vulnerable road users (VRUs) will behave.

What it all comes down to is that FSD will be able to make better decisions and plan its paths by making more informed, human-like decisions. That will be a big step towards improving V13 - which already has some very effective decision-making.

Larger Model and Audio Input

Tesla stated that FSD V14 will see larger model and context sizes, which coincidentally are listed in the upcoming improvements section of FSD V13.2.6.

Interestingly, Tesla says that context size is restricted by AI4’s memory, so it’ll need to be optimized. Tesla is already gathering audio data, and FSD V14 will be the first version to take advantage of it.

FSD V14 Release Date

We haven’t heard from Elon Musk or Ashok Elluswamy (Tesla’s VP of AI) about when FSD V14 will arrive. Ashok previously stated that FSD V13.4 would see audio inputs being used, but at Tesla’s earnings call, Musk said that audio input would become relevant in V14. It seems like Tesla may have scrapped V13.4 for V14.

Since Tesla is planning to launch their Robotaxi network in Texas this June, FSD V14 may be the version used for their autonomous taxi fleet.

Follow

Follow

Follow

Want to See the New Tesla Model Y? Here's Where It’s on Display

Not a Tesla App

Tesla’s refreshed Model Y has started hitting display and Tesla showrooms across North America and Europe, where you can now get up close and personal with the new Model Y and all of its improvements, including the smallest of details.

The refreshed Model Y surprised all of us by sporting its own unique look instead of just being a stretched-out Model 3. While it’s available to order now through the Launch Edition, which includes FSD and Acceleration Boost, deliveries won’t start for another couple of months.

The new front light bar and rear indirect lights make the vehicle unique and worth checking out. There are plenty of new items to check out in the updated interior as well, including new door panels, powered rear seats, and more.

Here’s where you can check out the new Model Y in North America and Europe:

North America

In North America, the updated Model Y can be seen in 19 showrooms across the United States and Canada. The photo below shows the 17 locations in the U.S. and two in Canada. We expect this list to grow as Tesla produces more vehicles and we get closer to the official launch.

New Model Y Displays in North America

Not a Tesla App

Europe

In Europe, the Model Y is already on permanent display at two locations - the Mall of Berlin in Germany, and at Tesla’s Amsterdam Europe Headquarters - both publically accessible. If you’re in the area, definitely drop by and take a look.

Tesla is also holding special events throughout Europe where they’ll be showing the new Model Y through select dates. place to check if you’re far away from the ones above is on Tesla’s Events List.

We recommend checking Tesla’s page to confirm any dates, as Tesla will be adding new locations until each showroom receives its permanent display vehicles.

Tesla will also be visiting major auto shows, including the Toronto and New York International Auto Shows with the new Model Y - so keep an eye out for those events as well.

Follow

Follow

Follow