To all of my U.S.-based subscribers, I hope that everyone had a wonderful Thanksgiving yesterday. And to all my subscribers outside the U.S., I hope you’re having a wonderful time with family and friends as we slide into the end-of-year holidays and festivities.
It’s incredible how quickly this year has gone. There is so much happening right now. The pace of technological advancement is unlike anything I’ve ever seen. And we’re in for an even more torrid pace in 2026.
We’re grateful for all of you, our subscribers, for joining us on this incredible journey that is leading us to artificial general intelligence next year, artificial superintelligence a few years after that, a return to the Moon, expeditions to Mars, and an abundance of cheap, clean energy to fuel it all.
It’s all happening in the next three to four years. Hard to believe, but it is…
Thank you for empowering us to carry out our mission of being the very best and broadest source of investment research and analysis in high tech on the planet. Without you, we wouldn’t exist.
Happy Holidays,
Jeff
How does Tesla deal with snow and ice when vision is impaired?
– Jeff O.
Hi Jeff,
This is definitely a question that comes up often when we talk about Tesla’s vision-based autonomous driving technology. And it’s a fair one. After all, how can a vision-based self-driving vehicle navigate the kind of weather conditions that make it nearly impossible to “see”?
As it turns out, the same way it handles pretty much any other road conditions… as well or better than most human drivers.
In one of the last AMAs where we tackled this question, my answer was more or less that if a human driver can safely navigate the road in whatever weather condition, so can my Tesla. And that was two significant full self-driving software upgrades ago, back in August 2024.
I recently covered a very recent version of Tesla’s FSD – v.14.1.4 – in The Bleeding Edge – Get Ready to Ride.
As I said in that issue, it’s just incredible…
I don’t know what else to say other than it is jaw-droppingly good. It’s absolutely stunning, just about flawless, extremely safe, and, in many regards, so very humanlike in its driving style.
It’s like a professional driver without any erratic driving or braking. Everything is precise, measured, and smooth.
What changed? There were clearly some major improvements to the performance of the neural network – the brain – of the full self-driving system.
It is obvious that Tesla made extraordinary efforts to train the FSD AI on fringe cases, outliers, what we might think of as the 0.1% of tail events that a vehicle might see only once in 100,000 miles.
Examples are things like what to do when a car from oncoming traffic crosses the double yellow lines into your lane. Or how to avoid a large piece of lumber in the middle of the road. Avoiding sideswipes from a reckless driver in an adjacent lane. Or a small animal running across a road. How to deal with a deer in the headlights. The list goes on and on.
The sort of circumstances humans encounter – and struggle with – on the road all the time, despite accounting for a very small amount of the overall driving experience.
Even before the most recent upgrade, though, I had handed over control to my Tesla out in snow, pouring rain, and fog, icy road conditions, and the performance of full self-driving was no different. Despite low visibility, the car had no problems navigating the roads.
And if the car did encounter driving conditions where the vision-based system couldn’t safely “see,” the FSD software simply indicates that we need to take the wheel. I’ve been driven in some torrential rain, as well as very heavy snow, where my own vision was hampered, but my Tesla had no problem at all.
The key point for us to keep in mind is that a Tesla has 8 exterior high-resolution cameras that provide a 360-degree, real-time view around the car. Our human eyes have strong vision for about 120 degrees and can perceive movement/objects to about 200 degrees.
My point is that the Tesla’s vision is far superior to our vision. Which is a way to say that if there ever was a circumstance when the FSD software couldn’t “see” and asked us to take over the wheel, it means that we shouldn’t be driving either.
I might be a bit crazy, but anytime the weather is bad, I jump into my Tesla to test out the latest version of FSD software. In fact, just yesterday, I received the download for the latest version 14.2.1.
You know what I’ll be doing the next few days…
Dear Jeff,
Thank you for the daily newsletter! The Bleeding Edge is the best free source for the latest information in tech I have found; you and your team do a great job with it.
In a recent Bleeding Edge issue, you said that Wikipedia is “heavily biased and often completely incorrect.” You also said to “Immediately stop using Wikipedia,” and to use Grokipedia as your “only source of an online encyclopedia.”
Often, Grokipedia copies Wikipedia articles almost word for word. It has also shown a tendency to cite unreliable sources, such as social media posts or personal blogs. Even ignoring this, the general consensus is that if Wikipedia can be considered to be a moderate left-leaning source, Grokipedia is at least as far right-leaning.
With all these issues, wouldn’t the best thing be to read both in tandem, rather than to “immediately stop using Wikipedia”?
The best way to find ground truth is to examine multiple perspectives through a critical lens before making a conclusion.
– Reed O.
Hi Reed,
You are correct, Grok has analyzed the contents of Wikipedia. Not all information on Wikipedia is inaccurate, as many topics aren’t used to push radical ideology. Grok uses the information and other widely available information to check and confirm the accuracy of information.
If something is inaccurate, unsubstantiated, or unbalanced, Grok will adjust. If it is all good, it can use the language.
As an exercise, I recommend that you look up “Climate Change” on both Wikipedia and Grokipedia. Read them both to understand how radically different the two responses are. Wikipedia is heavy with political narrative and fear, and Grokipedia is balanced, measured, and far more scientific.
Last month, Larry Sanger – the founder of Wikipedia and a libertarian – went public about how upset he was about how politicized and far left-leaning the leaders of Wikipedia had become.
He has been very vocal about how far Wikipedia has strayed from his original vision of having Wikipedia be a neutral source of truth on the internet.
He published a long diatribe on how to fix Wikipedia with “Nine Theses” as action items to address the horribly biased problems at Wikipedia. I’ve posted them below, and if you are interested, you can read Sanger’s thoughts on this issue, which I highly recommend, right here.
So, to answer your question directly, no, I could never recommend that anyone use Wikipedia. As per Sanger’s own position, critical information has been blacklisted, and there is no neutrality.
I also wouldn’t use Wikipedia in my research and analysis for my own subscribers. It would be a disservice to my subscribers to rely on such an unreliable and biased source of information.
It is also worth mentioning that Grokipedia is only on version 0.2 right now. Having a neutral, balanced source of information is not far-right. It is just neutral and balanced. It is also important for us to remember that by design, Grok’s underlying architecture is designed to be a maximum truth-seeking general intelligence.
That’s it, nothing more, nothing less.
And Reed, I completely agree with you. When analyzing or thinking about any topic, it is extremely useful to use multiple sources, look at it from several angles, and apply critical thinking to determine what is most accurate based on all information available at any given time.
This is precisely what my team and I do at Brownstone Research day in and day out.
The world is about to change radically next year, the year that I predict will be the year of artificial general intelligence (AGI). The implications are profound, and something that I have been covering and will continue to cover in The Bleeding Edge, and all of my other research and analysis.
Grokipedia is on an incredible trajectory despite only being released on October 27. It is literally just a month old, and it is already the best source of encyclopedic information. Just imagine how much better it will be a year from now.
And for those that don’t like the name, when it is “good enough,” it will be named Encyclopedia Galactica.
And that will be good for all of humanity.
As a member of Exponential Tech Investor, I am wondering if Jeff has been exploring graphene, the companies in the industry along with potential uses?
If yes, what are his thoughts, and does he have any guidance?
– Jeff L.
Hi Jeff,
Thanks for writing in. Yes, graphene is a popular topic for Brownstone readers and a material that I’ve been researching for over a decade.
I’ve covered it a couple of times in AMAs past, and it tends to pop up regularly in our feedback file. It’s understandable, given how often it is mentioned as a potential superconductor for various AI applications.
Very quickly, for newer readers, graphene is a highly conductive, pure carbon material that is often positioned as a breakthrough material for semiconductors due to its incredible conductive properties.
And true, graphene is incredibly conductive and can handle extremely high temperatures, making it attractive in terms of heat management in semiconductors.
The problem isn’t about what graphene is capable of, but the ability to manufacture and commercialize products made from graphene.
It’s simply not a good solution when we look at things from an economic standpoint, or when we take into full consideration some of the technical issues semiconductor manufacturers would encounter if they leaned into graphene-based chips.
It’s simply too expensive, too difficult a material to commercialize, and too energy-intensive for broad application in the industries where we’d see the greatest use cases for it.
If you’d like to read more about it, I encourage you to visit a few AMAs from our Bleeding Edge archive, where we discuss graphene:
I expect we’ll continue to hear about graphene as a potential “wonder material” for artificial intelligence, specifically in semiconductor applications. It’s a fascinating material.
But I would caution readers to be very skeptical of any companies that are soliciting investment claiming that they have cracked graphene and have a “wonder product” set to dominate an industry.
At the moment, graphene is simply not economically viable for most applications in industry. As I predicted in a recent AMA, I believe it may see some use in cases where costs aren’t really a concern. For example, defense applications where cost isn’t an issue, only performance.
But as always, my team and I have our eyes and ears on the materials industry and will be quick to let Bleeding Edge readers know if there are any changes or updates in the space.
And believe me, when there is a new material that is economically viable and can scale in terms of manufacturing, I’ll be pounding the table at Brownstone Research.
Jeff
The Bleeding Edge is the only free newsletter that delivers daily insights and information from the high-tech world as well as topics and trends relevant to investments.
The Bleeding Edge is the only free newsletter that delivers daily insights and information from the high-tech world as well as topics and trends relevant to investments.
