"For example, the current cities Waymo operates in do not have appreciable snow fall, and as a result neither the Waymo nor the human benchmark data include this type of inclement weather."
I'm happy to see this acknowledged, and hope it's a sign that they appreciate the difficulties of winter driving.
It’s wild to me how people are so fixated on this. Yes, obviously winter driving has challenges. And also obviously, the leading self-driving company has thought about that.
They’re preparing to launch and have already been testing in Chicago, detroit, Minneapolis, Denver, Philadelphia, Boston, NYC, and London. I think it’s safe to assume they’ve considered winter driving.
Ooh, that's a worthy challenge. Of course, I can imagine getting enough data on all of those cities and deciding to launch everywhere else but not Boston "because your roads are garbage and you all drive like you're impaired 24/7" :-)
They’re in Denver now (not available to the public yet) for this exact reason. Unfortunately we’ve had our least snowy winter in recorded history this year - so they only got a few days of real snow.
When was the last time you worried about someone cutting your brakes? A lot of times these hypothetical fears are disconnected from reality. Security is important, but people generally don't engage in destruction for destruction's sake so improving default safety levels has been a clear net positive for society so far. Maybe I'm being shortshighted and a future security exploit will change that, but it's not something I currently fear as someone whose car gets occasional OTA updates.
Cutting someones breaks requires physical access to the hardware.
Changing:
if (brakeDepressed()){
engageBrake();
}
To:
if (brakeDepressed() && currentTime < '5/6/26 4pm EST'){
engageBrake();
}
Can be deployed to thousands of vehicles, and would stop brakes from working during peak commute time on the East Coast.
Someone who can write out that code with that specificity should know there are countless technical and procedural ways to help prevent that sort of thing from actually making its way into consumer vehicles (or that OTA updates would be the only avenue to accomplish that). In a properly designed system, the only real fear here is a state-level attack. And I just don't think getting every Honda to crash at 4pm is a vulnerable enough attack vector to make this hypothetical worthy of much thought.
How do you know that a car is the result of a properly designed system before you get behind the wheel (or step in front of it?).
>the only real fear here is a state-level attack
Why isn't this a valid concern? We should just be fine with russia or china having the ability to remotely hack all of our cars and kill/spy on individuals, even critical members of our leadership? What about our own government? What about some terrorist launching formerly state-level malware from his basement with the help of AI?
Not only state actors. Vulnerability can be exploited by non-state actors. A terrorist getting hold of this capability to crash every Honda at 4pm introduces new challenges. The impact of 9/11 was not about how many people were killed. But it terrorized the population with that act. People stopped getting into flights. Imagine similar stuff with our daily routine cars.
State level actors have plenty of money to find any exploit around those protections and some need little incentive. They can hire a spy to cut my break line but their gain is much lower vs the cost. They don't care about me at all anyway except if I'm in a group of 100k people they can get at once.
> the only real fear here is a state-level attack.
This is blatantly false. In the real world there was a major recall after security researchers (not state actors) demonstrated that they could remotely interfere with safety critical systems. OTA updates without user involvement are a massive security vulnerability. So are internet connected safety critical systems. Neither should be legally permissible IMO.
> I just don't think getting every Honda to crash at 4pm is a vulnerable enough attack vector to make this hypothetical worthy of much thought.
Setting aside assassinations do you just have no imagination? There have been all sorts of crazy disgruntled worker sabotage stories over the years. Mass shooters exist. Political and religious terrorists exist.
For a specific mass scale state level hypothetical imagine that the US enters a hot war with a peer adversary for whatever reason. The next day during the morning commute the entire interstate system grids to a halt, the hospitals are completely overwhelmed, and the entire supply chain collapses for a week or so while we pick up the pieces. With a bit of (un)luck it might happen to catch an oil tanker in the crossfire while it was in a tunnel thereby scoring infrastructure damage that would take years to fix.
> should know there are countless technical and procedural ways to help prevent that sort of thing
Sometimes when I look at code it feels like I was led into a weird surprise party celebrating structure and correctness, only for everyone to jump out as soon as I get past the door to shout, “Just kidding - it’s the same old bullshit!” All that to say, we’re about as good or worse as anyone else, at our respective jobs.
> A lot of times these hypothetical fears are disconnected from reality.
Conversely, a lot of times people don't fear real dangers of reality until it bites them. "Hackers wouldn't care about me, and the single password I use on every website is super good and complicated."
> but people generally don't engage in destruction for destruction's sake
Generally true, but they do engage in destruction when there's profit to be made or when it becomes in their geopolitical interests, and sometimes that destruction is quite notable: Remember when it was safe to assume that passengers could passively wait out airplane hijackings?
Your average script-kiddie might not seriously consider cutting everyone's brakes simultaneously, Al Queda would have been giddy.
Software has an atrocious track record for security. Doubly so for hardware manufacturers. It only takes one smart cow to disable millions of vehicles vs a local knave cutting brake lines.
I yearn for the days of wrapped software where developers had to make a gold pressed release. Not, “we can patch it later”.
If you want to talk about society, then this is about systematic security not individual security. If someone somewhere can push a button and flash your car with OTA firmware to drive you off a bridge, political assasinations become a lot easier.
In fact, with all this data they are collecting, you wouldn't even need to be the next edward snowden to get this treatment. You could set the firmware to target, say, every left-wing voter in america.
You don't even need the own the car with such behavior. Everyone becomes a pedestrian eventually.
Could you elaborate (hopefully with real examples) of what it's like to be in the out group with few connections (or perhaps no connections) in regards to a particular good / service?
Then you get worse good and services. Lower quality or longer wait, or don't get it at all depending on the good. The effect isn't that different from being poor in a capitalist economy. In a capitalist economy, it's mostly money that determines what you can buy. In the Soviet blat-heavy economy, money didn't matter as much connections.
It was perfectly possible to have a decent salary but nothing to spend it on because the better items just aren't available. Maybe there's some delicacy you enjoy, or a special item you want like a cassette player and you could afford those if the store actually had them, but they don't. In that situation, your ability to buy more desirable items depended more on your connections or perseverance in doing things "the hard way" like queuing for hours to buy bananas, or recycling enough kilograms of paper to buy a book.
The Canadian Charter of Rights and Freedoms was established in 1982. We're still in the process of figuring out what it means (and as a living document, the interpretation will change over time).
It's messy. But I'd much rather that than need to ask "What would Pierre Trudeau think of this situation?"
The "bre" in "libre" is pronounced similarly to "zebra". Kinda. It'll get you in the ballpark, which is good enough for an Anglo.
"This Hour has 22 Minutes" had a great sketch where both a Francophone (Gavin Crawford impersonating Chantal Hebert) and an Anglo (I forget who) were stumbling over proper nouns from the opposite language. The joke was that both were trying too hard to pronounce things "properly". It came off as inauthentic and awkward.
The GPUs, sure. The mainboards and CPUs can be used in clusters for general-purpose computing, which is still more prevalent in most scientific research as far as I am aware. My alma mater has a several-thousand-core cluster that any student can request time on as long as they have reason to do so, and it's all CPU compute. Getting non-CS majors to write GPU code is unlikely in that scenario.
I provide infrastructure for such a cluster that is also available to anyone at the university free of charge. Every year we swap out the oldest 20% of the cluster as we run a five year depreciation schedule. In the last three years, we’ve mostly been swapping in GPU resources at a ration of about 3:1. That’s in response to both usage reports and community surveys.
Insane question, asked for the purposes of discussion: Would it make sense if those GPUs were top-of-the-line for years? Like if TSMC were destroyed?
Even then, I don't understand why being a landlord to the place were AI is trained would be financially exciting... Wouldn't investing in NVIDIA make a lot more sense?
> I'm sure there's alien civilisations that are more aggressive than us, but also ones that are less so.
What is the minimum amount of aggression necessary to evolve sentience? What is the maximum amount of aggression in an interstellar space-faring species? Where is humanity on that scale?
A super-aggressive species would likely self-annihilate before possessing sufficient energy to travel interstellar distances... So the jury's still out on us.
There are definitely categories of code where you could realistically expect lift-and-shift from C which you're confident is correct to safe Rust that's maybe not very idiomatic but understandable.
I believe Microsoft has a tool which did this for some bit twiddling crypto code. They have high confidence their C crypto code is correct, they run a process, they get safe Rust, and so their confidence transfers but now it's safe Rust so it just works in Rust software and you get all their existing confidence.
But it's never going to be all code, and it might well never be most code.
I'm happy to see this acknowledged, and hope it's a sign that they appreciate the difficulties of winter driving.
reply