We’ve shown plenty of very questionable Tesla Full Self-Driving (FSD) Beta videos a number of times before, and while I think that most of you, our athletic, witty readers, pretty much get the idea of what the reality of the state of FSD Beta is, I keep posting them because certain people from Tesla itself keep making claims about the system, claims like how it will achieve “full self-driving this year.” The person who claims this has a lot of reach online, so I think it’s worth continuing to put out some real-world reminders of where FSD Beta currently operates.
Now, it’s also worth noting that you can find plenty of videos of FSD driving in what appears to be a very competent manner, with few or no disengagements, and I’m happy to link to a couple of those, if you’d like to see that.
But the truth is that, while its nice to know that there are times and situations where it’s working well, people actually need to see the times where it doesn’t make the right decision, because these are the times that the limitations of the system are actually revealed, which helps people really understand what they’re dealing with: ALevel 2 system that requires constant, ready vigilance.
Complacency, and attributing too much competency and confidence to a very unfinished system is potentially far more dangerous than just, you know driving normally with the wet, pliant FSD (and other functions, like Wordle playing and making you horny/hungry) computer in your head and the two moist cameras swiveling in your skull.
So, with that in mind, here’s this recent video of a 4.5 mile drive in Boston:
If you’re curious about the route, the driver/Tesla owner, Taylor Ogan, provided a map of the route, which contains pretty normal city driving:
The problems seen in this video are not confidence-inspiring at all, and many of them are mistakes that, quite simply, no sober, non-simpleton human driver would make. Right at the beginning of the video, the Tesla attempts to steer into oncoming traffic lanes by turning on the wrong side of a truck.
Then, it does not appear to attempt to slow down for a huge red-and-yellow truck turning in front of it. Then, for no clear reason, the car attempts to steer again into the oncoming traffic lane, seemingly to try to pass the cars in front of it? It’s not clear.
It does seem to slow down effectively for a cyclist, though.
There’s more: Confusing turns, close calls, odd and dangerous choices. This is a far, far cry from anything that can be considered close to finished and is not even close to as good as even a pretty crappy human driver.
I don’t think it’s being too pessimistic to say that I’m not buying Elon Musk’s claim that it’ll be ready this year.
There’s been some interesting and revealing responses to the video, like this one from a prominent and quite orthodox Teslarian:
Even if we look past that confusing and deeply questionable corruption of Ogan’s name in there, the content here is revealing. Omar Qazi, the man behind the Whole Mars Catalog twitter account, isn’t actually questioning the content of the video but rather lamenting that Ogan has access to FSD at all.
This is, of course, a troubling way to think, as it implies that FSD should only be in the hands of those who are willing to always give it the benefit of the doubt and show only its successes online — a process that wouldn’t benefit the driving public and could be considered deceptive, even.
There’s also this sort of reaction:
Here’s the thing about the idea that this video could be misleading, or that the driver was somehow doing something wrong. First, how? The car is, ostensibly, doing the driving, with the person behind the wheel just watching and waiting to intervene if necessary. There’s no way — aside from some pretty elaborate and expensive movie magic — to fake what’s being seen here. The car made those bad choices.
And, as far as editing the video to show the bad moments, well, sure. But that doesn’t change that those moments happened and are important. Ogan clearly states it was on a 22-minute drive, but even if it was from a two-hour drive or a four-hour drive or whatever, everything seen here would still suck just as much.
What’s kind of the most frustrating part here is that there is some remarkable tech being seen here. The problem is with how it’s being presented, marketed, sold, and introduced to the public.
If this wasn’t a $12,000 option being sold for many real dollars long before it’s ready, if it wasn’t being deployed in pubic roads with amateur “safety” drivers, if it wasn’t consistently hyped up well beyond its actual abilities, then it would be a great R&D effort that could lead to something really remarkable.
But, that’s not how it’s being handled, and not where we are. Instead, we have unfinished software on public roads driving like an idiot while people online grow enraged and imagine dark reasons for why anyone would point out the obvious.
We’re in weird times.