It's intriguing how people's opinions about Elon Musk's predictions remain unchanged, even when they come true.
Take, for example, his 2016 prediction [1] that the Model Y would sell 500,000 to 1 million units annually, four years before its launch and even before the Model 3 was released. This would make it the world's best-selling vehicle, not just the top electric vehicle. How do people perceive such audacious forecasts? And how do they react when these predictions materialize?
Similarly, when Musk accepted a compensation package based on Tesla's value increasing tenfold within ten years, the company was still unprofitable and losing money. Remarkably, Tesla achieved this goal in half that. How do people view these ambitious commitments and their successful outcomes?
Recently, Musk claimed that Tesla's autonomous driving technology alone could elevate the company to a $5 trillion valuation, and with the addition of the Tesla Bot, this could rise to $30 trillion. What are people's thoughts on these projections?
And will they ever reconsider their skepticism?
Yes, some of his predictions are late (like FSD and the Roadster), but personally, I prefer someone who eventually delivers the seemingly impossible over someone who neither promises nor delivers.
Please tell me another car that uses 48V for the whole electric system. Indeed there are other hybrid cars that use a 48V but only for the hybrid part. The actual benefit of Cybertruck's 48V system is that it lowered the amount of copper needed and simplified the whole system.
Then look again at steer by wire. Indeed, RZ450e seemed to have it last year in reviews as an optional packaged but now if I look on their website, there's nowhere to be found. Also, Toyota bZ4x stil doesn't have it, the plan is by the end of 2024. So yeah, not from 2022 like you said... And of course for rear steering it was always used, but the novelty in Cybertruck is the front steer by wire.
I also did look Automotive Ethernet and there are indeed a lot of informations about it, but again, tell me which cars have it for the whole system. There are indeed some like Hyundai that use it only for Entertainment system or VW for the driver assist. The main advantage of Cybertruck is that it's being used for everything which translates to lower complexity and lower cost.
News is about how the guy uses the device successfully almost non-stop while awake. And how even though there were some issues with the retracted threads, the accuracy increased massively compared to the initial one.
But the title of the article says the implant experienced a problem.
Fake? Did they say anything non-factual? Seems to me they just chose to lead with the one new bit of information in the update since last time that is surprising and that people would want to know about in any new medical device: problems. They don’t totally trash the device, but I’m sure even people considering volunteering would be interested in the potential hiccups.
The way I see it is what Waymo + others (and even Tesla until last year pre v12) was the same as someone doing ChatGPT but not with generative transformers but with if statements. If user asks what is that, say this, etc.
And now Tesla changed the game completely and moved to a 100% NN based system which indeed requires huge amounts of data which they're the only who actually have. All the other companies have no chance of getting that unless they fit their cars with thousands of dollars of gear for free (because I think no one would pay extra to get it and they really need it in ALL cars).
So it's probably not a good idea in terms of monopoly, but I think it will be Tesla FSD that gets integrated in all the other brands.
Because I would want to put a NN as unreliable as GPT4 in control of my car ... They are struggling to put in guardrails and make it politically correct with text.
I guess AI truly is the new crypto because I'm seeing all the same hype patterns. And Tesla is always riding the wave.
I've heard from several folks working on self driving outside of Tesla that they're very concerned about how cavalier Tesla is being about safety. I recall hearing one story about an engineer who outright refused to ride in a Tesla with FSD enabled. All anecdotal, sure, but I've become very skeptical of Tesla's approach.
How many things does it not talk about? One of the challenges of self-driving is that it has to fail safely so they have to handle all of these uncommon situations a chatbot can just give up on.
In my opinion short term things are getting worse indeed but in the medium future (10+y) it's going to start improving.
What is really weird is that the "experts" completely lack the ability to recognize exponential growth (S curves) and model everything linearly. Especially episode 5 (~3:15) shows how coal usage has been massively over-estimated and not just a couple of years but 15 years and it's still going. And similarly how renewable energy sources are underestimated massively. The estimated time for going to 10TWh of renewables is ~2100, but it will most likely happen in 2030 (same episode ~10:00).
Of course, nobody says to stop worrying because things are actually bad. But just from the perspective of cost all energy will migrate to renewables and after doing that and price starting to decline to 0, we will be able to look into other ways to fix climate (e.g. recapture CO2 which currently is cost-prohibitive).
I used to be scared, but lately I'm starting to change my mind.
I'm not saying it's not an issue, it's just that I'm very optimistic for the (near) future. And most of the scare comes from some very wrong estimates of expert's inability to recognise an S Curve of progress and not a linear one.
I recommend everyone to watch the Brighter series of videos from RethinkX [1]
Especially in episode 5 there's some graphs with every year's estimate of coal use and renewables that are really crazily wrong. I really can't understand how can "experts" be so wrong for tens of years and keep going.
It's already cheaper to install PV & Wind instead of any other fossil fuel alternative. And the price is only going down with scale. Also battery prices are going down so the problem of renewables intermittency will be solved as well.
Also, because of the fact that we need to have more power installed than consumption (because of intermittency) we will have more and more extra power (what Tony Seba from RethinkX calls Superpower or what we hear lately in news that energy price is negative). And with that what we can do is dump that energy into CO2 capture that was not feasible before because it was generating more CO2 with the power it used than what it captured.
This was on NASA's reference roadmap around the year 2000. First X-33, then something air breathing and finally flying saucers powered by ground and space lasers.
Exactly in the same spot, I've recently gone freelance where I'm paid by the hour and I've yet to take a holiday cause I always think this week of holiday will cost me $$$ (huge amount of money compared to what a normal one week holiday would cost).
Sadly, I will probably burnout soon and ignore that thought.
> I've yet to take a holiday cause I always think this week of holiday will cost me $$$
I'm a long-time hourly freelancer, and it's absolutely true - the lack of revenue for that period will indeed cost you money. But not taking the holiday will cost you a vacation. I personally think about the value lost from not working, and the value gained from the vacation, and usually conclude that taking the vacation is the clear winner. It's all about establishing a proper value system and being honest with yourself about it.
I used charge hourly years ago. I had the same problem where taking a holiday made me think about losing money.
Now I charge a monthly rate based on the number of days a week I work (roughly equating one month to 4 weeks, granted they're not but it makes life easier).
Now I don't worry about taking time off because once I've clocked my days for that week I'm done I won't get paid more. It also means you know exactly what your income will be each month which is handy for budgeting.
Start charging day, week and month rates, along with retainer rates. For projects and clients you're confident in satisfying, also consider fixed prices for deliverables, revenue sharing deals, etc.
I took this list to gauge how many hours I want to work per year, and decided that 1500-1600 hours is enough, if it’s not, then the rate should be higher.
Damn thats a nice list. Today I learnt that despite living in the USA, I have managed to arrange my life so as to work about as much as the average German.
What if adding a timeout of 690ms gets you a HR complaint for sexual harassment? Sadly, true story...