Elon Musk’s AI startup, xAI (x.ai = grok), officially acquired X Corp. (the parent company of the social media platform X or x.com, formerly Twitter/x.com) in an all-stock deal. Both now operate under a unified holding entity, frequently referred to in corporate filings as X.AI Holdings Corp. (or simply xAI). Now SpaceX has moved to acquire or merge with xAI. This effectively brings the social media data from x.com, the AI development of x.ai, and the satellite infrastructure of Starlink/SpaceX under one "super-conglomerate" roof.
To my western ears, the speech doesn't seem all that shocking. Over here it's normal for the CEOs of financial services companies to argue they should be subject to fewer regulations, for 'innovation' and 'growth' (but they still want the taxpayer to bail them out when they gamble and lose).
I don't know if that stuff is just not allowed in China, or if there was other stuff going on too.
I can see the extended loss of face of China (real or perceived) at the time being a factor.
Edit: So, after posting a couple of admittedly quite anti CCP comments here, let's just say I realize why a lot of people are using throwaway accounts to do so.
Brain cycles aren’t a limited supply. Besides, you’ll get to feel a nice jolt of serotonin when you remember to sit backwards.
> I would imagine crashes on these high speed lines (which always seem to be run at a higher level of professionalism than the general networks) are rarest of all
If this crash is anything like the other ones, you might be surprised. Safety complacency tends to cause maintenance failures. Plus the low speed lines are less deadly since the total energy is proportional to velocity squared, and v is low.
In other words, it might be more helpful to look at it as "if they’re run at a higher level of standards, it’s because they have to be".
Statistically you’re probably right, but considering how many brain cycles we waste on non-essentials, it’s just as fun to waste them on this. That way you can start a nerdy conversation with your travel companions, and they can learn to travel without you next time.
> Plus the low speed lines are less deadly since the total energy is proportional to velocity squared, and v is low.
You're forgetting about the probability of a crash.
The vast majority of train crashes is due to an impact with a vehicle on a railway crossing.
However, high-speed rail is grade separated, so it doesn't have railway crossings, which means the main cause of crashes is fundamentally impossible.
In other words: Regular rail has a high rate of crashes (with a small number of fatalities each) due to car/truck drivers screwing up. High-speed rail has a low rate of crashes (with a large-ish number of fatalities each) due to catastrophic failure of track & train equipment.
Zero-risk bias at work. If it’s actually fun for you, don’t let anyone stop you, but I wouldn’t go as far as making it a confident general recommendation.
If you’re interested in this kind of thing, look up plainly difficult on youtube. He has more videos on train crashes than I’ve seen, and I’m embarrassed how many I’ve seen. Here’s one to get you started: https://youtu.be/VV2rIHEp5AM?si=sSBT9s49PqbLTGbt
There are a lot of safety lessons embedded in these videos, which is why I like them. I also did a double take when I heard "semaphore"; its history goes back far longer than the ~century of software engineering. https://en.wikipedia.org/wiki/Semaphore
Oh you silly duck! Semafor is a common word in a handful of other languages for things like traffic lights and such. I had to do a double take when I first saw it in a programming class.
Also hope you’re doing well it’s been a minute since our paths crossed on gdnet.
"Semaphore" is (old) Greek and means "sign (sema) bearer (phore)", and actually the meaning in railways and computing is more or less the same: in computing, a semaphore signals if a resource is in use; in railways, the resource is a segment of a railway line, and the user is a train.
I disagree, but we are all entitled to our own opinions, and I get that there are a lot of luddites on HN these days. The fact that you consider it vomit rather than useful information just says more about you than me. If there was just a wiki page on how railway terms were used in computing, I would have just linked that (search didn't turn up anything in the first few pages).
I don't want to speculate on this crash but my mental model for these things is that there's always a handful of factors that all align and converge to create an accident. Some factors are deep-rooted - and point to decisions made years ago - sometimes related to company culture. Theres always an element of operator error: someone ignored something due to inattention or sleepiness.
Social? A lot of the bars/restaurants people go to in the morning for coffee/breakfast usually have news on the TV, and people usually talk with each other when big news happens.
This morning, big debates about what happened, whose fault it is, how safe/dangerous trains are, anecdotes from the past and jokes. Somber but lively discussions. Benefit is social cohesion with your neighbours and compatriots :)
It’s been said that RL is the worst way to train a model, except for all the others. Many prominent scientists seem to doubt that this is how we’ll be training cutting edge models in a decade. I agree, and I encourage you to try to think of alternative paradigms as you go through this course.
If that seems unlikely, remember that image generation didn’t take off till diffusion models, and GPTs didn’t take off till RLHF. If you’ve been around long enough it’ll seem obvious that this isn’t the final step. The challenge for you is, find the one that’s better.
You're assuming that people are only interested in image and text generation.
RL excels at learning control problems. It is mathematically guaranteed to provide an optimal solution for the state and controls you provide it, given enough runtime. For some problems (playing computer games), that runtime is surprisingly short.
There is a reason self-driving cars use RL, and don't use GPTs.
Some part of it, but I would argue with a lot of guardrail in place and not as common as you think. I don't think the majority of the planner/control stack out there in SDC is based. I also don't think any production SDCs are RL-based.
Control theory and reinforcement learning are different ways of looking at the same problem. They traditionally and culturally focussed on different aspects.
I feel like both this comment and the parent comment highlight how RL has been going through a cycle of misunderstanding recently from another one of its popularity booms due to being used to train LLMs
While collecting data according to policy is part of RL, 'reductive' is an understatement. It's like saying algebra is all about scalar products. Well yes, 1%
RL is still widely used in the advertising industry. Don't let anyone tell you otherwise. When you have millions to billions of visits and you are trying to optimize an outcome RL is very good at that. Add in context with contextual multi-armed bandits and you have something very good at driving people towards purchasing.
More likely we will develop general super intelligent AI before we (together with our super intelligent friends) solve the problem of combinatorial optimization.
There's nothing to solve. The CoD kills you no matter what. P=NP or maybe quantum computing is the only hope of making serious progress on large-scale combinatorial optimization.
I like to think of RLHF as a technique that I, as a student, used to apply to score good marks in my exam. As soon as I started working, I realized that out-of-distribution generalization can't be only achieved from practicing in an environment with verifiable rewards.
The parent’s advice is toxic and mistaken. It’s a road to codependency. I’ve been with my wife 20 years, married 15. I would have said the same thing they said — I can’t do it all on my own, I need someone else.
Rubbish. And also dangerous rubbish. I’ve been weak for a long time simply because I hadn’t taken myself seriosuly. I literally believed that I couldn’t do it alone, which was wrong.
It was unfair to my wife to use her as an emotional support when she didn’t want to be. She’s been there for me a lot over the years. But when you tell someone that you can’t do it without them, it’s no longer their decision, and that’s unfair. Both to her and to me.
Please read Codependent No More, and especially Lost in the Shuffle by Subby. (I’ve identified a lot more with the latter.)
The point is, it’s okay to be having a rough time with your wife. Let go. Let her do her own thing. Stop caring so much. It’s okay for her to be upset and not want to help/have sex/go to an event/involve you/whatever the problem may be. The reason it feels rough is because you personally let it feel rough. Once I adopted that mindset, it became so much easier. And ironically my marriage improved.
Meds are also important. Make sure you’re on a good dosage of antidepressants if you need them, and a mood stabilizer. I recently started Latuda and dropped Seroquel per my psychiatrist, and it’s been night and day.
Lastly, keep trying to talk to people about your problems. I ended up reaching out to a random person on Twitter. They were kind and to my surprise happy to listen. It was one of the main reasons I was able to get through it all. The best person to talk to is a therapist, though I’d be happy to listen till you can find one.
You’re strong. You need to believe that. And you’re strong independently of your family or anyone else. Give yourself credit for getting as far as you have; that part has been important too.
Apparently “codependency” means something significantly different to what I guessed (which was interdependency, depending on each other). https://en.wikipedia.org/wiki/Codependency:
> In psychology, codependency is a theory that attempts to explain imbalanced relationships where one person enables another person's self-destructive behavior,[1] such as addiction, poor mental health, immaturity, irresponsibility, or under-achievement.[2]
> Definitions of codependency vary, but typically include high self-sacrifice, a focus on others' needs, suppression of one's own emotions, and attempts to control or fix other people's problems.[3]
I've heard codependency described as being dependent on others being dependent on you.
It probably has some rational basis in child rearing, it will benefit survival for the parents to be deeply dedicated to supporting their children, to the point where they receive a psychological reward from that dependency. But unhealthy when it comes to adult relationships, at least beyond a certain point.
There is a huge difference between acknowledging that humans are an inherently social species that usually needs comfort and psychologically benefits from an intimate relationship and straight up codependency, where you violate the boundaries of each other and thereby take away psychological safety.
I agree! The point is, don’t use your wife for your comfort and psychology benefits. Use the other people in your life. Especially when you’re having marital problems.
Just want to point out codependency--especially if you read Codependent No More--is not about being dependent on another person. That is dependent personality disorder perhaps.
Codependency is better described IMO as secondhand addiction. It was coined to describe the symptoms of people who live with alcoholics and other substance abusers and the destructive coping patterns they use to survive in the addict's wake. The codependent does not depend on the addict. In fact closer to the opposite.
Upvoted just for mentioning the book though. It was life changing for me.
I get what you're saying. A therapist is one of the types of people I had in mind, although that obviously isn't an option for everyone.
I agree that it's important to be able to have your own independent autonomy to properly function in a healthy relationship, especially a romantic one.
The point I was trying to make is perhaps more subtle than it came across, namely that webs of trust between humans (e.g. 'community') are, in my view, essential to being a fully actualized adult. If you aren't close to anyone, I think that means something is wrong which deserves further inspection, particularly within yourself.
Sorry for the somewhat harsh words. You have a point. The problem is that it’s way too easy to fall down the codependency rabbit hole when you start thinking of it as “I can’t do X unless someone else Y’s”. It was true for me, and I just wanted to make sure it wouldn’t be true for the poor fella going through marriage problems.
The trick and the trouble is that it’s easy to acknowledge the importance of being independent, especially in a romantic relationship, vs actually doing that in practice. After your 30’s your friends start to fade away, and one day I woke up without any except my wife. That was clearly a degenerate situation unfair to her, and expanding your social circle is something that should be done independent of whatever relationship you happen to be in. In fact, needs to be done.
Another complication / nuance is that, of course you should be serving and supporting your partner (and vice versa)! Its part of what makes relationships rewarding. They're not always 50/50 in all aspects.
The trick, as you say, is to know when that is crossing into something unfair. When it goes beyond something like who does the dishes or makes the most money into supporting the other person's core identity. Or when it becomes unsustainable / exhausting for other person. Identifying these issues can be difficult. It requires both partners to be in touch with their feelings and able to communicate openly.
