No. Absolutely not. The government owes its people a certain duty of care to say “just because you can doesn’t mean you should.”
LLM’s are good for advice 95% of the time, and soon that’ll be 99%. But it is not the job of OpenAI or any LLM creator to determine the rules of what good healthcare looks like.
It is the job of the government.
We have certification rules in place for a reason. And until we can figure out how to independently certify these quasi-counselor robots to some degree of safety, it’s absolutely out of the question to release this on the populace.
We may as well say “actually, counseling degrees are meaningless. Anyone can charge as a therapist. And if they verifiably recommend a path of self-harm, they should not be held responsible.”
I want this same analysis with more nuance about what negativity means. He mentions in the post that “technical criticism” counts as negativity.
There’s just a world of difference between “I don’t like React because I don’t want to write HTML in my JavaScript” and “React sux a$$”
Both are negative statements, but it doesn’t make sense to group them together.
Like…is this comment itself a “negative” comment? Maybe. But I want the author to improve and I think most people here do too…and that’s where HN really shines.
Yes, probably the core limitation of my analysis (see earlier comment). My classifiers are treating "I don't like React because I don't want to write HTML in my JavaScript" the same as "React sux a$$" and that's clearly wrong. The models I'm using were built for general sentiment analysis, not technical discourse. On a meta level, your comment itself is a perfect example - it's "negative" in that it criticizes my methodology, but it's exactly the kind of feedback that I was hoping for, so thanks!
Same reason they infiltrate airspace duringtraining, fly drones over airports, run submarines through ports.
Testing limits and tolerance, threatening what they could do in a real attack. Creating econocic pain in retaliation for support with a strong alibi to blame.
Boarding and detaining is a new escalation. How many cables cut before we consider military reaction? 3? 10? all of them?
Government buying more fire trucks is a poor measure for increased incidence of actual fires. They might just be signaling, fall for marketing, and lots of other things. Governments are run by people, and they are as fallible as any rando on HN.
If you make a confident claim with no evidence, in a discussion that is rooted in everyone wanting to believe something, then the rational response is to doubt harder.
If you want to research this, the terms to look for is something like "Russian hybrid warfare goals and objectives in Europe" (can also try greyzone warfare).
Put simply, the goal is to add friction to anything NATO tries to do to aid Ukraine, and/or generally to improve its posture against Russia, while also trying to crack apart and erode both broad based and elite support for acting against Russia.
The EU and US were an unassailable bastion of freedom, peace, and prosperity, with arguably the most solid political foundations in history in democracy, and the most solid alliance in history in NATO.
How do you defeat such a place? You turn up the heat, to describe it very generally. It means, n a sense, radicalizing the population, a classic solution to Russia's problem. That's what terrorists do: How do you cause the US to shoot itself in the foot: terrorize people into thinking they are unsafe and overreacting (even though 9/11 affected on small area of one city).
One way they turn up the heat is to spread ethnic hatred, social distrust, embrace of violence, and abandonment of those things that prevent those maladies: universal human rights, democracy, rule of law, etc.
You can see it in this thread: People rooting for warfare, abandonment of the rule of law, etc. - all by some minor, cost-effective actions, like cutting a cable.
The expensive action and infinitely more consequential action - the invasion of Ukraine - remarkably doesn't create the same outrage. That outrage would trigger the obviously best solution: Guaranteeing unlimited material and political support for Ukraine until they win the war.
That is, it's remarkable if you don't appreciate information dominance, especially with social media companies either abandoning all responsibility or openly aiding the radicalization. Russia can create radicalization directly too.
Agreed it's what they're doing but this looks more like "turning everyone against you". And you want your enemies to underestimate you (like Song or Kievan Rus' underestimated the Mongols) but the world doesn't underestimate Russia. Maybe it could have but WW2 and appeasement are still too fresh in memory.
> this looks more like "turning everyone against you"
Someday, in hindsight, that might be what it looks like. Though notice that Russia has the world's #2 and rising power, China, on their side (in a marriage of convenience, I'm sure).
But I think there's an implicit misapprehension. The West is the status quo power, and Russia is revisionist, and both play out the traditional roles in that ancient game: Russia is challenging - trying to revise - the status quo order that puts the West on top. When you challenge the status quo order, lots of people invested in it don't like you; people born to the status quo can't imagine another way and reject you reflexively (and people also reflexively reject change).
This happens with tech innovation - status quo tech and disrupters. Sometimes it looks, in hindsight, like the disrupter 'turned everyone against them'. Sometimes people join them and they look like visionaries and leaders, and eventually elder statespeople. They are none of the above; they are just revisionists and sometimes it works out, sometimes it doesn't.
> WW2 and appeasement are still too fresh in memory.
I find that hard to believe. I have no data, but my impression:
First, there has been plenty of appeasement since WWII. NATO already appeased Russia on Georgia long ago, on parts of Ukraine (e.g., Crimea) a decade ago. The West threatened but then let the Soviets crush dissent in Hungary and the Prague Spring in the mid-20th century. Détente, arguably a degree of appeasement, was US policy in the 1970s. Trying to bring China into the global order was the policy until the mid-to-late 2010s. There are many more examples.
Second, the rush to war and the abandonment of human rights as the foundation, the rule of law, international law, international institutions, the outlawing of war, NATO, the rejection of nationalism, etc. etc. ... are all abandonments of the WWII generation's lessons, principles, and accomplishments. They built the postwar order. They conducted the Nuremberg trials, specifically a demonstration of the rule of law, justice, and human rights - and some Nazis were found not guilty.
It’s remarkable to me that a major new competitor in online distributed learning hasn’t already happened, considering the obvious LLM application.
But this press release makes me sad. At one point both of these companies had big visions for how online learning should happen. To read the announcement, it sounds like they’re being held hostage by a management consultant. There is so much gobbledigook and so little clarity about how to help people learn.
Yeah ditto. I don't know when it happened, but the Coursera courses I tried at first (around 2012 I think?) were very high quality -- I thought it was clearly a competitor to traditional brick and mortar.
Then a few years later, checked it out and there were thousands of courses, many clearly without as much thought or effort.
I am not as familiar with the other online schools that focus on quality (like WGU). I am surprised they have not eaten traditional schools lunches, since the actual quality of instruction is often very variable (I am a former professor, for the most part profs have little oversight in how they run classes). Market for lemons maybe?
Another aspect I am surprised at is that the big companies have not just started their own schools. UT-Dallas where I was at for a few years was basically started to help train up folks for Texas Instruments. (RAND Pardee school is kind-of an exemplar, although that is not focused on software engineering.)
I debate sometimes I shouldn't bother with hiring seniors and just train up everyone. If you have 10k software engineers does it not make sense to just have that level of training internally?
> Then a few years later, checked it out and there were thousands of courses, many clearly without as much thought or effort.
Thousands, and no decent way of separating the wheat from the chaff. Their filtering options suck. I'm also a bit disappointed that (most? all?) of their courses don't feature interactive exercises the way Khan Academy does. I mean I get they started out as basically a repository of recorded lectures, but i.e. a Linear Algebra course is pointless without practicing problems. A few overly simplistic multiple choice questions are the "best" I've seen on on Coursera.
The meaningful competitor wrt. raw educational content is freely available OpenCourseware made available under CC licenses, which prevent any after-the-fact rugpull. Of course online learning also has a big service-provision and perhaps certification component, which is where specialty platforms like Coursera and Udemy may have a real advantage.
The platforms lost because they enshittified and everyone left because of it, not because YouTube existed (it already did when they started). Compare the ancient "legacy" stuff that Coursera had to the stuff it has today. Little wonder nobody actually wants what they're selling.
As another comment here said:
> Those courses that were basically “we’re a top university and we let someone record the class from the back” were a literal life changer. Honestly, that was all I wanted.
The moment they stopped doing that, everything went to shit and this is the natural end result.
> It’s remarkable to me that a major new competitor in online distributed learning hasn’t already happened, considering the obvious LLM application.
I think it would be hard to make it work, without devolving into 50% slop. As in, it would still require very substantial continuous effort by dedicated experts, to provide a high-quality offering.
My grandfather’s Apple account was blacklisted too but I was less sympathetic to him because he genuinely sends spam email from his personal account (it’s politically motivated).
One day he was bricked from his accounts because he ran afoul of Apple’s ToS. The problem then was I couldn’t feel sure that he hadn’t actually done something which a reasonable person would say should result in account closure.
Paris’s case is much more strange, because it feels more likely to be a false-positive.
There is no legal right to have an account with Apple or Google, and I’m not sure I want there to be. But so much of our lives are built on these services and these stories erode our trust that the services themselves can handle the responsibility of adjudicating acceptable use. We need our digital accounts to be robust in the very long-term, even when there are bad actors who want to do all manner of bad things. And we need to feel confident that a properly empowered human reviewed the case and can articulate the reasons for a ban. When we charge a person with a crime, we tell them what the crime was and give them due process to fight it. I’m not sure I want the courts to decide these questions but we need some more due process when it comes to account termination.
> There is no legal right to have an account with Apple or Google, and I’m not sure I want there to be.
There shouldn’t be a legal right to an account, but there absolutely should be a legal right to sit down with someone from the company to plead your case, understand why the account was locked, and at least be given the opportunity to gather your things if they decide not give you a second chance.
If you get evicted from an apartment they don’t just change the locks and keep all your stuff…
There should be a legal right to a clear explanation and a mechanism of appealing these decisions with an external organisation. I think it’s unreasonable to expect that they should be able to delete users this casually with everything that is tied to your devices.
You could make it so costs for arbitration could be paid up front by the person appealing and then if the account deletion was deemed wrong the company refunds said user. Could probably apply to monetisation on YouTube that I see withdrawn for very dubious reasons too.
>arbitration could be paid up front by the person appealing
We need a constitutional amendment that prevents binding arbitration agreements, which removes judicial review from public accessibility.
There absolutely should be a legal right to pursue this through the courts (which require a response from the company, to avoid default judgment).
----
My main PiHole blocks all of *.google.* & *.apple.* for many reasons. My exploration into PiHoles began a decade ago, after Google pulled a similar response-less account termination (without explanation). This left me unable to update a blog (with several million annual impressions), with no recourse [0].
[0] Unlike OP's situation, I was able to download most of my writing/photos, only because they were public-facing (website).
We have these systems - they are called courts. The subject is in Australia and so am I, I can file a case up to around $100k USD for $150 in filing fees.
If Apple doesn't respond they will lose by default and possibly be held in contempt.
Are you sure this is available in the UK or US? I’m pretty sure all the legal agreements you sign when joining say the decisions of Amazon or YouTube or Facebook are final so are these Terms illegal?
You can file a claim for whatever you want in a court. You have a much higher chance of winning if your claim involves something like you paid a company for a service and they didn't give you it - and now you have damages.
If the company says they have a contract with you where you agreed to this, that's not a very good defense against the Consumer Legal Remedies Act or the Unfair Competition Law, one or both of those provide statutory damages, these are California laws that get enforced every day. You find a Superior Court in CA, pay a few hundred dollars to file, and let Apple respond.
It would be the same story in the UK as it is in California or Australia.
it seems like car-makers themselves feel burdened to make their own self-driving tech, as opposed to outsourcing the software to a third party.
Dell and HP don’t make operating systems…it seems like having a handful of companies focused on getting the self-driving part right without the need to also specialize in manufacturing would be beneficial.
My first inclination was to be bullish on Rivian, and there’s no question that their vehicles are beautiful. But is there anything to suggest they have an advantage over Tesla or other automakers when it comes to self-driving?
Most traditional OEMs in the automotive industry are integrators. They write a specification, and a Tier 1 supplier provides a black box matching that spec. (Tier 1 in turn provides a spec to its suppliers and integrates their parts)
This has several consequences
- Tier 1 suppliers are waiting on input/approval from OEM before proceeding with projects
- Tier 1 suppliers don't necessarily have the capital to do work "at risk", even if they could build the part without approval/specifications. (TBH - some do)
- Each layer of the supply chain lacks context on the whole project and product line and cannot achieve efficiencies outside of the scope of its contract.
These haven't really been a problem for mechanical parts and E/E parts that have well-defined functions and interfaces and have a lot of re-use from previous generations. It works really well with Kaizen (incremental innovation).
To outsource it, a traditional OEM would need to completely specify the behaviour of said self-driving system, baseline the specification, put out the requests for quotation, etc. Tier 1 then needs to analyse the spec, estimate it, break it down in to sub work packages, work with its suppliers, etc. From an optimisation point of view, this is really inefficient partitioning of the problem space. For greenfields development, an emergent specification via experimentation may be better - but that won't fit in traditional V-model sub-contract OEMs/Tier 1s use.
That flow doesn't need to be followed; the suppliers could raise/allocate capital and build the self-driving stack "at risk" - and this seems to be done (Tier IV, Waymo, etc). But as it's new technology, I assume Rivian think they can do better by themselves and can get the capital for the development as part of an integrated solution while they are smaller it might seem they should not waste limited capitial that way - but integration will save a lot of inefficiency in sharing specifications across boundaries, full system integration and deriving emergent specification via experimentation rather than some MBSE folly.
Not only is Rivian betting on an integrated platform being important for their own cars long term, they’ve also essentially sold that portion of their business to VW. They are investing in the software platform for a lot more cars than just the rivian branded ones.
Level 2 is simpler and more mature, so it should be easier to specify, package and integrate.
It can follow the traditional OEM outsourcing route, where OEM has high-level models and gets suppliers to implement the details.
e.g. I can find public information that Subaru use Veoneer cameras, Xilinx chipsets, but defines their own algorithms. I would speculate they have an outsourced company convert algorithms to FPGA netlists/embedded code.
(On the other hand, I know other OEMs have a more complex mesh of joint ventures)
Aside from Mercedes-Benz, level 2 is the best we have, though?
Edit: Re-reading your post, you were talking about the future. You make good points; I will be interested to see how it plays out. Probably as long as the technology remains radar and cameras, it will continue to be easily outsourced.
HP did write their own operating system: HP RTE. It wasn't until decades later, after the platform became commodified and they stopped designing their own chips, that they went with someone else's.
And of course, Microsoft made their own cards back in the day, and they still make the XBox as integrated hardware.
This technology is way too early for commodification. Right now, Rivian is a data play.
Their platform means they have consistency other providers don't have. They have data from the existing trucks on the road, and they'll roll out these sensors long before they roll out self-driving. Cleverly they've also pitched these as "adventure" vehicles, which means they'll have some data from rarer situations, not just highways. Off-road performance, for example, will add anomalies that they can use to stress-test self-driving code. If a car could handle areas without roads, it is less likely to kill people if a mudslide happens. Or a shadow falls across the bridge.
> as opposed to outsourcing the software to a third party.
All about margins and data protection
> Dell and HP don’t make operating systems…
They come from an era where this style of thinking didn't really exist and now they're in an era where market share almost entirely prevents new players/ideas from entering. If it was 2025 and OS market share wasn't static and the PC was just taking off, we'd probably see far more attempts at OS development to better monetize the PC products the vendors sell.
It’s probably a sensible concern that if they use someone else’s tech, they’ll be subsidizing that company’s eventual mastery of the self-driving space, who will then be able to control pricing. The only long game, I imagine, is to create your own self-driving tech, so that your own customers are investing in your own long term success, not someone else’s.
One aspect is that Tesla is all cameras, whereas Rivian sees it as important to have multi-sensor suites (cameras, ultrasonic, radar, and in Gen 3: lidar). TBH as a customer I prefer to know that the latter is protecting me instead of just cameras.
They could have a better driving assistance package than 99% of other cars on the road for 1/10 the price by using OpenPilot as the LKAS, or installing a Comma in the car.
Real shame nobody has taken that approach, not even a fork
Comma had made essentially no efforts to meet the requirements of automotive systems the last time I looked at them. They would be an incredibly risky supplier for systems that could easily come under regulatory scrutiny.
It's an... interesting approach, They essentially reduce the surface area as much as possible. I don't buy that it's enough, but, again, interesting to see what they do.
Besides, a big OEM could pour an army of developers and turn the Comma approach into an ASIL D, it would be quite a lot of work but within the realm of possibility.
I recently ordered a Comma Four, which I will install BluePilot on and use with my Lightning. I'm looking forward to seeing how competent it is. Gets good reviews, at least, so I'm hoping.
I do wonder if there will be many more iterations, though, with so many manufacturers switching over to an encrypted canbus and locking out the control method comma uses.
:)
reply