In the case of vibe-coded slop like OpenClaw it's not a question of some vague notion of "code quality", it's a case of the software shitting the bed and not working anymore, with no recourse of fixing it. (Neither humans nor LLMs have the context window to analyse and fix tens of millions of lines of code slop.)
> and Anthropic banned OpenClaw usage
If OpenClaw wasn't broken it would just use a standard token API.
But see above - as software it is fundamentally broken and unfixable.
That doesn't seem to be true, the global expectancy is 73 and only a dozen of the wealthiest countries are mid 80s. I wouldn't be worried about dying at 70, but maybe I'm built of tougher stuff.
You say my claim is not true then proceed to acknowledge it is true. To be clear about the fact: about 1/4 of all countries have a life expectancy >80, and as you said, more than a dozen in mid-80s.
Well, no, no country on earth has an expectancy in the mid 80s, I was being generous with your wording. 40-50 countries have expectancy between 80-84, which is early 80s. Most countries are sub-80.
I don’t think that’s an accurate or useful characterization of modern AI like Claude at all. It is not simply regurgitating knowledge. It applies its knowledge to create bespoke solutions to the problem you pose to it, and is able to self evaluate its progress towards the completion criteria. If you don’t think that counts as “problem solving”, your definition would exclude nearly all knowledge work and engineering.
People underestimate the vastness of training data (internet) and overestimate their ability to recognize if something is really bespoke. Not to say the no problem solving is happening, because there are many problems that we inefficiently solve again and again and the LLMs are making the solutions more accessible to everyone with a subscription.
> It applies its knowledge to create bespoke solutions to the problem you pose to it, and is able to self evaluate its progress towards the completion criteria.
It imitates applying knowledge. The imitation may be uncanny, but assigning LLMs intentionality and ToM is a category error.
Does "applying knowledge" necessitate human-like intentionality and theory of mind? If you insist it does, and this is a category error, then we need a new category.
By analogy, consider that many have referred to classical, deterministic computing as some kind of "thinking" for the last half century+. Does this stop being kosher when the computer has an uncanny propensity for human language? Perhaps, but the computer is still clearly chewing through problems that would have required a lot of human thinking (e.g., arithmetic) in ages past.
I haven't seen any genuine proposals for words to replace the human mind analogues, let alone proposals that the anglosphere would plausibly adopt en masse.
The article does not claim they have achieved recursive self improvement... just that it appears to be a plausible outcome given the progress of AI development in the past few years.
I don't know about you, but AI advancements have brought extraordinary improvements to me personally in my ability to be productive, in much the same ways the article outlines. I find it deeply satisfying to be able to "get ideas out of my head" faster and tackle more meaningful problems.
FWIW, it deeply concerns me how much power and capability is being centralized in the hands of so few, especially Anthropic. I, for one, hope these advancements can be scaled down to something I can have full sovereignty over and trust... in my own home.
It depends. If you are in a disadvantaged class it is very likely going to err towards a dismal result long term. However if you are a privileged intellectual these models can accelerate and expand your horizon. It isn't the end, surely. It is, however, both impressive and depressive simultaneously and that perspective only depends on your point of view.
But when the bar to entry is beyond expertise in a field or subfield, how does an individual ever hope to attain an unexplored space to explore?
It may be the beginning of thinking, but to many who view things on a longer timeline. It starts to look like it will breakdown the frameworks of which are required to get to that position. Otherwise, you just end up retreading explored ground. This removing the joy of discovery from any humans hand/mind.
Recently I've found my mind reawaken. It's about asking good questions now. The models can find the answers, but you have to know what to ask. Sometimes the model is wrong and you have to challenge it to find an alternative. Being able to explore problem spaces quickly is interesting.
Are you saying this result is uninteresting and therefore AI slop or puffery? Obviously OpenAI has a motivation to "market" the accomplishment as much as possible, but surely you agree it IS a remarkable achievement?
I'll let the mathematicians in the field determine the level of "interest" in this result, but saying "you may want to make sure you are sitting down" is pure puffery.
> has a motivation to "market" the accomplishment as much as possible
I am so sick of HN promoting unethical behaviour as virtuous due to it's financialization worship at the foot of "valuations".
> but surely you agree it IS a remarkable achievement?
If you could define the bounds of "remarkable" I could answer this question.
It's remarkable, its not out of the bounds of the pattern of success that AI has had with math recently to the point that people should sound alarm bells.
A lot of the weight this holds is the fact that it's an old problem and that its difficulty hinges on the lack of investigation the disproof side of hypothesis. The model basically took a contrarian path and found tools and methods that support that a disproof is viable. So the (unquantified amount of) mathematicians out there were all dedicating their resources on the notion that this can be proved. Some with hindsight would say that if they a had team of experts who are driven to the goal of disproof that this would have been achievable by humans, and one of the mathematicians of the paper state as much,this still has value in terms of reliability measurement, and possibly human-aided endeavors when the methods scrounged by the model can be used in other solutions.
IRC is distributed and federated. Not only are there countless networks, each network has countless servers, and each group of servers that are up and can see each other can operate on their own, all the way down to a single server, or up to any subset up to all.
When a peering connection goes down and the network splits, maybe some people in the group disappear, or maybe from your point of view everyone else disappears.
Maybe the remaining subset of other users is already good enough because it's enough to continue what you were tallking about and who you were talking with, or if not, you have the option to just try some other servers until you find where everyone alse is. Were "server" is an actual seperate instance of the server software operated by an independant person, hosted on whatever kind of hardware or vm they set up, connected to whatever network they are on, not what Discord calls a "server".
Even if the entire group of say freenode servers goes down somehow (even though that's not really possible) there is still undernet and 400 other nets. Even without prior coordination it would be essentially trivial for the users to all just go looking for, or create on the spot, the same channel on some other net, and basically everyone finds each other again almost effortlessly. And that's if something unbelievable actually happens, let alone the normal minor breaks that actually happen once in a while.
This is entirely different from being wholly at the mercy of the single entity Discord.
Freenode had full-network outages periodically. ddos attacks, infrastructure failures etc. and when those happened, the practical experience was the same... people waited it out. Nobody coordinated a mass migration to undernet or stood up alternative servers for a few hours. (It took much bigger issues - social/organizational/political, not technical - to catalyze the mass migration.)
You're making an argument about the virtues of decentralization - and I agree, decentralization is great! Just in practical reality, freenode (not IRC itself) had exactly the same failure mode as we just saw today.
I'm not really arguing the superiority except as a side effect of answering the question "isn't it the same?" No, it's not the same at all.
When irc servers go down or networks split, the users just hop on other servers if they actually want to keep talking with the same specific people they were.
The earlier comment posed the scenario of being high and dry when they needed to do basically support work with users.
If they don't care then they don't bother, but that doesn't change the fact that if the communication mattered to you, then you are not stuck the way the earlier comment said. You are only as stuck is you feel like. And not just because of the hyperbolic technicality that you can stand up your own new server. You could, but you never need to. There are countless servers already.
And there is no coordinated "mass migration" operation. The channel breaks, you just go browse a few other servers, or if it's exceptionally bad, maybe another network. It's effortless for each user. It takes 2 seconds. You're not trying 1000 other networks blindly either. You already know a few more popular/likely suspects to try first. And so does everyone else, and even when you guess "wrong" and go to dalnet and everyone else is on undernet, there will still be a channel on dalnet with a topic or a user telling you where to go. It's all just not a big deal. It took way more effort to write this comment explaining it than to actually deal with a net split and get back in communication with more or less everyone that was in whatever channel broke.
Just speculating, but I don't think it's unrelated. Discord heavily utilizes Cloudflare, and Cloudflare uses Let's Encrypt for a certificate issuance. If they happened to have a certificate signing dependency in some operational rollout today, I think it could explain it. Certainly the timing is very correlated.
For domains where they handle the certificates, Cloudflare utilizes multiple CAs, to avoid such a single point of failure: I’ve seen Cloudflare managed certificates issued by Let’s Encrypt, Google Cloud, Sectigo, and SSL.com.
Cloudflare does provide the option for customers to manage their own certificates, which would make it the customer’s responsibility to have alternatives issuers when needed.
I guess we'll find out but it would be surprising if they use Let's Encrypt for their backend services. The front door is issued by Google Trust Services.
On my account they always serve Google issued certificates. There is also Let’s encrypt certificate but it is not used though. I guess that’s a fail-safe.
In Cloudflare Enterprise you can pick either or leave it on auto. Iirc there's a 3rd option but I don't know if it's still supported (Terraform and SDKs used to have it in the enum)
I would say a better analogy is using Google… you can use it as a tool to seek information and deepen your understanding. But it requires your brain to be engaged and to be putting that stream of knowledge into practice.
I would say far more likely is that its creator was acqui-hired and Anthropic banned OpenClaw usage.
The reality is that AI is both capable of producing sloppy code and capable of cleaning it up, if directed to do so, just like humans.
And, just like humans, code quality is very rarely the make or break factor between success and failure in business, much less popularity.
reply