> well I can't leave a task with the LLM and come back to it tomorrow
You could actually just do that, leave an agent on a problem you would give a junior, go back on your main task and whenever you feel like it check the agent's work.
Everything I’ve read about experiments where they’ve tried this have been massive failures. The AIs always get stuck and can’t make further progress at some point when given the full responsibilities of a human employee.
It lacks the ability to self correct and do all the adjacent tasks like client comms etc. So if I come back to it in the afternoon I may have wasted a day in business terms, because I will need to try again tomorrow. What do I tell the client, sorry the LLM failed the simple task so we will have to try again tomorrow? Worse, lie and say sorry this 2 hour task could not be achieved by our developers today. Either way we look incompetent (because realistically, we were not competent, relying on a tool that fails frequently)
I'm sorry but I'm not familiar with the context you mention, have not worked in a job where I had to communicate with clients and I find it hard to imagine a job where a junior would have to communicate with a client on a 2 hour task. Why would you want a junior to be the public face of your company?
Digital agencies often trust their developers to interface with their clients. These are tight, trusting relationships built between two technical or tech adjacent teams in most cases. To say they are a public face would misunderstand the relationship a bit, both teams have juniors and our juniors have high expectations of professionalism and execution.
> I dont want default language. I understand multiple of them. And it is even ridiculous that I have to set it up.
For Silicone Valey, it is difficult to comprehend, that people may be speaking more than 1 language.
That's why you get programs in other languages than intended (phone set to English - get the English version of app) or they "offer" ( ok/not now) to translate.
That list would be incomplete. Americans at least don't tend to "helpfully" automatically proxy their whole site through Google Translate when they detect foreign IPs.
The most baffling thing is that we aren't talking about Hurrah-Americans here. We are talking about Google, which is full of Indians on all levels of the company. They, if anyone, should have understanding of multilingual people, and yet... such an incredible mess, which is still not fixed after many months.
Never gonna happen. Cash has no intrinsic value exccept maybe for use in fire / toiler paper. GPUs while currently inflated in price will always find enough value. Their price might go down 50-75% but never 99%
That's the meaning of intrinsic calor - the device can do what it can do, regardless of market conditions. Today it has the value of fifty teraflops, and tomorrow it still does, unless it breaks. However, intrinsic value cannot be measured in dollars.
And yet we're talking about electronics here, they don't have sentimental value and just because compute capacity is unused there are no guarantees that it will be used, even at a per unit cost approaching €0.
I'm sure that farmers during the Great Depression were also consoling themselves with the "intrisinsic caloric value" of their corn.
As I said, the intrinsic value of a GPU is not measured in €. In fact, the lower the sale price gets, the better a deal it is, not worse - you get the same intrinsic value for less extrinsic cost.
There are also intrinsic costs, mostly power consumption.
Food calories are cheaper to convert into something useful. It's not like GPUs, once bought for peanuts, turn into perpetual motion machines. They need power, cooling, a whole infrastructure built around them.
GPUs would have taken the world by storm already in the roughly 30 years since they've been around.
Even for GenAI it's likely ASICs take over at some point if we really care about performance.
GPUs used to cost 20% of what they cost know and Intel and AMD make perfectly serviceable GPUs for most PCs. NVIDIA top of the line GPUs won't suddenly be plugged in to lowly laptops.
Yes, lots of companies will buy them for cheap, but these AI beasts also have OpEx costs. Not every alternative use is worth the money and there are 0 guarantees that the alternative costs cover the gap. NVIDIA sell 80% of GPUs for AI now.
I think people don't realize just how big this bubble is.
On the other hand some notable open source leaders seem to be abrasive assholes. Linus, Theo, DHH, just three examples who come to mind. I think if you have a clear vision of what you want your project to be then being agressively dismissive of ideas that don't further that vision is necessary just to keep the noise to a low roar.
Even Linus doesn’t act that way anymore. Here’s him a few years ago:
> This week people in our community confronted me about my lifetime of not understanding emotions. My flippant attacks in emails have been both unprofessional and uncalled for.
> Especially at times when I made it personal. In my quest for a better patch, this made sense to me. I know now this was not OK and I am truly sorry. The above is basically a long-winded way to get to the somewhat painful personal admission that hey, I need to change some of my behavior, and I want to apologize to the people that my personal behavior hurt and possibly drove away from kernel development entirely.
> I am going to take time off and get some assistance on how to understand people's emotions and respond appropriately.
He took time off and he’s better now. What you call “political correctness” is what I and others call “basic professionalism”. It took Linus 25 years to understand that. I can only hope that the people who hero worshipped him and adopted a similar attitude can also mature.
If you’ll notice, he called the code garbage, not the author. Judging by how bad the code was, I think this interaction was fine. This actually shows the progress Linus made in improving himself.
But you got to give it to him, he does seem to be really good at catching deficiensies early that may accumulate to become serious bugs or security vulnerabilities in the future. Sure, being an asshole is not ok, but being assertive is a must for a person in his position.
> Makes me wonder how much to the mass strife and confusion of the internet is simply down to people not knowing what the words they use mean?
Or being intentionally misled about them. People who enjoy being awful in various ways have a vested interest in reframing the opposition as "political correctness" in order to make it easier to dismiss or ridicule. The vast majority of usage of the term "political correctness" is in dismissing or ridiculing it.
It has everything to do with political correctness. Honest, blunt language is now de-valued in favor of passive, sanitized, AI-slop language that no longer conveys important information. The revised post forgot to mention the critical point of the bloated, buggy Javascript framework because it would offend someone here.
Prefer a blunt, honest dick over a passive, polite liar anyday.
Hmm I don’t think any of the revisions are about being PC but rather not making juvenile comments. Linus has definitely made a lot of harsh inflammatory comments to others, I don’t think it’s the right thing to do and shows his character but at the same time for me at least it comes across as a smart pompous jerk who says things in the wrong way but at least usually has some kernel of a point.
The Zig comments come off has highly immature, maybe because they are comments made to unknown people, calling folks losers or monkeys just crosses some line to me. Telling someone to stfu is not great but calling groups of people monkeys feels worse.
Calling the devs of Actions "monkeys" has nothing to do about being un-PC or not. It's just plain rude and deeply insulting. It has no place in an a public announcement such as this.
Also, Torvalds was rightfully called out on his public behaviour and he's corrected himself.
Perhaps he should be. This idea that we should tolerate terrible things and only respond to them politely seems to produce bad outcomes, for some mysterious reason.
Any analysis of Github's functionality that begins and ends with blaming individuals and their competency is deeply mistaken while being insulting. Anyone who has ever worked at a large company knows exactly how hard it is for top performers to make changes and it's not difficult because the other people are stupid. At least in my experience, almost everyone holding this "they must be stupid" opinion knows very little about how large organizations make decisions and knows very little about how incentives at different levels of an org chart leads to suboptimal decisions and results. I would agree with you that being overly polite helps no one, but being correct does, and what they initially wrote isn't even right and it's also insulting. There's no value in that.
I am not convinced of this. Being rude and insulting someone’s intelligence is rarely a good trait. Linus got away with it due to the unique circumstances: leader of an incredibly popular open source project and a gatekeeper to a lot of access to it.
My argument against how he handles things has always been that while it may seem effective, we do not know how much more effective he would be if he did not curse people out for being dumb fucks.
And it doesn’t seem like this is a requirement for the job: lots of other project leaders treat others with courtesy and respect and it doesn’t seem to cause issues.
The reality is that it is easy to wish more people were verbally abusive to others when it isn’t directed at you. But soon as you are on the receiving end of it, especially as a volunteer, there is a greater than not chance that you will be less likely to want to continue contributing.
I think this is a good way to put it and I agree with it. Linus is a jerk and I would never want to work with him. Doubly so with zig maintainers who call other groups of people losers or monkeys. Shows a clear lack of maturity and ability to think.
Eh. Linus has a long history of abusive behavior towards other Linux contributors but also apparently apologized for it and started amending his ways. The Zig person I do not know by reputation, let alone in person. One post that he later chose to amend based on feedback is not enough for me to pass that kind of judgement. If anything, the fact that he updated it shows the opposite of lack of maturity. Adults can get frustrated. What they do with it is what matters.
Zero clue what your point is so please help me understand.
I was agreeing with your stance and adding my own anecdote that it’s a turnoff with the way those posts were originally formatted. Not people I would want to work with. If you do that’s fine. This is not star wars and simply my own choice as it’s everyone else.
I also cannot think of a time in my adult life I wanted to call out a group of people as losers or monkeys i n public.
My point is that Linus and the Zig guy are in different categories in my mind. I think it is a bit naive to lump them into the same category.
I would definitely classify the tiki torch wielding white nationalists as losers publicly, for example. In fact I have a hard time thinking of a better term for them. It could also apply to the fairly famous liar and criminal, the disgraced Congressman George Santos. Or any person who decides to flash kids at a playground, or beats his wife and children.
I think the Zig guy was a little over-dramatic with his initial post. He did change his mind, so in my book that's better than not. Linus did too, just after many years of bad behavior. My point is that your replies were painting the world with only black and white and there is a lot of gray area in between. Sometimes public shame is a valid way to do discourse. Often times it isn't. But it's not a "always" or "never" thing.
I did not realize we were lumping Microsoft engineers alongside white nationalists and pedos. Sure folks like that I can see people using descriptions like that.
> I also cannot think of a time in my adult life I wanted to call out a group of people as losers or monkeys i n public.
I guess that makes this your first time:
> Sure folks like that I can see people using descriptions like that.
All in all I think we generally agree that being respectful is better than being rude. And that some people who do not have respect also do not deserve respect. Shall we just leave it at that?
Then stop replying if you want to leave it at that? I have only agreed with your original statement and then you keep questioning my opinion. You are trying to pick over my words for no reason. Note I said I can see people using that language. I did not say myself. And of course why would I even think about pedos in the context of rude comments made to an unknown group of Microsoft engineers.
My opinion, I have no desire to work with people that write comments calling other engineers monkeys or losers. I have seen that behavior before and it’s not people I like to work with.
Because one person is judging that "terribleness" before being entitled to flame, changes to that person influence their ability to objectively make that assessment.
Say, when their project becomes popular, they gain more power and fame, and suddenly their self-image is different.
Hence it usually being a more community-encouraging approach to keep discussions technical without vitriol.
Flaming is unnecessarily disruptive, not least because it gives other (probably not as talented) folks a license to also put their worst impulses to text.
He represented his community with insulting words to the world. In higher ranks of IT it is all about communication. With his lack of proper words he showed these leaders, who decide about the adoption of Zig, that they do not want to communicate with him/the Zig community.
As a project/tech leader he is in the business of communications. He recognized this. See link in the article.
there's a big gulf between being politically correct and not being a jerk. In this case the community reps can present their concern, motivation and decision without insulting people. It's also not a smart or valid comment; give me any organization over 100 people and I can find something deeply flawed that it hase produced or a very bad decision. Do I then tag everybody who currently works for that organization as "a brain-dead idiot" or similar?
Eager to do what? If it sucks it sucks, but that's a very childish way to frame it, no one did anything on purpose or out of spite. That kind of silliness hurts the image of the project. But bad translation I suppose.
One can avoid being asshole even if it is not strictly speaking necessary. In fact, if you are an asshole when it is not necessary, then you are an asshole.
Not calling other software engineers 'losers' is not about political correctness. They're "losers" because they take their product on a path you don't like? Come on. Linus can be emotional in his posts because Linux is his "child".
Several ways, although I'm not sure whether the below will happen:
1. Paid ads - ChatGPT could offer paid listings at the top of its answers, just like Google does when it provides a results page. Not all people will necessarily leave Google/Gemini for future search queries, but some of the money that used to go to Google/Bing could now go to OpenAI.
2. Behavioral targeting based on past ChatGPT queries. If you have been asking about headache remedies, you might see ads for painkillers - both within ChatGPT and as display ads across the web.
3. Affiliate / commission revenue - if you've asked for product recommendations, at least some might be affiliate links.
The revenue from the above likely wouldn't cover all costs based on their current expenditure. But it would help a bit - particularly for monetizing free users.
Plus, I'm sure there will be new advertising models that emerge in time. If an advertiser could say "I can offer $30 per new customer" and let AI figure out how to get them and send a bill, that's very different to someone setting up an ad campaign - which involves everything from audience selection and creative, to bid management and conversion rate optimization.
So I don't necessarily disagree with your suggestions, but that is just not a $1T company you're describing. That's basically a X/Twitter size company, and most agree that $44B was overpaying.
It's not that OpenAI hasn't created something impressive, it just came at to high a price. We're talking space program money, but without all the neat technologies that came along as a result. OpenAI more or less develop ONE technology, no related product or technologies are spun out of the program. To top it all off, the thing they built, apparently not that hard to replicate.
ChatGPT usage is already significantly higher than Twitter at its peak, and there is a lot more scope activity with explicitly or implicitly commercial intent. Twitter was entertainment and self-promotion. Chatbots are going to be asked for advice on how to repair a dish washer, whether a rash is something to worry about, which European city with cheap flights has the best weather in March for a wedding, and an indefinite stream of other similar queries.
I might get worried when mainstream computers won't be able to run Linux. Until then.. I'm not worried.
Seems there are efforts to bring openness to platforms that inherently have an interest to resist it and while the progress is slow.. there is progress
I think this might be the way forward, Claude is great at project managing.
I’m already telling Claude to ask Codex for a code review on PRs. or another fun pattern I found is you can use give the web version of Codex an open ended task like “make this method faster”, hit the “4x” button and end and up with four different pull requests attacking the problem in different ways. Then ask Claude to read the open PRs and make a 5th one that combines the approaches. This way Codex does the hard thinking but Claude does the glue
I view it more as fun and spicy. Now we are moving away from the paradigm that the computer is "the dumbest thing in existence" and that requires a bit of flailing around which is exciting!
Folk magic is (IMO) a necessary step in our understanding of these new.. magical.. tools.
I won't begrudge anyone having fun with their tools, but folk magic definitely isn't a necessary step for understanding anything, it's one step removed from astrology.
I see what you mean, but I think it's a lot less pernicious than astrology. There are plausible mechanisms, it's at least possible to do benchmarking, and it's all plugged into relatively short feedback cycles of people trying to do their jobs and accomplish specific tasks. Mechanical interpretability stuff might help make the magic more transparent & observable, and—surveillance concerns notwithstanding—companies like Cursor (I assume also Google and the other major labs, modulo self-imposed restrictions on using inference data for training) are building up serious data sets that can pretty directly associate prompts with results. Not only that, I think LLMs in a broader sense are actually enormously helpful specifically for understanding existing code—when you don't just order them to implement features and fix bugs, but use their tireless abilities to consume and transform a corpus in a way that helps guide you to the important modules, explains conceptual schemes, analyzes diffs, etc. There's a lot of critical points to be made but we can't ignore the upsides.
I'd say the only ones capable of really approaching anything like scientific understanding of how to prompt these for maximum efficacy are the providers not the users.
Users can get a glimpse and can try their best to be scientific in their approach however the tool is of such complexity that we can barely skim the surface of what's possible.
That is why you see "folk magic", people love to share anecdata because.. that's what most people have. They either don't have the patience, the training or simply the time to approach these tools with rational rigor.
Frankly it would be enormously costly in both time and API costs to get anywhere near best practices backed up by experimental data let alone having coherent and valid theories about why a prompt technique works the way it does. And even if you built up this understanding or set of techniques they might only work for one specific model. You might have to start all over again in a couple of months
> That is why you see "folk magic", people love to share anecdata because.. that's what most people have. They either don't have the patience, the training or simply the time to approach these tools with rational rigor.
Yes. That's exactly the point of my comment. Users aren't performing anything even remotely approaching the level of controlled analysis necessary to evaluate the efficacy of their prompt magic. Every LLM thread is filled with random prompt advice that varies wildly, offered up as nebulously unfalsifiable personality traits (e.g. "it makes the model less aggressive and more circumspect"), and all with the air of a foregone conclusion's matter-of-fact confidence. Then someone always replies with "actually I've had the exact opposite experience with [some model], it really comes down to [instructing the model to do thing]".
You could actually just do that, leave an agent on a problem you would give a junior, go back on your main task and whenever you feel like it check the agent's work.
reply