EternalFury

That type of reasoning really reminds of the Internet bubble.

Either you believe in emergent behavior or it’s only very good at recognizing patterns. Which is it? I made a test called LLMLOLinator (stupid name) and I can tell they cannot stray from the probability distributions they learned during training. So, I am not confident in so-called emergent behaviors.

Impressive quality

My point is that it’s not the spirit of the games. Ancient Greeks didn’t even keep a record of any metrics. Back then, there was no assumption that “modern” competitors should exceed the performance of those who competed decades or centuries before.

This type of focus leads to unhealthy expectations and doping in my opinion.

World records don’t matter in the Olympics. Medals go to those who show up and prevail.

I never expected this quick jab to trigger such reactions.

Yes, it’s glue. True for Ruby, PHP, Perl as well.

I just think it’s disingenuous to associate the word “fast” with anything implemented with it, granted that C is 70 times faster than it. I mean, me saying it’s slow compared to C is less hyperbolic than saying it’s fast.

And I think Python is perfectly adequate for plenty of use cases, so is PHP, so is Ruby. I like to be mean to JavaScript, but even that is fine for web dev.

Python is fast compared to something? Maybe fast enough to generate HTML.

People always conclude I hate AI. But it’s only a matter of ROI. The amount of capital investment must be matched with increased sales or reduced expenses. If reduced expenses come via job elimination, it doesn’t last long in a consumer society. Increased sales through new product categories are an option, but costs will need to come down and reliability will need to increase. Hence the current hopes.

It’s healthy for GS to express a contrarian opinion.

If GPT 5 doesn’t meet high expectations, another winter could soon arrive.

It’s hard to make LLMs ignore what they were trained to generate. It’s easy for humans. Isn’t that an obstacle on the path to AGI? I was doing trivial tests that demand LLMs to swim against their probability distributions at inference time, and they don’t like this.

You are most likely right. I just fear spreading complexity everywhere. But I also want to take advantage of the vast compute available at clients. So, I am conflicted. Vue is probably a good compromise. Please share your best references on Inertia/Vue.

It’s hard to call awk, grep, sed, lex, programming languages, don’t you think? It would be even harder to say they popularized regular expressions, considering their very very limited reach. I agree with the rest.

I would pick PHP for its intended use, web dev without fat clients. Laravel is the best framework I have ever used…and I have used them all.

I was there then and I am still here today. Ahah.

Perl was the language best suited for text manipulation because of its early popularization of regular expressions. As such, it was excellent for a lot of tasks in « DevOps », even though that term wasn’t a thing then (everyone carried their own water).

I remember deploying a web forum that was coded in Perl once. I remember having to debug large Perl scripts and that was a nightmare. I remember a time when Python was a new odd thing and the long-bearded hacking elite joked about it.

But, all this will be lost in time, like tears in rain…

Just remember one thing: The mess you make with any language is yours, not the language’s.

If it's done right, by documenting a mature interface and its inputs/outputs, no one needs to know what atoms make up the slice of bread that gets buttered.

We were able to resist the pressure until they turned 12 in both cases. But it was not enough, we should have held the line until 16 or 18.

Peer pressure is the biggest factor. There is probably a threshold beyond which your kids are going to give their attention away because so many of their friends are doing it. I once hosted a birthday party where after the cake was cut, all 7 kids were on their phones, barely interacting. That was one of the most dystopian thing I have experienced.

As a parent, I wish people who monetize attention had not hurt the development of my children’s social skills.

We really tried to curb their engagement with attention-channeling apps and devices, but it got to a point where restricting was enough to label parents as harsh, strict luddites.

I suspect it will take a generation or two to revert the effects of that attention hijacking.

You get Founder Liquidity because you managed to convince people who do just as much work that they couldn’t do just as much work without you. LOL I am so cynical.

If it passed The Area 101 Test, it would already be amazing, as this is a trivial test that goes against the fundamental principles of LLMs.

So much is riding on GPT5, if it disappoints, it’s another AI winter.

Let’s not be too picky. This is a good thing.

Good luck. It’s much easier to talk about it. The last open OS I have seen reach a semi-mainstream level of adoption was started in the early 90’s, more than 30 years ago, by some Linus guy.

Alas, "summarization" is a crapshoot. It sometimes summarizes essential facts out of the original text. I think it's due to limits on output tokens. And chunking cannot solve this because you cannot summarize well if you don't take in the whole thing. Aggregating summarized chunks doesn't produce good results.

It works very very well on Amazon reviews, though, because these are small and thin.

I am always reminded of Severance, the TV show, when I think of LLMs. Data refinement is what is happening massively right now. Basically, take data harvested on the Internet, which may be of poor quality, and use a teacher-student approach to reformulate it in higher quality. Then use that synthetic data to train new models. It's a bit like the biblical multiplication of the bread loaves. How much farther can it get LLMs? I don't know.

During the next turn, though, a lot of the same synthetic goop is going to turn up on the Internet. That's probably why OpenAI is spending the big bucks to secure access to original content producers. Now, here is to hope "original content producers" are not going to depend on LLMs to generate that original content.

I think training data would be plentiful if autonomous drones roamed the face of the earth, constantly recording audio and video. There is nothing as good as reality to learn what real is. The representation of reality on the Internet has become a matter of opinion for a lot of people.

My hero! I would never have stayed in “computer things” had I not read Donald’s volumes.

I too spent time crafting the perfect tool and sharpening the saw. I wish I had spent that time building things with existing tools, as I believe it was a form of procrastination.

At least, it was true in my case.

Sometimes you don’t know where things will end up, so you should repeat yourself with the expectation things will diverge as complexity creeps in. If the system reaches stability and there’s duplication at that point, you can hunt it down by identifying common concepts.

Pretending to know from the start where things will end up is silly and leads to very abstract abstractions.

Does performance degrade over time like with all other Windows installations?

What I like about Linux and macOS is that performance remains stable and battery life is excellent.

LLMs are going to get a headache with that repo.