stingraycharles

They also famously hate nvidia since 2008 and would prefer to use TPUs, at least historically

Should add China to the list of pretty much all of them as well. It’s a silly take.

“Google doesn't care that much about alignment”

I don’t think this is necessarily true, did we all forget how much Google cared about alignment that their AI wasn’t able to render a white polar bear?

As always, benchmarks rarely paint the whole picture. It also seems like this article is somewhat biased, eg when Fable and Kimi are close but Fable wins it’s “dead heat”, but when Kimi wins it’s “Kimi wins”. GPT 5.6 seems to be missing as well.

I am really eager to give Kimi K3 a try, but I’ll reserve my judgement until I’ve worked with it for at least a few days.

We only get Copilot. I’m not very happy.

Fable is still the same model, it’s still a great model, and to be honest all these articles writing and speculating on how the LLM industry is going to evolve are not that insightful nor interesting.

I don’t think one should pay much attention to them.

But we're talking about AI writing and/or influencing other people's online discourse, not?

I don’t think this is correct, as we’re talking about all online discourse being influenced by AI, and you don’t know what to trust anymore, rather than generating AI output yourself (which is what you are referring to).

I think in this context, they mean seeking the primary source of the information by hand.

I mean, you have those kind of luxuries even in the poorest of countries in Asia, it’s just that there’s still a huge discrepancy between rich and poor, city vs countryside.

It’s not difficult to find areas in all these countries that are significantly less developed than Spain/Portugal’s underdeveloped areas. It’s just not as black and white as you seem to suggest.

(I come from EU but have been living in various countries in Asia for over a decade)

Yes, but a grand prize result being awarded to an AI slop submission is not it. It deteriorates the legitimacy of the whole contest if all that matters is convincing an AI rather than critical reviewers.

“It seems Google is repeating its mistakes of chat apps (they had gazillions of those at some point) with AI.”

It’s just a product of how Google is organized, it’s very distributed with less top down product management than other organizations.

It’s interesting that they still haven’t seen reason enough to fix this, I can only assume it works well for them in other ways.

“Google is failing hard at convincing their own users that they are as good.”

It’s interesting as they very quickly caught up after their own “code red” a few years back, and ever since seem to be going full Google and fragmenting themselves into irrelevance.

They do have their tensor processors, though, so they have a huge advantage there.

Since we’re bikeshedding: that’s not what yak shaving means.

Murder and speeding both involve violations of the law, there’s a big difference though.

There is no evidence for these types of claims. They likely need to retain data for legal purposes (I think all of them are under injunctions from court cases), but there’s no way they will be breaching contracts with all these enterprises just for a little bit of data. Those contracts are their lifeline.

Why would anyone want to risk any money going any way with these stocks?

Tesla isn’t behaving rationally, there is no way to tell whether SpaceX will behave rationally or not.

It seems like it’s becoming somewhat more rational, but all these stocks just seem to be incomprehensible to me.

Can’t believe the abstract has an em dash, a “not X but Y” and a “rule of three” in the first sentence. This is ridiculous.

I don’t know what’s up with this thread, for some reason people insist that language packages and distro packages are nearly the same.

There must be a lot of packages for languages that are useful that simply never make it into Debian stable. “Where’s the patch” really isn’t an answer for this.

There’s plenty of evidence that people used to sleep very early, and have a period of activity in the night, before having a second period of sleep.

I follow the same schedule as you do, and also work remotely, and usually take a single 1-2h nap somewhere between 12pm - 3pm. It makes me have two moments of “morning productivity”, which works very well for me.

In the end, listen to your body.

Because they both have very different management processes (centralized vs distributed) and release cycles. They solve different problems.

It’s not about quantity but target audience and use case. They solve different problems.

They are literally solving the same problem

No they’re not. Distro packages cater to end users and have very different release cycles and maintenance processes.

Distro packages are managed top-down (pushed by maintainers), while language packages are managed bottom-up (pushed by authors), so to say.

I think that’s up to debate, and my point is that debating whether free software counts as “supply” or not is really not that interesting.

What do you mean? AI is being trained on all data available, so obviously it’s being trained on data from 2017/2018 as well.

Or do you mean that the patterns that AI is showing today were already present in communication around that time?

That seems obvious as well, as AI has a lot of repetitive patterns that come from all kinds of periods, of which “load-bearing” is just one.

I mean, that’s just arguing over whether or not the definition of “supply” implies “compensation”, which isn’t very interesting imho.

The grandparent’s point remains the same, the software ecosystem and its supply chain or however you want to call it is a hot mess.

Nah, it’s full of LLM-isms. There’s a lot of focus on “there’s not just one profile, there are five”, “it’s not A, but B!” and other “X is to Y what A is to B” style content that obsesses over comparing different concepts.

Well for full disclosure, I lead a team of forward deployed engineers at a database company. The role typically means that our engineers are embedded within the customer for extended period of times, and they work on basically devops, software engineering + some more traditional solution architecting, which is basically what the article describes.

They use LLMs in similar ways that regular engineers use. This is an engineering role, not a product / project management role. I don’t think this role is anything super special that will be revolutionized in any different way than that other engineering roles are affected.

In the end their value add is that they’re both embedded within the customer’s and our company, they’re our eyes and ears within the customer. Their purpose is not to make sales demos, their purpose is to make our software actually work properly for the customer’s needs.

LLMs don't really have anything to do with this, other than LLMs being useful for pretty much any (tech) role.

Yeah this is an extremely poorly written article. They didn’t even bother to add a “rewrite this article to make it sound less than AI”.

It’s also using a bazillion words to make a point that could be summed up in a single paragraph: there’s a huge variance in the number of tokens required to encode the same content, with code leading the charts.

To be fair, most of this was already known, and Anthropic communicated very clearly about the different tokenizer they started using.

Their compute is also mostly 1:1 correlated to the number of tokens, so I don’t believe in the conspiracy that this is just to inflate prices.

It was way too technical to disqualify it as not being technical writing. Writing can serve multiple purposes.