barnacs

It turns out LDL has different subfractions based on size. Normally, the liver produces large ones which shuttle fat around the bloodstream, lose their payload and shrink then the liver reabsorbs them and the cycle continues.

The problem appears to begin when these LDL molecules hang around for too long. They are damaged by external factors in the bloodstream (eg. glycation due to high blood glucose, oxidation due to fragile polyunsaturated fats) and not recognized anymore for recycling. That's when they can start forming macrophages.

tl;dr LDL by itself is an incorrect measure of cardiovascular risk. Big pharma still profits by selling everyone LDL reducing drugs.

one authentic source: https://www.youtube.com/watch?v=fVLZA0qp-wc

Mathematics is a language. If anything, it's much more well defined and formal than most others. Train on enough examples and statistical autocomplete gets you places. I'm surprised how anyone would even consider this intelligence?

I hold my stance that LLMs are stochastic parrots.

Making the parrots ever more complex and training on ever more data produced by intelligent, creative beings may make them more useful or convincing but does at no point give rise to intelligence or creativity.

Similar to your contractor example, I like to think of the useful way to use LLMs as doing "grep -C" (show lines surrounding the match).

You provide some context you are looking to expand on and let the LLM autocomplete provide you with relevant terms and concepts that have historically come up in that context and may or may not be actually useful. Then you do due diligence to learn about those related concepts, determine their applicability and distill it all down into a solution.

It will be interesting to see when ads and propaganda are seamlessly (but purposefully) integrated into the output of LLMs. Who will decide what you should want or think? Will you even notice?

As for now, autocomplete is only as good as the training data. Once humanity collectively stop being autonomous beings and generating novel ideas, it all comes to a halt. LLM suggested ideas and preferences are nothing more than some mashup/average of what came before. The ability to actually think may become a rare treasure.

I remember making a Qt app for a friend that would upload dropped files via ftp and copy the link to the clipboard. Good old days!

This is getting way too dystopian for my taste. People in the know need to stop pushing the narrative that this is somehow anything more than statistical autocomplete.

I don't think most people have been inconvenienced enough yet. ID verification is invasive enough and should cause enough friction to push another bunch over the edge.

Hopefully this will give yet another push towards decentralized, open source services. Platforms where noone and everyone is responsible and the state does not get to decide the rules.

Just a reminder: LLMs are statistical models that predict the next token based on preceeding tokens. They have no feelings, goals, relationships, life experience, understanding of the human condition and so on. Treat them accordingly.

Get with the newspeak, it's called "sideloading" now and your corporate overlords get to dictate the terms.

In the end, your newly renamed "department of war" is just going to waste a bunch of your taxpayer money to purchase some useless overpriced tech from their cronies. My symphaties to all citizens.

I highly recommend self-hosting a meta-search engine to get away from any single provider without losing the benefits of either. I have been using searx[0] happily for years.

[0] https://github.com/searxng/searxng

I don't think using git should necessarily be taught as a part of a CS education. Any self respecting engineer will be capable and have the curiosity and motivation to dig into it on their own. CS should give them the prereqs to do so, such as hashing, graphs, trees, etc.

If the fields of a structure or the string keys of an untyped map don't match then you don't have compatibility either way. The same is not true for restricting the set of valid values.

edit: To put it differently: To possibly be compatible with the nested "Circle" map, you need to know it is supposed to have a "Radius" key that is supposed to be a float. Type definitions just make this explicit. But just because your "Radius" can't be 0, you shouldn't make it incompatible with everything else operating on floats in general.

Every time you introduce a type for a "value invariant" you lose compatibility and force others to make cumbersome type conversions.

To me, invalid values are best expressed with optional error returns along with the value that are part of the function signature. Types are best used to only encode information about the hierarchy of structures composed of primitive types. They help define and navigate the representation of composite things as opposed to just having dynamic nested maps of arbitrary strings.

Agreed. "Be the change you want to see". In this case, both as a user and a professional.

stop using Google, Amazon, Meta products

That's the easy part. What do you do about stuff like face recognition and cameras everywhere? Should you hide your face every time you go out? Should you not speak because there might be a mic around picking up your voice?

This is only going to get worse. We can't trust companies or governments to respect our privacy. We can't trust each other to keep the data recorded by our devices private.

It seems like the fight for privacy is a lost cause. What do we do?

Pretty much every single country in the world has taken on more debt than they can serve. And the 2008 crisis wasn't triggered by Greece either. The private creditors should have taken the loss. And that holds true for the rest of the world.

Right, but next time your data may be threatened by some European idiot rather than one from the US.

To quote from my other comment:

In my view, data can only be protected by its rightful owner. And for that, we need education, not regulation.

I happen to live in the EU so I may have a slight clue what I'm talking about.

But if you want an authority on the subject, look up Yanis Varoufakis and how sovereignty and democracy worked out for Greece when shit hit the fan.

other important aspects

like what?

As a European citizen, I can assure you, my options are getting ever more limited. Several global companies have kicked me off their platforms recently due to all the regulations they can't be bothered with. Those that make an effort to comply are by default required to submit to the EU surveillance system. At the same time, I have no illusions that any of this would somehow protect my data from the NSA and the like.

In my view, data can only be protected by its rightful owner. And for that, we need education, not regulation.

As if the surveillance and regulation by the unelected EU bureaucrats was any better for the European citizens...

I remember the good old days when a "vape" was just a sturdy housing for a rechargable battery, some heating wire, cotton and juice. The power was determined by the resistance of the coils you built. Those things would last forever.

I'm looking for a language optimized for human use given the fundamental architectural changes in computing in the last 50 years. That way we could skip both the boilerplate and the LLMs generating boilerplate.

That's probably why a boundary (like MCP) is useful. Imagine maintaining the critical application logic the "old fashioned way" and exposing a MCP-like interface to the users so that they can have their LLM generate whatever UI they like, even realtime on the fly as they are engaging with the application. It's a win-win in my mind.

the ability to produce useful output beyond the sum total of past experience and present (sensory) input.

Isn't that what mathematical extrapolation or statistical inference does? To me, that's not even close to intelligence.

Thinking is using what you know to come to a logical conclusion

What LLMs do is using what they have _seen_ to come to a _statistical_ conclusion. Just like a complex statistical weather forecasting model. I have never heard anyone argue that such models would "know" about weather phenomena and reason about the implications to come to a "logical" conclusion.

I guess our definition of "thinking" is just very different.

Yes, humans are also capable of learning in a similar fashion and imitating, even extrapolating from a learned function. But I wouldn't call that intelligent, thinking behavior, even if performed by a human.

But no human would ever perform like that, without trying to intuitively understand the motivations of the humans they learned from, and naturally intermingling the performance with their own motivations.