Even the question of “who” is a fascinating deep dive in and of itself. Consciousness as an emergent property implies that your gut microbiome is part of the “who” doing the thinking in the first place :))

So, first of all, thank you for the cogent attempt at responding. We may disagree, but I sincerely respect the effort you put into the comment.

The specific part that I thought seemed like a pretty big claim was that human brains are “simply” more complex neural networks and that the outputs are based strictly on training data.

Is it not well established that animals learn and use reward circuitry like the role of dopamine in neuromodulation?

While true, this is way too reductive to be a one to one comparison with LLMs. Humans have genetic instinct and body-mind connection that isn’t cleanly mappable onto a neural network. For example, biologists are only just now scraping the surface of the link between the brain and the gut microbiome, which plays a much larger role on cognition than previously thought.

Another example where the brain = neural network model breaks down is the fact that the two hemispheres are much more separated than previously thought. So much so that some neuroscientists are saying that each person has, in effect, 2 different brains with 2 different personalities that communicate via the corpus callosum.

There’s many more examples I could bring up, but my core point is that the analogy of neural network = brain is just that, a simplistic analogy, on the same level as thinking about gravity only as “the force that pushes you downwards”.

To say that we fully understand the brain, to the point where we can even make a model of a mosquito’s brain (220,000 neurons), I think is mistaken. I’m not saying we’ll never understand the brain enough to attempt such a thing, I’m just saying that drawing a casual equivalence between mammalian brains and neural networks is woefully inadequate.

That’s a strong claim. Got an academic paper to back that up?

This is why I strictly refer to these things as LLMs. That’s what they are.

I’m happy with the Oxford definition: “the ability to acquire and apply knowledge and skills”.

LLMs don’t have knowledge as they don’t actually understand anything. They are algorithmic response generators that apply scores to tokens, and spit out the highest scoring token considering all previous tokens.

If asked to answer 10*5, they can’t reason through the math. They can only recognize 10, * and 5 as tokens in the training data that is usually followed by the 50 token. Thus, 50 is the highest scoring token, and is the answer it will choose. Things get more interesting when you ask questions that aren’t in the training data. If it has nothing more direct to copy from, it will regurgitate a sequence of tokens that sounds as close as possible to something in the training data: thus a hallucination.

Ah, I misunderstood then, sorry. But still, even with all the investment in the world, LLM is a bubble waiting to burst. I have a hunch we will see truly world-altering technology in the next ~20 years (the kind that’d put huge swathes of people out of work, as you describe), but this ain’t it.

There’s an upper ceiling on capability though, and we’re pretty close to it with LLMs. True artificial intelligence would change the world drastically, but LLMs aren’t the path to it.

Actually, that’s not the real reason patents are public. The reason is to allow everyone to freely use the patent after the expiry.

The tradeoff is supposed to be the inventor gets exclusive use for a decade in exchange for detailing exactly how the thing works for everyone else.

For the families who can afford it, daycare is the replacement.

“It takes a village to raise a child” is an old expression for a reason. Historically (EDIT: And today in most of the world), parents wouldn’t take care of their kids 24/7. They would have parents, siblings, neighbours and friends to help share the load.

The idea that parents and parents alone do 100% of everything to raise a child is a very modern western thing.

I don’t know the details of this app, but if it’s specifically US streets and notes on households there, then GDPR does not apply, as they’re not mapping EU households. GDPR is only invoked if the personal information of Europeans is at risk.

I hate the term intellectual property. It’s a word used to describe vastly different concepts with vastly different legal backgrounds and problems.

Copyright is theoretically a good thing, giving an artist or writer the time to profit from their work before the work becomes public domain, incentivizing the work. The current international agreements around it are absolutely bonkers thanks to Disney. The fact that the copyright persists after death, let alone for a century, is complete madness. The artist obviously can’t profit from their work after they’re dead. It’s an absolute shameless cash grab that destroys culture.

Patents are also theoretically a good thing, allowing companies to release specifications of machines that allow for 10 years of exclusive use. Without patents, companies would hide their designs as trade secrets. It guarantees that after a decade, the designs will be publicly available for anyone to see. They need to be much more heavily restricted in what you can put patents on though. Patenting a specific machine design is fine, patenting molecules or math breaks the entire system. Software patents are blatantly absurd and broken.

EDIT: Should also mention that 3D printers are a patent success story of the system working as intended. Patented in 1986, the inventor made good money making expensive machines with his own company. In 1996, the patent expired and we had an explosion of competing machines, getting ever cheaper and more effective. Everybody won. The inventor made bank for his decade of exclusivity, and then everyone benefited from the design being public domain, free for everyone to use.

Trade secrets, the protection of specific recipes, client lists and strategies, can be abused to protect companies against disclosing information that may be very pertinent to their customers and governments. The Coca-Cola recipe or lists of clients as a trade secret is fine imho, but they can also abuse trade secret law to hide systems that lie about your car’s emissions.

Trademarks help protect consumers against knockoff brands that pretend to be what they’re not. This is the least abused type of “IP”. This doesn’t mean there aren’t bad actors out there registering tons of different trademarks to squat on those designs & names, hoping to force a new company to pay up to use the name. Trademark squatting could theoretically be solved by annulling the trademark if the company isn’t actively using it. Trademarks are currently much too easy to maintain.

All of this to say, lumping all of these different laws into “IP” is not useful at all when talking about the goals of the different legislations, what they’re trying to do, and how they fail.

I think the best counter to this is to consider the zero learning state. A language model or art model without any training data at all will output static, basically. Random noise.

A group of humans socially isolated from the rest of the world will independently create art and music. It has happened an uncountable number of times. It seems to be a fairly automatic emergent property of human societies.

With that being the case, we can safely say that however creativity works, it’s not merely compositing things we’ve seen or heard before.

I really don’t understand this whole “learning” thing that everybody claims these models are doing.

A Markov chain algorithm with different inputs of text and the output of the next predicted word isn’t colloquially called “learning”, yet it’s fundamentally the same process, just less sophisticated.

They take input, apply a statistical model to it, generate output derived from the input. Humans have creativity, lateral thinking and the ability to understand context and meaning. Most importantly, with art and creative writing, they’re trying to express something.

“AI” has none of these things, just a probability for which token goes next considering which tokens are there already.

To add to what others have said, the real source of load is browsing. It’s much better than a few months ago, but it is still the case that users actively on the web frontend is the biggest load.

If you are actively using a single digit number of accounts simultaneously, no issue. A bot farm using 100,000 users simultaneously is a big problem.

Would need some kind of server setting to specify categorisation. Maybe open a GitHub issue for a server setting to do this, and then it could be displayed in the join-lemmy site

Copy-pasting from my other answer here:

Trite answer: When it’s done

More in-depth answer: Currently there’s no set date. It depends on how quickly they can tear out all the WebSockets code and replace it with simple HTTP (that’s the BIG change, will fix a lot of different things), and then test those changes. The hot_rank fix has already been merged, that’s done, but they want a stable, cohesive release with all the good stuff.

Current estimations I’ve seen range from 1-2 weeks, but it all depends on how fast they can get it coded and tested.

Known bug, will be fixed in 0.18. Certain edge cases cause the hot_rank to not decrease, meaning they never leave the front page.

For the time being, yeah. There are multiple github issues open with potential aids to community discoverability and usability, but between lemmy.ml being on fire (recently quenched after moving server… for now), trying to rip WebSockets out of the code before instances hit levels where that’ll cripple performance, and dealing with the ~150 other issues per project that have been opened this week, they’re… a little preoccupied.

Nope. You would just have to subscribe to each community you want before it can hit your All feed. Communities don’t get pulled in until at least 1 person on the instance subscribes.