"Lots" and "one that is widely accepted" are not the same thing at all,

Which is why I specifically said "There are widely accepted ones too."

There's no point in discussing intelligence without first nailing down which definition the interlocutors are using.

Absolutely.

There is value in recognizing the problem, not dismissing it. There are lots of defnitions, many of which are fuzzy, because people who want to argue that humans (or specific groups of humans) are "intelligent" while others are not generally have to define intelligence in a very ad hoc way to support their argument. People who are interested in studying the subject generally have reasonable definitions. The former don't like the latter's definitions because they fail to provide the desired absolute threshold they desire.

My point was that the LLMs claim to know everything but when they don't know something they just make things up where a human is likely to say "sorry, don't know,"

me pointing out that much of that "knowledge" is garbage. That's it.

Hm. Those don't quite sound like the same thing. It's like you've been caught saying something silly and you're engaging in a bit of revisionism. On a public, threaded message board with no edit button, no less. If one were being uncharitable one might point out that the quite human phenomenon of "digging yourself in deeper" often follows the one of making shit up rather than saying you don't know.

That's not entirely true. We don't have a good idea of how biological brains (simple or complicated) learn. There are some hints of how they might feed back error signals but we don't know in detail. There is the possibility, and people love to latch onto it, that brains are doing something that works better than gradient descent. There isn't really a good reason to believe that, and quite a few not to, except for handwavy comparisons like the summary makes.

Certainly the hardware is different, and that's where the obvious differences originate. The brain isn't really any more analog than your laptop though.

There are lots of definitions of what "intelligent" means. There are widely accepted ones too. There isn't really one where the human brain is, and everything else isn't.

Well, let's see... first of all, the brain is not digital

It's not?

My point was that the LLMs claim to know everything but when they don't know something they just make things up where a human is likely to say "sorry, don't know," or at least say "My guess is X" rather than "Oh yeah, it's X. Definitely X. What do you mean it doesn't exist? Oh, sorry, I just made that up. My bad."

That was good for a laugh this morning. Especially posted on Slashdot right after the guy claiming that what the brain does is uncomputable. Because... it is, that's why.

You can train an LLM to say it doesn't know, or even build pretty good objective confidence metrics. We don't, because people don't like it. We don't like it in humans either.

Everyone who's ever made that claim about something in the real world has been wrong. Why do you think this is different?

Free open source models that cost a tiny fraction as much for the same capabilities?

I can understand your stance if you think that AI is just a fad. But for those who believe that AI is going to kick off a new industrial revolution, it's the difference between that new industrial revolution being in the hands of people like Musk and Altman (who are profiting off of having scraped the commons), vs. the public as a whole.

What open source is capable of, when it comes to AI models, depends entirely on how much compute it has available.

Prompt:

Hey ChatGPT, who is being reasonable in this argument?

---
Rei: " ... We think of bees as mindless drones (literally, we took the very word!), but they're all individuals each acting on their own.... "

Synonymous Homonym: "Drones are specifically the male bees. Most bees are not drones."

Rei: "That's not what I said. I said that the word "drone", as in a mindless unthinking being, is derived from drones, as in male bees. I did not say "all bees are drone bees"."

Synonymous Homonym: "What you actually said:

We think of bees as mindless drones (literally, we took the very word!)

Suggesting that the word "drone" means "mindless bee.""

Response:

Rei is being more reasonable here.

The key issue is whether Rei's original sentence implied that all bees are drones or that "drone" as a metaphor comes from bees.

Rei wrote:

"We think of bees as mindless drones (literally, we took the very word!)"

The parenthetical "literally, we took the very word!" naturally reads as:

The figurative word drone (meaning a repetitive, unthinking worker) comes from the bee term drone.
Therefore, our cultural association of bees with "dronelike" behavior is reflected in language.

That is a claim about etymology, not about bee taxonomy.

Synonymous Homonym's first correction:

"Drones are specifically the male bees. Most bees are not drones."

is technically true, but it doesn't actually address what Rei was saying. Rei wasn't claiming that all bees are male drones.

Then Synonymous Homonym shifts to:

"Suggesting that the word 'drone' means 'mindless bee.'"

This doesn't accurately paraphrase Rei either. Rei never said the word means "mindless bee." The sentence only suggests that the metaphorical sense of "drone" is derived from the bee term.

If I were judging the exchange, I'd say:

Synonymous Homonym's factual statement is correct: drones are male bees.
But it's responding to a claim Rei didn't make, making it a misreading.
Rei's clarification directly matches the ordinary reading of the original sentence.

So the stronger interpretation is that Synonymous Homonym misunderstood the original remark, and Rei's correction is justified.

Or to put it more simply: you're attacking a straw man (a claim I never made) because you cannot defend your actual scientifically inaccurate comments about bees having half-hour memories.

You are confusing "forgetting" with "disruption".

I am not.

Your entire argument was that bees have short-term memories. The example you cited demonstrates precisely the opposite - that bee memory is if anything too durable>/i>, such that bees will refuse won't even bother searching for their hive if it has been moved, that they always go back to the spot where they remember it being last.

However, what you are choosing to ignore is that, once a bee has been accepted into a different hive, it becomes part of that hive

That's like claiming that if you move to a different city, you've "forgotten" where you used to live.

When a bee, due to a navigation error or its previous hive being gone, enters a different hive (drifting), it absorbs the pheromones of the new hive. It is now chemically marked as belonging to that hive. Chemicals are communication to bees, just as much as vocal commands are to you. Bees do not learn all of the other bees in their hive, they don't have some sort of rolodex. If you smell like hive A, you're allowed into hive A. If you smell like hive B, you're allowed into hive B. Bees don't particularly "care" which hive they're in; they have their own individual motives and drives, which simply involve being in "a" hive. Once they're marked as belonging to hive B, they can no longer enter hive A (at least not safely).

Note in the above what has nothing to do with any of this? Memory. It's just about smell. Memory is about where the hive can be found after foraging (which is also about memory) - and it remains, even after drifting (they'll continue to return to the same spot - again, even if the new hive is moved). Smell is about which hive you can enter. Or for a summary version:

1) A bee leaves the hive to go foraging

2) It remembers where the best spot to visit is (usually from having gone there before, but occasionally from having seen a waggle dance) and what flowers (shapes, smells, sizes, etc) will be yielding best there at what times of day, and what areas to NOT go to, where there may be threats. This information persists for days, weeks, or even the bee's entire life. It can target an area to an accuracy of a couple hundred meters, and then begins a search.

3) When done, it returns back to where it remembers that the hive should be (this memory is highly persistent, and can only be reset by an orientation flight.

4) The bee starts by using the sun and broad navigational features as with outbound flights to get to within a couple hundred to a few dozen meters (the "visual catchment area"), then gradually switches to small-scale features and searching. This is all based on memory.

5) For the final approach, the bee relies on a mix of sight (remembered), sound (generic), and smell. The latter is not a learned trait, it's "whatever you happen to smell like". While it's usually described as recognizing the smell of their sisters, that's not exactly right. The actual underlying mechanism not so much learning what something does smell like as it being unable to detect what they do smell like

The mechanism the same as how humans become unable to notice their own body odour or perfume: sensory adaptation. Because they're constantly smelling themselves, their brain learns to tune out their own smell. However, it doesn't tune out the smells of others. When they return to their own hive, the scent is something that they're adapted to tune out. But when they arrive at a different hive, they're hit with a scent that they're not adapted to, and that they can detect.

If you want to put it in human terms, the underlying mechanism is "this hive thinks you're stinky, that one doesn't smell you because you've all been around each other for so long".

If you want to call sensory adaptation "forgetting", then you're going to need to call human sensory adaptation "forgetting" as well. And again, none of this has anything to do with actual memory tasks, such as navigation and how to find the best flowers. Bee memory is exceptional with them.

Who said the word "small"?

(My personal hot take is that, both for copyright reasons ("Purpose and Character of the Use", aka for-profit, is a critical factor in determining copyright violation, such as from scraping), and general moral reciprocity argument (closed commercial models extracts profit from the commons without giving back), closed source trainers should fundamentally be required to give back to the commons in some meaningful way)

... some public ownership of compute, to offer out for grants to Open Source development, so that the commons can better keep up with the closed source providers?

I'm sorry that the genitals of people you neither know nor are dating bother you so much, dude.

Do you also spend this much time thinking about the genitals of children, like so many of your ilk?

Bluesky knows full well it's not operating a real federated service

Better tell that to Blacksky, Eurosky, etc.

The vast majority of people stay on the primary PDS, relay, etc namely because Bluesky hasn't proven itself to be some evil overlord pursuing insidious goals. If that were to ever occur, people would just migrate. Unlike with ActivityPub (Mastodon), ATProto allows for true migration. Your content isn't tied and linked to a specific server - it's more like a URL on an arbitrary domain, and you can just change the "domain" (the PDS). Everything is timestamped and cryptographically signed, so if you download a backup of your content, you can just reupload it somewhere else and it continues to remain linked into the whole ecosystem.

More to the point, primary Bluesky servers have gone down and third parties like Blacksky remained operational, very much demonstrating that the network is federated.

Also, re: this from the header:

" and by the end of October last year, it had reportedly seen a 40% drop in daily mobile active users over the past 12 months."

... is cherry picking. If you graph users, you'll see that - like most sites - new users tend to arrive in big "spikes", triggered either by events at other social media platforms, or major news cycles (such as elections). Most new users to a site are not "sticky". Some drop off in days, some in weeks, some in months, etc, but this slowly levels out, and the rest are "sticky". With Bluesky, usually half or so of new users stick around, which is an unusually high percentage. If you measure from a new-arrivals spike, of course you see a "dropoff", but you see that for any site. The question is, how is the long-term trend of users that stick around? If you cancel out the spike pattern, Bluesky has a long-term population of around 600k daily posters / 1M daily likers / 300k daily followers.

What you can say is there haven't been any big new user spikes since late 2024 / early 2025. That said, there kinda was some serious news going on in late 2024 / early 2025....