Genuine apologies for being combative. I got a little worked up at the insinuation that I am somehow trying to dispose of ethics. I do think that you could stand to be less combative yourself; even if you think I have absolutely nothing to offer you, you must think there's some value to having this discussion if you've gone this long, and I have to choose to continue as well.

I did in fact read the Wolfram article when it was posted here a few weeks ago. I appreciated the more straightforward overview of the architecture in the link I shared without the expository information. Referring to a model like ChatGPT as a lookup-table is obfuscating, and that claim is not made by an experts, including in the Wolfram article. I would maybe permit that it's something like a randomized lookup table, in that if its randomization happened the same way every time then any sequence of inputs and outputs would result in the same eventual output. But that's ignoring both the randomization and the feedback loop of rereading of its own outputs, not to mention the presence of a neural net in the model.

Lookup tables aren't mentioned in the article I linked at all. There's plenty of description of the tokenization and encoding of input data, which is superficially similar. The presence of the neural net, the attention scheme, and the encoding of the relative position of words in an input phrase allows for interaction between all the words in a sequence, and that interaction is tempered by all the sequences in its training data.

I don't mean to skip over the fact that ChatGPT itself is not learning anymore. I agree that it's important, but I am discussing a hypothetical scenario that will become reality before long. It's perfectly capable of continuously learning from its conversations (although highly inefficiently) should OpenAI choose to do that, although there's obvious logistical reasons that they don't want a bunch of random people inputting its training data for them.

To say that LLMs are entirely Markov chains is a misapprehension. An LLM like ChatGPT is not memoryless even in its current form, because of its internal feedback loop. If you would instead argue that the state we're referring to is not the most recent statement but instead the full conversation, then I would argue a human speaker IS comparable to a Markov chain in any particular conversation. The human speaker obviously differs in that they can both update their "model" over the course of the conversation and carry those updates forward into future conversations, but the hurdles for a computer model accomplishing that are logistical, not inherent. Am I missing something there? Even Wolfram says:

What ChatGPT does in generating text is very impressive—and the results are usually very much like what we humans would produce. So does this mean ChatGPT is working like a brain? Its underlying artificial-neural-net structure was ultimately modeled on an idealization of the brain. And it seems quite likely that when we humans generate language many aspects of what’s going on are quite similar.

When it comes to training (AKA learning) the different “hardware” of the brain and of current computers (as well as, perhaps, some undeveloped algorithmic ideas) forces ChatGPT to use a strategy that’s probably rather different (and in some ways much less efficient) than the brain. And there’s something else as well: unlike even in typical algorithmic computation, ChatGPT doesn’t internally “have loops” or “recompute on data”. And that inevitably limits its computational capability—even with respect to current computers, but definitely with respect to the brain.

It’s not clear how to “fix that” and still maintain the ability to train the system with reasonable efficiency. But to do so will presumably allow a future ChatGPT to do even more “brain-like things”.

ChatGPT is not a brain, but I don't think you can reasonably claim it has nothing in common with one. And I still don't see any reason why a brain-like model could not be created.

I'll leave aside the ethical questions (no beliefs here!) since I don't think that we'll come to an agreement. I trust that you'll be kind to my dog, and that's all that really counts.

- We only have a rough idea how brains think

- We have an explicit and complete idea how LLMs work

- What we do know does not match how LLMs work

- Attempting to run an LLM the way the brain works fails

I don't understand how this isn't going back on your earlier claim that complexity doesn't equal intelligence? Our lack of understanding of how the brain works doesn't provide any evidence of it possessing some supernatural faculty that a computer couldn't (with currently non-existent but feasible tech) replicate. And your fourth claim here is demonstrably false: even ChatGPT is based on a neural net. Obviously this does not mean it is working the way the brain works, but it is absolutely complex enough to exhibit emergent behaviors that we could never make sense of.

In fairness to you, ChatGPT could be written as a set of instructions, but that's only because it's no longer learning in its current state. The same could be said of a snapshot of a human mind.

Atlas really is as simple as a bunch of code.

You're responding to a point that I haven't made here, and drawing a false equivalence between discussions of hardware vs software. The discussion as we began it was to imagine a sophisticated learning model placed inside hardware that allows it to gather information about the world around it, the "access to real-world, embodied referents" mentioned in the article.

Okay, where do you draw the line? 'cuz the line has to be drawn.

You have to draw the line too! Once again, I am not arguing that ChatGPT is alive, nor that it is conscious. I do believe that being alive and being conscious are not mutually exclusive, due to my beliefs about consciousness, which is clearly something we are not going to agree on.

I find that ethical individuals have the ability to query their ethics, and unethical individuals have the ability to query the existence of ethics.

If you think that what I'm doing is the latter and not the former then it might not even matter what I'm typing here. I'm concerned that our ethical system is largely based on the idea of being nice to humans and things that are sufficiently like humans, precisely because it leads to Type 1 and Type 2 errors of being cruel to dogs or falling in love with chatbots respectively. And referring to myths in your "facts don't care about your feelings" argument walks a really bizarre line.

Yeah and taking an astrological view of the solar system allows for the position of Mars to influence my luck. That doesn't make astrology factual, accurate or useful.

This is a ridiculous comparison, as astrology is provably false and any theory of consciousness is necessarily unprovable one way or another. Certainly some or more plausible than others, but you need to subscribe to some idea of what consciousness is to even begin to have this discussion, and you're only espousing negative views on consciousness save for fairy-tale appeals to your feelings that humans have got it and computers don't.

Panpsychism appeals to me because I agree that it's obvious that humans are conscious, and it's obvious that my dog is conscious, and it sure seems like chickens are conscious but it starts to get fuzzy around there. It would be ridiculous to lay down some arbitrary line between chickens and another species, or to decide that some particularly gifted chickens are conscious but not others. It seems equally strange to me to finally decide that viruses are conscious, but deny consciousness to other self replicating processes just because viruses have DNA.

What seems most sensible to me is to then conclude that consciousness is not binary, but something all things have to some extent. This is based on how I "feel," but so is every other concept of what consciousness is. But that's what this whole discussion is about.

Atlas sees position, velocity and force. That's it.

I guess what I'm getting at is the idea that one could take an equally reductionist view of the human mind, it's just that our brains are optimized beyond the point of being possible to interpret. We see, we hear, we feel, we smell, and all that information is plugged into a complex logical system, along with our memories, categories, and any evolved instincts to dictate our actions. And I think that the systems you're describing in the paragraph preceding this quote don't lack computer analogs. If you'd like to get into the technical weeds of that I'd be interested in pursuing it.

researchers saying "it can't do this" and the general public going "but it sure looks like it does, clearly you're wrong because I want this to be true."

What about Christopher Manning, the other computational linguist mentioned in this article?

I should be more precise about what I'm trying to say, cause I'm certainly not one of those nuts who believes that LaMDA or ChatGPT are sentient. I'm engaging with you because you seem knowledgeable on a subject which I find my self at odds with many really smart people that I agree with on lots of other stuff.

The central disconnect that I'm interested in learning more about is the idea of humans as exceptional in deserving of our respect and compassion. In this article this stance is presented as an almost a priori abhorrent view of humanity, one that should be met with a sigh and look towards the camera. I'm gathering that you agree with Bender in wanting to posit humanity as an axiom.

I kind of take a panpsychist view on consciousness, and fundamentally what I'm arguing is that that perspective allows for manmade constructs like computer programs to attain some degree of consciousness. I'm curious if/where the disconnect (touching David's finger) arises there in your view, or if we're simply arguing from different sets of premises.

With all that being said, I can absolutely understand a lack of interest in engaging on this topic or seeing it as intellectually frivolous given that we aren't even able to convince people to treat other humans with respect.

I agree with a lot of what you're saying, but wonder about the claim that we think "ramp, ramp, box" watching this video. I think that discussions about what qualifies as artificial intelligence benefit from more careful consideration of what we consider to be our own intelligence. We do think ramp, ramp, box, but we can only think that after taking in a ton of visual data and attempting to make sense of it with our previous experience. We have a heuristic for ramp developed over a similar feedback loop, like gently correcting your kid when they call a raccoon "kitty". What if an extension of LLMs was developed that made use of long term memory and the ability to generate new categories? Why do you think that they'll never be able to work that way? And moreover, what is the model trying and failing to replicate in your mind? An individual human speaker, or some sort of ideal speaker-listener, or something else?

distressingly enjoyable to listen to!

A favorite of mine :)

Someone asked me what my favorite album of 2019 was so far, and I realized I've been keeping terrible track of new releases! The two that occurred to me right away were UFOF, the new album by Big Thief, and a very recent EP by a neat band called floral print, but I'm curious of other people's favorites. What am I missing out on?

Neat stuff

might enjoy this!

Amazing song, amazing album, and one of the best songwriters of all time. Awesome to hear this again!

yay

Been listening to this album a lot lately. Very dreamlike for me

Love John Cale. Another beautiful track from the same album:

Listening to Lou Reed's "Live: Take No Prisoners." It's pretty great, for most of the songs he doesn't even sing the lyrics, just rants about something for like 10 minutes.

Also listening to an album by a weird New York band/collective called PC Worship. Allegedly, they build their own instruments out of literal garbage lying around their apartments. They're pretty good at any rate.

https://pcworship.bandcamp.com/track/odd

And also listening to Tyvek, a power punk band from Detroit that's REALLY stoked to be from Detroit. I think they're sweet.

"Rocket from the Tombs was always doomed," said singer David Thomas. "Everything from Cleveland was doomed. Rocket from the Tombs is totally inconsequential and irrelevant. Pere Ubu is totally inconsequential and irrelevant. That is the power of Cleveland. Embrace, my brothers, the utter futility of ambition and desire. Your only reward is a genuine shot at being the best. The caveat is that no one but your brothers will ever know it. That's the deal we agreed to."

Been listening to a lot of Bill Fox, a legendary Cleveland artist. The Mice's first and only full length LP, Scooter, got a 5 star rating from NME, which was a big fuckin deal at the time, especially for a Cleveland band. They were all set up to go on tour to Europe and get shit kicked off, and then the morning they were supposed to fly to Liverpool, Bill breaks up the band for some unknown reason. He had a short solo career, and then a few years later he just stopped playing music completely. Since then he and The Mice have faded into obscurity outside of Ohio. Very interesting character.

http://www.joehagan.net/images/pdf/fox.pdf

I'm glad someone else liked it as much as I did!

Ah, that's sweet. Dream collab

I remember the video of Tyler and Earl meeting him, and Earl was on the verge of tears

Yup yup yup. Definitely a "we're all gonna die, and I feel fine" kind of feeling, which is sorta positive? as morbid as it sounds

What fantastic use of color! Like an impressionist Pieter Bruegel. Really beautiful!!

It kinda reminds me of The Stranger, how the unlikely glimmer of hope is what makes your impending doom so much harder to deal with. This won't change the outcome of the election. PLEASE just chop off my head already.

But not nearly as easy to appreciate as your preantepenultimate post, about labor unions and DJT.

Yeah, I took a few liberties with the title. Sue me.

If you're not careful, Donald Trump might do just that.

hubski reader app double posted T_T

I love this! Definitely looking into this band when I get home. Thanks for the rec

I love this! Definitely looking into this band when I get home. Thanks for the rec

I'm glad you asked! The whole album Denton After Sunset is really great, but Landlocked State is definitely my favorite track. The frontman for TNCK, Andrew Savage, is now in a more well known band called Parquet Courts, who I adore. One of the Parquet Courts songs I've been coming back to lately and that has a pretty similar sound to Landlocked State is Instant Disassembly:

It's uniquely terrifying that people like this exist, and uniquely 21st century that they're able to find like minded people and meet up.