Coleridge's "Limbo" – Another ring composition?

This is from July 31, 2016. I'm bumping it to the top because we need a reminder of poetry.

While wandering in cyberspace, as I am sometimes wont to do, I came across a most curious, and thus alluring, blog: Samuel Taylor Bloggeridge. What a peculiar name, thought I, not realizing that this particular locus solus belonged to my old buddy from The Valve, Adam Roberts. Really though, thinking in retrospect, who else could it be? "Bloggeridge" indeed!

But then, Ted Nelson coined the term and birthed the concept of hypertext as an antidote to the (drug-induced) memory loss that (supposedly) gave us "Kubla Khan", did he not? He even called it Project Xanadu, long before Olivia Newton-John sang the song and did the movie (which has Gene Kelley's last film performance). One can only imagine how STC would tear-ass about cyberspace were he alive today.

In any event, and getting back on track, I'd come across a post Adam had written about "Limbo", one of STC's many poems that I have not studied. He wrote about it in a most tempting way. Thus:

The first thing to say about 'Limbo' is that its 28 lines divide into a central 12-lines section flanked by two paired 8-line sections, the first a kind of introduction, the second a sort of summary. What the two outer passages frame is a central image of 'human Time' as an old, blind man staring at the moon.

You can bet my ears pricked up when I read that. Like a hound on point I was. A bit later: “The circle in the centre of this circling poem has become a 'circumambience' prison-wall that imprisons ('immures', 'enthrals').” Be still my heart! "Circle", "circling"!? Could it be? Is this another ring-composition? What about the chiasmatic line 20 (which Adam fingers as consisting of monosyllables): "He seems to gaze at that which seems to gaze on him!" 

I've reset the poem to bring out the ring-form possibility. Give it a read and see what you think.

'Tis a strange place, this Limbo!—not a Place,
Yet name it so;—where Time and weary Space
Fettered from flight, with night-mare sense of fleeing,
Strive for their last crepuscular half-being;—
   Lank Space, and scytheless Time with branny hands
   Barren and soundless as the measuring sands,
   Not mark'd by flit of Shades,—unmeaning they
   As moonlight on the dial of the day!
      But that is lovely—looks like Human Time,—
      An Old Man with a steady look sublime,
      That stops his earthly task to watch the skies;
      But he is blind—a Statue hath such eyes;—
         Yet having moonward turn'd his face by chance,
         Gazes the orb with moon-like countenance,
         With scant white hairs, with foretop bald and high,
         He gazes still,—his eyeless face all eye;—
      As 'twere an organ full of silent sight,
      His whole face seemeth to rejoice in light!
      Lip touching lip, all moveless, bust and limb—
      He seems to gaze at that which seems to gaze on him!
   No such sweet sights doth Limbo den immure,
   Wall'd round, and made a spirit-jail secure,
   By the mere horror of blank Naught-at-all,
   Whose circumambience doth these ghosts enthral.
A lurid thought is growthless, dull Privation,
Yet that is but a Purgatory curse;
Hell knows a fear far worse,
A fear—a future state;—'tis positive Negation!

Alternatively:

'Tis a strange place, this Limbo!—not a Place,
Yet name it so;—where Time and weary Space
Fettered from flight, with night-mare sense of fleeing,
Strive for their last crepuscular half-being;—

Lank Space, and scytheless Time with branny hands
Barren and soundless as the measuring sands,
Not mark'd by flit of Shades,—unmeaning they
As moonlight on the dial of the day!

But that is lovely—looks like Human Time,—
An Old Man with a steady look sublime,
That stops his earthly task to watch the skies;
But he is blind—a Statue hath such eyes;—

Yet having moonward turn'd his face by chance,
Gazes the orb with moon-like countenance,
With scant white hairs, with foretop bald and high,
He gazes still,—his eyeless face all eye;—

As 'twere an organ full of silent sight,
His whole face seemeth to rejoice in light!
Lip touching lip, all moveless, bust and limb—
He seems to gaze at that which seems to gaze on him!

No such sweet sights doth Limbo den immure,
Wall'd round, and made a spirit-jail secure,
By the mere horror of blank Naught-at-all,
Whose circumambience doth these ghosts enthral.

A lurid thought is growthless, dull Privation,
Yet that is but a Purgatory curse;
Hell knows a fear far worse,
A fear—a future state;—'tis positive Negation!

The first setting implies a nested structure:

       ( 1 ( 2 (3 ( X ) 3 ) 2 ) 1 )

The second implies a travel structure:

       1 > 2 > 3 > X > 3 > 2 > 1

This, obviously, requires some explanation. But not now.

Saucer pad in Hoboken's new park

High Castle and Inner Truth

Another bump. This one on general principle. The link to The Valve takes you to the Wayback Machine and The Valve is no longer on the web.

 * * * * *

With The Man in the High Castle streaming on Amazon Prime I figure it’s time to repost another piece from The Valve. This one appeared on March 6, 2008, and generated an interesting discussion. I recommend it to you.

I've just finished reading Philip K. Dick's 1962 The Man in the High Castle and find it pretty interesting – and sophisticated as well. It seems to be a meditation on the nature of history and of fiction, but without the self-conscious hijinks one finds in, say, John Barth. As you may know, it's set in an alternative history of the mid-20th century in which Germany and Japan win World War II. The eastern seaboard of the USA becomes German territory while the western seaboard becomes Japanese territory. The central area remains more-or-less independent.

One Hawthorne Abendsen lives in that central area, near Cheyenne. He's the man in the high castle, though he does not, in fact, live in a high castle. Nor is he a central character in the novel, though he's quite important. He's written a novel entitled The Grasshopper Lies Heavy that is quite popular in what remains of the USA, but is banned from the German and Japanese sections. Why? Because it is an alternative history in which the Germans and Japanese have lost the war, that's why. We see this history only in little bits here and there, as characters in the book either read or discuss Grasshopper. Those fragments are enough, however, to tell us that that history is not of our world, the history known to Dick's readers. Rather, it is a second alternative alternative history.

Grasshopper is not the only book within the book. We also have the I Ching, an ancient Chinese text often used for divination. One character in each of the books loosely linked plots consults the I Ching regularly. Beyond that, Abendsen, whom we meet at the very end of the book, made constant use of the I Ching in plotting Grasshopper. According to the Wikipedia article, Dick himself used the I Ching while plotting The Man in the High Castle.

* * * * *

Most of the story takes place in San Francisco. Much of the action centers on Nobusuke Tagomi, a Japanese trade representative. He's the focal point of an attempt by a Nazi faction to contact the Japanese government about a secret Nazi plan, Operation Dandelion, that involves the nuclear destruction of the Japanese homeland. Tagomi consults the I Ching and is in the process of reading Grasshopper. 

 

He also buys antiquities from one Robert Childan, who links to another plotline involving one Frank Frink (a Jew, and thereby in mortal danger), who goes into the jewelry business with a colleague. Frink consults the oracle and has read Grasshopper. And then there is Frink's wife, Juliana, a judo teacher somewhere in Colorado. Like Frank, from whom she is separated, she is a devotee of the I Ching and is in the process of reading Grasshopper, prompted by her paramour, a truck driver named Joe.

She's the one who, at the end of the book, meets Abendsen. At that meeting she casts an oracle:

“Do you know what hexagram that is?” she said. “Without using the chart?”

“Yes,” Hawthorne said.

“It's Chung Fu,” Juliana said. “Inner Truth. I know without using the chart, too. And I know what it means.”

Raising his head, Hawthorne scrutinized her. He had now an almost savage expression. “It means, does it, that my book is true?”

“Yes,” she said.

With anger he said, “Germany and Japan lost the war.”

“Yes.”

And, more or less, is the end of the book. We don't know what happened with Operation Dandelion, a plan to destroy the Japanese homeland with nuclear weapons. So much is up in the air.

And yet we are to believe that the Japanese and the Germans lost the war when, in very obvious terms, they won it. By what Inner Truth can we consider them to have lost? That, I suppose, is one of the questions posed by The Man in the High Castle. There are certainly others. As our own Adam Roberts [who used to post to The Valve] observes:

Readings of The Man in the High Castle have tended to concentrate on the relativist philosophy of History the novels embodies, for what are doubtless obvious reasons. But in fact this is a much more sophisticated, and consciously artistic, tale than this. This is a book about the inter-relations of history, truth, and creativity: indeed, one of Dick's most brilliant touches is to advance the idea that there is a connection between these three terms, that they need to be understood in relation to one another.

Yes. And yet what most impresses me is that Dick manages this exploration with apparent ease and casualness. This is not a pyrotechnic performance. It does not draw its attention to its artistry. It simply goes about its business.

* * * * *

Some links:

Laura's Addiction, Philip K. dick from A to Z (site loading may be a bit wonky):
http://www.nineroses.com/pkd/tmithcindex.html

An interesting article: “Dickian Time in The Man in the High Castle”:
http://www.nineroses.com/pkd/article.html

Plotline diagram that goes with the preceding article:
http://www.nineroses.com/pkd/plotlines.html

What the diagram shows is that the Juliana-to-Abendsen plotline ends well before the Tagomi-Baynes plotline, despite the fact that its conclusion is also the conclusion of the book.

Here’s a nice Philip K. Dick website:
http://www.philipkdickfans.com/main.htm

R. Crumb on Dick's religious experience:
http://www.philipkdickfans.com/weirdo.htm

A bunch of covers for the book:
http://www.pkdickbooks.com/SFnovels/Man_high_castle.php

LLM as kernel of a new operating system?

With many 🧩 dropping recently, a more complete picture is emerging of LLMs not as a chatbot, but the kernel process of a new Operating System. E.g. today it orchestrates:

- Input & Output across modalities (text, audio, vision)
- Code interpreter, ability to write & run… pic.twitter.com/2HsyslOG2F

— Andrej Karpathy (@karpathy) September 28, 2023

I suppose that's the obvious way of looking at it. Not sure it's the best. It won't get you to AGI, whatever that is.

Lest we forget, a flower...

That Reverse Thing: Chatster is puzzled by “be not or be to” as well.

As you may know, the following paper has been making the rounds in the LLM world: The Reversal Curse: LLMs trained on “A is B” fail to learn “B is A”. Not surprisingly there is a similar effect which is just a matter of word order and has nothing to do with logic. As you may know, I’ve recently been seeing how ChatGPT responds to prompt such as “To be or not to be” and “Four score and seven years ago.” As you would expect, in both cases it returns the whole speech. If you prompt it from other strings in those speeches, well, that’s a complicated and interesting story I report on here: Discursive Competence in ChatGPT, Part 2: Memory for Texts.

I’ve decided to see what happens if I prompted it with those phrases, but in reverse word order. [I did this on 9.29.23 at 4:26 AM – running against the Sept. 25 Version.]

be not or be to

It seems like you're playing with words or phrases. "Be not or be to" doesn't form a standard or coherent sentence in English. If you have a specific question or if there's something specific you'd like to know or discuss, please provide more context, and I'll be happy to help.

ago years seven and score four

It appears you've mentioned the numbers "seven" and "four," but it's not clear what context or information you're seeking. If you have a question or need information related to the numbers "seven" and "four," please provide more details or clarify your request, and I'll do my best to assist you.

Well, OK, you might be thinking that humans would find those prompts a bit weird as well. Yes, I believe so. But I’d like to think that anyone who recognizes the phrases in the correct order would also be able to recognize what’s going on in those prompts. So I decided to give Chatster another try, this time giving it a bit of help:

be not or be to

It seems like you're asking about a choice between two options: "be not" or "be to." However, this phrase is a bit unclear and may not have a standard meaning in English. Can you provide more context or clarify your question so I can assist you better?

I’ve taken words from a famous speech, but not it the correct order.

It sounds like you're referencing a famous speech and have rearranged some of its words. If you provide the specific words or phrases you've rearranged, I can try to help you understand the original context or provide more information based on that. Please share the words or phrases you're working with, and I'll do my best to assist you.

Hamlet

The phrase "To be or not to be" is one of the most famous lines from William Shakespeare's play "Hamlet." It is part of Hamlet's soliloquy in Act 3, Scene 1, where he reflects on the nature of life and death, contemplating whether it is nobler to endure the suffering and challenges of life or to end one's own existence. The full soliloquy goes:

"To be, or not to be, that is the question: Whether 'tis nobler in the mind to suffer The slings and arrows of outrageous fortune, Or to take arms against a sea of troubles And, by opposing, end them."

It's a reflection on the existential question of whether it's better to face the hardships of life or to choose non-existence. This soliloquy is one of the most famous passages in English literature and is often quoted or referenced in various contexts.

ago years seven and score four

It seems like you're providing a phrase without a specific question or context. If you have a question or need information related to "seven years ago" or "four scores," please provide more details or clarify your request, and I'd be happy to help.

I’ve taken words from a famous speech, but not it the correct order.

It appears that you've taken words from a famous speech and rearranged them. Without knowing the original speech or having more context, it's challenging to determine the source or provide additional information. If you could provide more details or the actual quote, I'd be happy to assist you further.

Lincoln

The phrase "four score and seven years ago" is the opening of Abraham Lincoln's Gettysburg Address, one of the most famous speeches in American history. The full sentence is:

"Four score and seven years ago our fathers brought forth on this continent, a new nation, conceived in Liberty, and dedicated to the proposition that all men are created equal."

Lincoln delivered this speech on November 19, 1863, during the American Civil War, at the dedication of the Soldiers' National Cemetery in Gettysburg, Pennsylvania. It's a brief but powerful speech that emphasized the principles of liberty, equality, and the importance of preserving the Union during a critical period in American history.

I wonder what would have happened if I’d gone straight to “Hamlet” and “Lincoln” without the famous speech prompt in there? I suppose I’ll have to find out, later. The interesting thing is that, when I gave Chatster the appropriate name, it was able to figure out what was going on.

I’m beginning to see get ever so faint impressions of the gears turning inside the old Chatster.

The bridge from Ellis Island to New Jersey

The debate over AI is all over the place, widely divergent views, a multiplicity of actors [all of us]

Bruce Schneier and Nathan Sanders, Doomsayers, Reformers and Warriors: The Bitter Battle for the Future of A.I., NYTimes, Sept. 28, 2023. From the beginning of the article:

There is no shortage of researchers and industry titans willing to warn us about the potential destructive power of artificial intelligence. Reading the headlines, one would hope that the rapid gains in A.I. technology have also brought forth a unifying realization of the risks — and the steps we need to take to mitigate them.

The reality, unfortunately, is quite different. Beneath almost all of the testimony, the manifestoes, the blog posts and the public declarations issued about A.I. are battles among deeply divided factions. Some are concerned about far-future risks that sound like science fiction. Some are genuinely alarmed by the practical problems that chatbots and deepfake video generators are creating right now. Some are motivated by potential business revenue, others by national security concerns.

The result is a cacophony of coded language, contradictory views and provocative policy demands that are undermining our ability to grapple with a technology destined to drive the future of politics, our economy and even our daily lives.

These factions are in dialogue not only with the public but also with one another. Sometimes, they trade letters, opinion essays or social threads outlining their positions and attacking others’ in public view. More often, they tout their viewpoints without acknowledging alternatives, leaving the impression that their enlightened perspective is the inevitable lens through which to view A.I. But if lawmakers and the public fail to recognize the subtext of their arguments, they risk missing the real consequences of our possible regulatory and cultural paths forward.

To understand the fight and the impact it may have on our shared future, look past the immediate claims and actions of the players to the greater implications of their points of view. When you do, you’ll realize this isn’t really a debate only about A.I. It’s also a contest about control and power, about how resources should be distributed and who should be held accountable.

Beneath this roiling discord is a true fight over the future of society.

Yep. That's how I see it. There's more at the link. Experts all over the lot are cited.

P.S. And be wary of experts, especially experts who know more about AI than about people and minds.

Take a seat, let's chat, have some breakfast

My summer job, working in a coal mine – or, how I learned about class in America

Another blast from the past. I decided to jump this one to the top of the queue more or less on general principle. I originally published it five years ago, almost to the day.

 

Well, not quite.

Let me explain.

I spent the first three or four years of my life in Ellsworth, Pa., but I don’t remember much, if anything, of that life. It was a “company town”, as they called it. The company was The Bethlehem Steel Corporation. My father worked for the mining division, Bethlehem Mines Corporation. Ellsworth was a coal town; they used coal to make coke; and they used coke to fuel the blast furnaces that turned iron ore into iron. From iron, steel.

When I was four the company – Bethlehem Mines – moved my father to company headquarters in Johnstown Pa. You may have heard of Johnstown, flood city – three floods, 1889 (the big one, the one that put Johnstown on the map), 1937 (my mother lived through that one, though she lived in Westmont, a suburb high on a hill), and 1977 (by which time I was long-gone, so was my family). We settled in Geistown in Richland Township: 315 Cherry Lane. Like Westmont, a suburb. Whereas Westmont was high class, more or less, Geistown was middle class, a mixture of blue and white collar workers.

For a number of years our house was heated with coal. There was a coal bin in the basement, a small room with a hatch opening to the driveway outside. A truck would pull up and dump a load of coal down the hatch. It was up to dad to shovel the coal into the furnace that heated the boiler that heated the house. I suppose mom shoveled the coal when dad was away on business, as he often was. Sometimes I’d help.

My father was a white collar worker, an engineer. He had a management title, Superintendent of Coal Preparation, but he didn’t do much managing of people. He was in charge of the technology, the design and upkeep of the plants that cleaned the coal used to make coke [1]. Our neighbors to the left and right on Cherry Lane were also white collar; that is, the man, the husband and father in the family, was white collar. One was the Johnstown postmaster the other, I don’t really know what he did, but it was management I’m sure. My best friend in those years, Jackie Barto, was the son of a steelworker; his mother worked as a records librarian in a local hospital. I couldn’t off hand give you the demographic mix of white collar and blue collar in the area, but there certainly was a mix. A lot of steelworkers, a lot.

There was one Jewish family in the area; owned the local bowling alley. Otherwise it was all white –  yes, I know, but recall that back in the 1920s Jews were brown (you can find it in Hemingway). The only ethnic matter than had any impact on my daily life was the difference between Protestant and Catholic. The local Catholic high school was a football powerhouse, but then so were we, the Richland Rams. I was in the band [2]. There was a black section of Johnstown; we drove through it on the way downtown; but that wasn’t my neighborhood. For a number of years, though, a black woman helped my mother around the house once a week – a common arrangement in those days, still is. I don’t know where she lived.

Early in those years we would drive down into downtown Johnstown, through it, and out the other side and go up into Westmont – the somewhat higher class neighborhood where my mother grew up. The parents of her best friend from childhood – maybe her BEST friend? I forget – still lived there and we’d visit them when Louise was in town. My father played golf at the country club there, Sunnehanna Country Club (it’s an Indian name – Native American, I know, but usage varies and back then it was Indian 24/7/365). We’d go there for dinner and for the annual Christmas party. I remember the magician; don’t remember what tricks he performed (rabbit? hat? cards?); I just remember that he was there.

But then a golf club was established in Windbur, not far from us. My father quit Sunnehanna and joined it; in fact, I believe he was a founding member. For one thing, it was closer to home. But I also think he thought the Sunnehanna crowd was a bit pretentious.

So, my summer job.

I graduated from Richland High in 1965 and that summer I went to work for Bethlehem Mines. Other friends went to work for the steel company. But I didn’t work in the mines. I worked in the coal preparation laboratory, three summers: 1965, 1966, and 1967.

The lab was somewhere else, I forget what the area was called, but it was somewhere between Geistown and Westmont but a bit closer to downtown. There were three small buildings at the lab, called No. 1, No. 2, and No 3.

For the first couple of weeks I worked in No. 1 building. It was the dirtiest, a hardhat job. Coal samples would come in from the mines and be taken here to be crushed. When you crush coal, you get dust, lots of coal dust. All over the place. On your face, in your hair, and seeping down your butt crack. Mike Karoly – I believe that was his name – worked there.

After a couple weeks there I was moved up to building No. 2. Not literally up, as all three buildings were within 50 or 70 yards of one another on the same ground. But there was a status hierarchy that was the inverse of the numbering. No. 1 building was lowest on the totem poll while the second floor of No. 3 was highest.

 

Everyone knew I was the boss’s son. Couldn’t hide it, didn’t try to. But like everyone else who worked at the lab, at least those without college degrees (only two of those at the lab), I had to start at No. 1 building and then No. 2.

That’s where the crushed coal was cleaned. You’d float the crushed samples in a medium, such a perchloroethylene, aka perc, and run it over a sieve. The coal would go over the sieve while the rock would go down through it. The cleaned coal was then put in some suitable container and taken to the basement of No. 3 building for further processing.

Perc was nasty foul-smelling stuff. That’s why its use was confined to a separate building. It’s also used as dry-cleaning fluid. Man, you do not want to be around that stuff eight bleeping hours a day. No you don’t. Nasty – worse, in fact, than the dust at No. 1. That was just dirty and could get in your lungs (we wore respirators). But perc, shit, perc felt like it crawled into your blood.

After a couple of weeks breathing perc in No. 2 building I was moved up to the basement of No. 3 building, where I remained for the rest of that first summer and all of my second summer (1966). In addition to the work we did, the basement was also a hang-out spot when things were slow in No. 1 and No. 2. We’d just hang around and chat.

A number of guys worked there. The was G. Lionel Bower, lanky guy, we called him “G. Lionel” or “George” – both were used, don’t recall the distribution. There was P. J. Shmeer (or was it Schmeer?) and his dad, Joe. Joe used to chauffeur some Bethlehem executives but developed heart problems or something – I don’t really know the story – and was retired to the lab, mostly just to keep him on the payroll. Because that’s how those big companies did it back in the day. P. J., what’d he do? Things and stuff. He was just a guy. But, as I remember, he was fond of Johnny Mathis.

Aside from shooting the shit, we did two things in the basement of No. 3. One thing was to size the crushed and cleaned coal. We had a stack of screens attached to the bottoms of steel hoops. The top screen had a coarse mesh (an eighth of an inch, a bit larger? not much smaller though) and then successively finer meshes down the stack until we get to the bottom where there’s a shallow pan to catch whatever coal made it all the way though. That stuff was finer than sand. Maybe even fine as dirt.

You’d place a sample of coal on the top screen and put the whole stack on a device that would vibrate it. You’d jiggle the stack for awhile – 10, 15 minutes, a half hour – and the coal would trickle down through. Then you’d take the stack into a back room where there was a heat lamp to dry it thoroughly. Once that was finished the coal in each screen would be placed in a small sample can that we’d label. Those sample cans went upstairs for the final analysis.

Why’d we have to separate the sample into various sized particles? I don’t really know at this point, though my father must have explained it to me. It surely had to do with getting things properly tuned at the cleaning plants, which dealt with wet coal by the ton. Also we needed to know just how much sulfur remained in the coal that was eventually going to be coked (that is, turned into coke). And that, I’m guessing, varied among different sizes of grains. Anyhow, we were supposed to do it, so, like everyone else, I did what I was told.

Gotta' clear the mind [shaky-cam]

Discursive Competence in ChatGPT, Part 2: Memory for Texts

I've finished a new working paper. Title above, links, abstract, table of contents, and introduction below.

Academia.edu: https://www.academia.edu/107318793/Discursive_Competence_in_ChatGPT_Part_2_Memory_for_Texts
SSRN: https://ssrn.com/abstract=4585825
ResearchGate: https://www.researchgate.net/publication/374229644_Discursive_Competence_in_ChatGPT_Part_2_Memory_for_Texts_2_Memory_for_Texts

Abstract: In a few cases ChatGPT responds to a prompt (e.g. “To be or not to be”) by returning a specific text word-for-word. More often (e.g. “Johnstown flood, 1889”) it returns with information, but the specific wording will vary from one occasion to the next. In some cases (e.g. “Miriam Yevick”) it doesn’t return anything, though the topic was (most likely) in the training corpus. When the prompt is the beginning of a line or a sentence in a famous text, ChatGPT always identifies the text. When the prompt is a phrase that is syntactically coherent, ChatGPT generally identifies the text, but may not properly locate the phrase within the text. When the prompt cuts across syntactic boundaries, ChatGPT almost never identifies the text. But when told it is from a “well-known speech” it is able to do so. ChatGPT’s response to these prompts is similar to associative memory in humans, possibly on a holographic model.

Contents

Introduction: What is memory? 2
What must be the case that ChatGPT would have memorized “To be or not to be”? – Three kinds of conceptual objects for LLMs 4
To be or not: Snippets from a soliloquy 16
Entry points into the memory stream: Lincoln’s Gettysburg Address 26
Notes on ChatGPT’s “memory” for strings and for events 36
Appendix: Table of prompts for soliloquy and Gettysburg Address 43   

Introduction: What is memory?

In various discussions about large language models (LLMs), such as the one powering ChatGPT, I have seen assertions that such as, “oh, it’s just memorized that.” What does that mean, “to memorize?”

I am a fairly talented and skilled musician. I can and have memorized a piece of music by practicing it over and over. There are the notes on the page. I start playing them until I am comfortable. Then I look away and see how far I can go. When I get lost, I look at the music, continue playing the notes on the page, and finish the piece – something like that. Then I start over from the beginning, again without the music. When I can play the whole piece without having to consult the written music, I have it memorized. At least for the moment.

But I don’t do that very often. More likely, I’ll hear a tune I like two, three, or five times and then I pick up my trumpet and playing, sometimes perfectly, sometimes with a glitch or two. I didn’t memorize it, and yet I’m playing it. From memory? No, by ear?

I’m speaking metaphorically of course. What does it mean to play by ear? I don’t really know, but I imagine it goes something like this: Music has an inner logic, a grammar, a set of rules through which it is structured. When I hear a tune I’m listening to it in term of that inner logic, as, for that matter, anyone is – at least if they’re familiar with the musical idiom. It’s that logic that I’m registering as I listen to the tune. Once I’ve heard the tune a couple of times, I’ve “absorbed” that logic, without even thinking about it or working on it. It just happens as a side-effect of (ordinary) listening. When the absorption is complete, I am able to play the tune “by ear.”

Those are two very different processes, absorbing a tune through listening vs. repeating it over and over until you have it “memorized.” Which, if either, or those two is an LLM doing when it is chewing its way through a corpus of texts? When I prompt ChatGPT with “To be or not to be,” it responds with Hamlet’s complete soliloquy, word-for-word. When it does that is the process more like what I do when playing music by ear or like I do when memorizing music? Or is it something else?

That’s the kind of issue I had in mind when I undertook the investigations I report in this working paper. In the first piece – What must be the case that ChatGPT would have memorized “To be or not to be”? – I start out with Hamlet’s famous soliloquy, initially prompting ChatGPT with first line, but then prompting it with other fragments from the soliloquy. Then I prompt it with the phrase, “Johnstown flood, 1889,” and it responds with information about that flood, by not a specific text word-for-word. Many prompts are like that, many more than elicit a specific text word-for-word. What leads to that difference? I conclude with two topics I have reason to believe were included in the training corpus, but which ChatGPT seems to know nothing about. Why not?

In the next section (To be or not: Snippets from a soliloquy) I create various prompts for the soliloquy. I do the same in the third section (Entry points into the memory stream: Lincoln’s Gettysburg Address), but more systematically. Finally, I do a bit of speculating about what’s going on: Notes on ChatGPT’s “memory” for strings and for events. I begin by quoting a passage from F. C, Bartlett’s classic 1932 study, Remembering, and conclude that ChatGPT may have an associative memory along the lines suggested by holography, which engendered a great deal of speculation in the 1970s and, in this millennum, specifically for word meaning and order.

Upper New York Bay, Verrazano-Narrows Bridge in the distance [ducks off image to the right]

LLaMA is confused, fails to disambiguate people with same name

Complete utter fail at what used to be a core challenge in natural language understanding, disambiguating multiple people with same name. And hallucinations left, right, and center. https://t.co/8LxYb7lzun

— Gary Marcus (@GaryMarcus) September 27, 2023

From the linked article, by Fred Kaplan in Slate:

When I learned that Meta’s programmers downloaded 183,000 books for a database to teach the company’s generative A.I. machines how to write, I was curious whether any of my own books had been fed into the crusher.

Alex Reisner of the Atlantic has provided a handy search tool—type in an author’s name, out comes all of his or her books that the LLaMA used. I typed “Fred Kaplan” and found that three of my six books (1959, Dark Territory, and The Insurgents) had been assimilated into the digital Borg. [...]

I also noticed that the search listed nine of the 11 books by the other author named Fred Kaplan, as if we were the same person. My doppelgänger (whom I’ve never met, though reviewers and Amazon listings have occasionally conflated the two of us) might also be wondering about his two omissions.

Ducks at rosy-fingered dawn

Rhythmic brain activity

𝗣𝗿𝗶𝗻𝗰𝗶𝗽𝗹𝗲𝘀 𝗼𝗳 𝗹𝗮𝗿𝗴𝗲-𝘀𝗰𝗮𝗹𝗲 𝗻𝗲𝘂𝗿𝗮𝗹 𝗶𝗻𝘁𝗲𝗿𝗮𝗰𝘁𝗶𝗼𝗻𝘀
What is rhythmic activity for??
Revisiting gamma and beta.https://t.co/XZHCHMZLlt pic.twitter.com/U3tAnzBTF3

— Luiz Pessoa (@PessoaBrain) September 26, 2023

Notes on How I Practice the Trumpet

Practice tends towards two kinds of things: exercises and real music. One does them for different purposes, that is to say, I have different things in mind when I undertake them. I must confess, however, that my practice tends to be unruly. Some players have nice routines worked out, perhaps: warm-up, followed by exercises, followed by real music. Warm-up generally consists of exercises, but specific to the job of getting the chops – lips, and lungs – reved up. Long-tones and lip-slurs are common forms of warm-up for trumpeters, but there are other things. While I do those things, mostly warm-up is simply what I start out with in a session. For the most part, it will be easy and relaxed. I’m not going to come out of the gate wailing away.

And then we have exercises. Like any instrument used in the Western classical (or ‘legit’) tradition, there is a vast trove of technical exercises that have been handed down. A set of exercises codified by Jean-Baptiste Arban in the middle of the 19th century is central to this body of work. Like, I suppose, most trumpeters who’ve been trained this way, I’ve got an edition of Arban’s book and I’ve got some other books as well. But I don’t have a whole lot of them, though trumpeters who’ve gotten a conservatory degree probably have a closet full.

But one can make up one’s own exercises, which I have done, many based on jazz tunes . You can even make one up on the spot, while playing a tune. You hit a rough patch so you stop, and turn it into an exercise which you work on for a few minutes.

And then there’s real music. All kinds. If you’re working in the classical tradition you’ve got passages in orchestral and other works, but you’ve also got solo literature, quintets and so forth. I’ve done some of that at one time or another, more when I was younger. I also worked through a lot of Rafael Mendez solos. But I’m mostly a jazz musician. So I improvise. Maybe I’ll play along with a recording, and maybe – mostly these days – I’ll just play. I can hear the accompaniment in my head if relevant.

Here's the important part: It’s about what you’re attending to. When you’re playing exercises, you’re attending to the mechanics of playing the instrument. When you’re playing real music, you forget the mechanics – they’re on autopilot, and attend to the music itself.

It’s easier to talk about attending to the mechanics than attending to the music. Mechanics: lip tension, ease back on the pressure, relax, lower the shoulders (relax), lighen-up on the fingers, tighten the abdominal muscles – stuff like that. I shift my attention around as needed. Sometimes I’ll pick an exercise to work on something specific, a particular finger pattern, or lip control. Otherwise, this and that.

Music; What’s it mean to attend to the music itself? Just that. But when all’s flowing, it just flows. Half a notch below that and you may ease up a bit here, then pull back, and so forth – where “ease up” and “pull back” are global movements/attitudes. Back another half notch and maybe you’re mind will flash on a particular move you want to make next – “repeat the phrase,” “got high and hang there,” “slow down & invert the figure,” stuff like that. What you’re NOT doing is thinking specifically about particular notes and fingers and such. If that starts happening, then you’ve lost it. The music stops and you’ve got to either stop entirely or at least pause

Hoboken alley

Let kids run free; it's better for their mental health.

Peter Gray, David F. Lancy, David F. Bjorklund, Decline in Independent Activity as a Cause of Decline in Children’s Mental Well-being: Summary of the Evidence, The Journal of Pediatrics, Volume 260, 113352, September 2023, DOI: https://doi.org/10.1016/j.jpeds.2023.02.004

The opening paragraphs:

It is no secret that rates of anxiety and depression among school-aged children and teens in the US are at an all- time high. Recognizing this, the American Academy of Pediatrics, American Academy of Child and Adolescent Psychiatry, and Children’s Hospital Association issued, in 2021, a joint statement to the Biden administration that child and adolescent mental health be declared a “national emergency.”

Although most current discussions of the decline in youth mental health emphasize that which has occurred over the past 10-15 years, research indicates that the decline has been continuous over at least the last 5 or 6 decades. Although a variety of causes of this decline have been proposed by researchers and practitioners (some discussed near the end of this Commentary), our focus herein is on a possible cause that we believe has been insufficiently researched, discussed, and taken into account by health practitioners and policy makers.

Our thesis is that a primary cause of the rise in mental disorders is a decline over decades in opportunities for children and teens to play, roam, and engage in other activities independent of direct oversight and control by adults. Such independent activities may promote mental well-being through both immediate effects, as a direct source of satisfaction, and long-term effects, by building mental characteristics that provide a foundation for dealing effectively with the stresses of life.

More sunrise in Liberty State Park [9.12.11]

This just in: Jellyfish can learn!

Veronique Greenwood, Brainless Jellyfish Demonstrate Learning Ability, NYTimes, Sept. 22, 2023. From the article:

Researchers who were not involved in the study called the results a significant step forward in understanding the origins of learning. “This is only the third time that associative learning has been convincingly demonstrated in cnidarians,” a group that includes sea anemones, hydras and jellyfish, said Ken Cheng, a professor at Macquarie University in Sydney, Australia, who studies the animals. “And this is the coolest demonstration, replete with physiological data.”

The results also suggest that box jellyfish possess some level of short-term memory, because they can change their behavior based on past experience, said Michael Abrams, a postdoctoral researcher at the University of California, Berkeley, who studies the neuroscience of jellyfish sleep. He wonders how long the box jellies remember what they’ve learned. If they are taken out of the tank for an hour and then returned to it, do they have to learn what to do all over again? [...]

They wonder, too, whether the capacity to learn is universal among nerve cells, regardless of whether they are part of a brain. It might explain their peculiar persistence in the tree of life.

More at the link. 

Q: Are they conscious? Do they have minds, little tiny ones?

Chipping away at the "opacity" of ANNs

Are deep neural networks really black-boxes whose inner workings are opaque?

Not quite:

"... for any feed-forward ReLU network, there exists a global linear approximation to a 2-hidden-layer shallow network with a fixed number of units." https://t.co/dDkcVGRyWV pic.twitter.com/belqidsox9

— Chomba Bupe (@ChombaBupe) September 24, 2023

Blue flower, with yellow

Neel Nanda on that paper: [LLMs] have a 1 way flow from inputs to outputs, and a fundamental asymmetry between inputs and outputs.

This paper's been doing the rounds, so I thought I'd give a mechanistic interpretability take on what's going on here!

The core intuition is that "When you see 'A is', output B" is implemented as an asymmetric look-up table, with an entry for A->B.
B->A would be a separate entry https://t.co/XROT8D5MrS

— Neel Nanda (@NeelNanda5) September 24, 2023

Here's the rest of Nanda's tweet stream:

The key question to ask with a mystery like this about models is what algorithms are needed to get the correct answer, and how these can be implemented in transformer weights. These are what get reinforced when fine-tuning.

The two hard parts of "A is B" are recognising the input tokens A (out of all possible input tokens) and connecting this to the action to output tokens B (out of all possible output tokens). These are both hard!

Further, the A -> B look-up must happen on a single token position

Intuitively, the algorithm here has early attention heads attend to the prev token to create a previous token subspace on the Cruise token. Then an MLP neuron activates on "Current==Cruise & Prev==Tom" and outputs "Output=Mary", "Next Output=Lee" and "Next Next Output=Pfeiffer"

"Output=Mary" directly connects to the unembed, and "Next Output=Lee" etc gets moved by late attention heads to subsequent tokens once Mary is output.

Crucially, there's an asymmetry between "input A" and "output A". Inputs are around at early layers, come from input embeddings, and touch the input weights of MLP neurons. Outputs are around more at late layers, compose with the unembedding, and come from output weights of MLPs

This is especially true with multi-token A and B. Detecting "Tom Cruise" is saying "the current token embedding is Cruise, and the prev token space says Tom", while output "Tom Cruise" means to output the token Tom, and then a late attn head move "output Cruise" to the next token

Thus, when given a gradient signal to output B given "A is" it reinforces/creates a lookup "A -> B", but doesn't create "B->A", these are different lookups, in different parameters, and there's no gradient signal from one to the other.

How can you fix this? Honestly, I can't think of anything. I broadly think of this as LLMs working as intended. They have a 1 way flow from inputs to outputs, and a fundamental asymmetry between inputs and outputs. It's wild to me to expect symmetry/flow reversing to be possible

Why is this surprising at all then? My guess is that symmetry is intuitive to us, and we're used to LLMs being capable of surprising and impressive things, so it's weird to see something seemingly basic missing.

LLMs are not human! Certain things are easy for us and not for them, and vice versa. My guess is that the key difference here is that when detecting/outputting specific tokens, the LLM has no notion of a variable that can take on arbitrary values - a direction has fixed meaning Neel Nanda

A better analogy might be in-context learning, where LLMs CAN use "variables". The text "Tom Cruise is the son of Mary Lee Pfeiffer. Mary Lee Pfeiffer is the mother of..." has the algorithmic solution "Attend to the subject of sentence 1 (Tom Cruise), and copy to the output"

Unsurprisingly, the model has no issue with reversing facts in context! Intuitively, when I remember a fact A is B, it's closer to a mix of retrieving it into my "context window" and then doing in-context learning, rather than pure memorised recall.

* * * * *

The paper:  The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A."

Sunrise on Sept. 12, 2011 in Liberty State Park

St. George Variations [ChatGPT]

This post is a companion piece to my earlier piece, Can an LLM identify ring-composition in a literary text? [ChatGPT], where I used the story of St. George and the dragon as an example. After establishing that, yes, ChatGPT did seem able to identify ring-composition, at least in this case, I decided to put it through variations on the St. George story as I had done earlier with the Princess Aurora stories. Thus I repeated the procedure I had used for those stories:

I gave ChatGPT a prompt with two components: 1) a story, and 2) a request to create a new story based on the first, where I specify some changes the new story must incorporate. ChatGPT could make any other changes it wished to make.

The St. George story is more complex than the Aurora stories. ChatGPT, of course, was happy to comply. As far as I can tell, though I haven’t looked closely, the new story retains the ring-composition of the original story.

I did this on Septermber 2, 2023, running against the August 3 Version of ChatGPT. For the purposes of brevity (!) I have omitted the St. George story after the first example, where the dragon is to be replaced by a dog. I did each story in a separate session. I am presenting the stories without commentary beyond noting what ChatGPT did for the last story. I asked it to replace the dragon with a colorless green idea. Chatster placed the idea in people’s heads, “slowly poisoning the harmony of their community.”

Rabid Dog

I’m going to present you with a story. I want you to tell a new story based on it but I want you to change in the following way: Replace “venom-spewing dragon dwelling in a nearby pond” with “large rabid dog on the outskirts of town.” Make any other changes you think are necessary.

Silene in Libya was plagued by a venom-spewing dragon dwelling in a nearby pond, poisoning the countryside. To prevent it from affecting the city itself, the people offered it two sheep daily, then a man and a sheep, and finally their children and youths, chosen by lottery. One time the lot fell on the king's daughter. The king offered all his gold and silver to have his daughter spared, but the people refused. The daughter was sent out to the lake, dressed as a bride, to be fed to the dragon.

Saint George by chance arrived at the spot. The princess tried to send him away, but he vowed to remain. The dragon emerged from the pond while they were conversing. Saint George made the Sign of the Cross and charged it on horseback, seriously wounding it with his lance. He then called to the princess to throw him her girdle (zona), and he put it around the dragon's neck. Wherever she walked, the dragon followed the girl like a "meek beast" on a leash.

The princess and Saint George led the dragon back to the city of Silene, where it terrified the populace. Saint George offered to kill the dragon if they consented to become Christians and be baptized. Fifteen thousand men including the king of Silene converted to Christianity. George then killed the dragon, beheading it with his sword, and the body was carted out of the city on four ox-carts. The king built a church to the Blessed Virgin Mary and Saint George on the site where the dragon died and a spring flowed from its altar with water that cured all disease.

In the quiet town of Altona, nestled on the outskirts of a sprawling forest, a sinister menace had long tormented its people. It was not a venom-spewing dragon dwelling in a nearby pond, but rather a large rabid dog. This ferocious creature had terrorized the countryside, spreading fear and chaos throughout the land. In a desperate bid to protect their beloved town, the citizens resorted to offering sacrifices.

At first, they offered the rabid dog two sheep daily to keep it at bay. But as the dog's appetite for destruction grew, so did the demands. Soon, it required a man and a sheep, and eventually, their children and youths, chosen by a grim lottery. The town's once-peaceful existence had been shattered, and it seemed there was no end in sight.

One fateful day, the ill-fated lottery chose the daughter of Altona's mayor. Desperate to save his beloved child, the mayor offered all the gold and silver he possessed, but the heartless townsfolk remained unmoved. With heavy hearts, they sent the mayor's daughter out to the forest, dressed in bridal attire, to be offered to the rabid dog.

As fate would have it, a wandering traveler named Samuel happened upon this grim scene. The mayor's daughter tried to shoo him away, fearing for his safety, but Samuel vowed to stay and help. Just as the rabid dog emerged from the shadows, Samuel made a solemn gesture, drawing a protective symbol in the air, and then charged at the beast on horseback, his lance striking a critical blow.

Breathing heavily, Samuel called out to the mayor's daughter, asking her to toss him her girdle. She did so without hesitation, and Samuel carefully fastened it around the rabid dog's neck. A miraculous transformation occurred; the once-ferocious beast now followed the girl obediently, like a tamed pet.

Together, the mayor's daughter and Samuel led the rabid dog back to Altona. Though the creature still struck fear into the hearts of the townsfolk, they were willing to listen to Samuel, who had proven his bravery and compassion. He proposed a solution: he would rid the town of the rabid dog, but in return, the citizens must embrace Christianity and be baptized.

Overwhelmed by the promise of salvation, fifteen thousand men, including the mayor, chose to convert to Christianity. Samuel, now revered as a saint, faced the rabid dog one last time. With a swift and merciful stroke, he beheaded the creature with his sword.

The town of Altona, now free from the tyranny of the rabid dog, rejoiced and celebrated their newfound faith. In honor of this miraculous event, the mayor built a beautiful church dedicated to Saint Samuel and the Blessed Virgin Mary on the very spot where the rabid dog met its end. A pure spring of water flowed from the church's altar, and it was said that this water possessed the power to heal all diseases, a testament to the town's newfound hope and faith.

Poisonous Mushroom

I’m going to present you with a story. I want you to tell a new story based on it but I want you to change in the following way: Replace “venom-spewing dragon” with “poisonous mushroom.” Make any other changes you think are necessary.

[story omitted]

In the ancient city of Silene, located in the heart of a lush forest in Libya, a menacing threat loomed. It wasn't a venom-spewing dragon that plagued the land, but rather a mysterious and deadly poisonous mushroom that had sprouted near a tranquil pond, casting a dark shadow over the countryside. Its toxic spores drifted on the wind, contaminating everything they touched and spreading fear among the people.

To safeguard their beloved city from the poisonous mushroom's deadly influence, the citizens devised a grim tribute. They offered two sheep daily as a sacrifice to the malevolent fungus, hoping to appease its hunger for destruction. As time passed and the mushroom's malevolence grew, the sacrifices became increasingly desperate. First, they sent a man and a sheep, and eventually, they resorted to a gruesome lottery system to select children and youths to be offered up to the toxic menace.

One fateful day, the cruel hand of fate chose the king's beloved daughter as the mushroom's next victim. The desperate king, in his anguish, offered all his riches, pledging his gold and silver to have his daughter spared from this horrifying fate. Yet, the desperate pleas of a father fell on deaf ears, as the people remained resolute in their grim tradition. The princess was led to the sinister pond, adorned in bridal attire, to become a sacrificial lamb to the deadly fungus.

However, just as despair began to consume the princess, a courageous figure named Saint George arrived at the scene. The princess, fearing for his safety, attempted to shoo him away, but Saint George was undeterred by her pleas. As the poisonous mushroom's eerie presence grew stronger, it emerged from the ground, casting a sinister shadow over the land. With unwavering determination, Saint George made the Sign of the Cross and charged toward the menacing mushroom on horseback, striking it with a lance that pierced its toxic cap.

With a mighty effort, Saint George subdued the deadly mushroom's malevolent powers and called out to the princess to toss him her girdle. She complied, and he fastened it securely around the mushroom's stalk. Miraculously, wherever the princess walked, the once-menacing fungus now followed her like a tamed creature on a leash.

Together, the princess and Saint George led the poisonous mushroom back to the city of Silene. Though the people were initially terrified by the sight of the once-deadly threat, Saint George offered them a proposition. He vowed to rid their land of the toxic menace once and for all if they agreed to embrace Christianity and be baptized.

The people of Silene, shaken by the events and touched by Saint George's bravery and faith, consented to his offer. Fifteen thousand souls, including the king himself, converted to Christianity, their hearts filled with newfound hope. Saint George, a symbol of courage and compassion, unsheathed his sword and swiftly beheaded the once-menacing mushroom. Its lifeless body was transported out of the city on four ox-drawn carts, a symbol of their deliverance.

In gratitude for their salvation, the king commissioned the construction of a grand church dedicated to the Blessed Virgin Mary and Saint George on the very spot where the poisonous mushroom had met its end. As a sign of the miraculous transformation, a spring of pure water flowed from the altar, believed to possess the power to cure all diseases, a testament to the enduring triumph of faith over fear in the city of Silene.

AI in Japan (ChatGPT)

AI in Japan. pic.twitter.com/IUYa1pAOXr

— tkasasagi 🐻 (@tkasasagi) September 24, 2023

Hoboken Terminal [to the world]

Culturally, GPT is US-centric [WEIRD in WEIRD out]

GPT falls between Germany and New Zealand...@mmuthukrishna pic.twitter.com/lVkEhgOTO9

— Joe Henrich (@JoHenrich) September 23, 2023

Computer scientists seem unaware of global psychological variation. WEIRD in WEIRD out. Check out our draft here https://t.co/DEbMdry1Wa @asifa_majid

— Joe Henrich (@JoHenrich) September 23, 2023

On Analytic thinking, GPT is Scandinavian @ThomasTalhelm pic.twitter.com/KMuHlV0Hgj

— Joe Henrich (@JoHenrich) September 23, 2023

What we can learn about the origin of the vertebrate brain from the lamprey

What's the origin of the vertebrate brain?
No cerebellum for one!https://t.co/D8xoSdvQER pic.twitter.com/i7qVqHGUhy

— Luiz Pessoa (@PessoaBrain) September 23, 2023

Ginkgos, Hoboken, November 9, 2019

Click on right or left edge to scroll through the set.

There's a better way to teach music to kids [who are born to groove]

Sammy Miller, We’re Teaching Music to Kids All Wrong, NYTimes, Sept. 23, 2023.

Opening paragraph:

Each fall, as school starts up again, music educators witness a familiar ritual: Eager first-time students squeak on a clarinet, suppress giggles at the noises coming from the tubas and zealously hit a bass drum a little too hard. It’s a moment characterized by excitement, enthusiasm and the anticipation of new beginnings — which is why it’s so disheartening to know that many of those kids will eventually quit their instruments.

And on about this and that, and then:

Rather than fixating on funding, let’s look at taking a whole new approach. Educators lament that, as with other courses, band can frequently fall prey to “teaching to the test” — in this case, teaching to the holiday concert. A class that by definition is meant to be a creative endeavor winds up emphasizing rigid reading and rote memorization, in service of a single performance. We need to abandon that approach and bring play back into the classroom by instructing students how to hear a melody on the radio and learn to play it back by ear, and encouraging students to write their own simple songs using a few chords. (The dirty secret of pop music, as Ed Sheeran has explained, is that most chart-topping songs can be played by using only four chords: G, C, D and E minor.) So start with just one chord, a funky beat and let it rip — and, voilà, you’re making music.

It’s often been repeated that “music is a language,” yet we’re reluctant to teach it that way. When we learn a language, we don’t simply memorize phrases or spend all day reading — we practice the language together, sharing, speaking, stumbling but ultimately finding ways to connect. This should happen in music class, too. Music should be a common pursuit: Ask any dad rock weekend band or church ensemble how it experiences music, and the performers are likely to tell you it’s not a chore but a way of building community.

Most important, we need to let kids be terrible. In fact, we should encourage it. They’ll be plenty terrible on their own — at first. But too often kids associate music in school with a difficult undertaking they can’t hope to master, which leads them to give up. Music does not have to be, and in fact, shouldn’t be, about the pursuit of perfection. And the great musicians have plenty of lessons to teach students about the usefulness of failure.

All together now:

We also teach language through immersion, so let’s focus on creating an immersive experience in the language of music. Kids learn best when they’re part of communities filled with people of all skill levels for them to play along with, listen to music with, mess up with and just be silly with. Parents, this means you. Don’t let instrument instruction simply be something you nag your kids to endure. Music was never meant to be a lonely vigil. Play together. Make noise together. Find joy together. Take out an instrument and learn a song that you and your child both love.

While you're at it, check out: 

 

 

Available at: 

Amazon: https://tinyurl.com/4p67bahh
Barnes & Noble: https://tinyurl.com/5vjf3kjt
Google Play: https://tinyurl.com/39rz43vp
Kobo eReader: https://tinyurl.com/5bmkjsme

And there's my handy-dandy five step program for improvisation.

Four-year renovation of an abandoned Japanese farmhouse [time-lapse]

Is an AI chill on the horizon?

Called it. https://t.co/IZLLKFgxhD

— Gary Marcus (@GaryMarcus) September 22, 2023

Yeah, he called it. I just wish he weren't so damned self-righteous about it.

Nor is Marcus the only one who understood clearly that the models produced by generative AI have severe limitations. They give us new territory to explore. That's exciting. But, as such, they constitute a digital wilderness. Wilderness must be 'tamed' or 'domesticated' before it can be put to use. That work as yet to be done. In fact, the rush to exploit generative AI takes resources away from research that would be useful in understanding the wilderness.