Monday, March 18, 2024

GPT, the magical collaboration zone, Lex Fridman and Sam Altman

I was making one more run around the web before I buckled down and got back to a major writing task, when I came across the brand-spanking-new conversation between Lex Fridman and Sam Altman. Lex is Lex, and an interesting guy, and Sam is, well, he's interesting to me, but – there was a hint of megalomania at the end of that NYTimes story from Mar. 31, 2023, that rubbed me the wrong way, and all the AI hype – he IS the CEO of OpenAI. So it seemed to me that I just had to listen in, not the whole thing – and I could legit play solitaire while listening – and so I did, skipping over stuff.

But then the conversation hit an interesting patch. So – and I'm not going to try to re-create the context – they're talking about GPT-4 at roughly 46:03:

Altman: what are the best things it can do

Fridman: what are the best things it can do and the the limits of those best things that allow you to say it sucks therefore gives you an inspiration and hope for the future

Altman: you know one thing I've been using it for more recently is sort of a like a brainstorming partner
and for that there's a glimmer of something amazing in there
I don't think it gets you know
when people talk about it
it what it does they're like
ah it helps me code more productively
it helps me write more faster and better
it helps me you know translate from this language to another
all these like amazing things
but there's something about the like kind of creative brainstorming partner
I need to come up with a name for this thing
I need to like think about this problem in a different way
I'm not sure what to do here
uh that I think like gives a glimpse of something I hope to see more of

um one of the other things that you can see like a very small glimpse of is
when it can help on longer Horizon tasks
you know break down some multiple steps
maybe like execute some of those steps
search the internet
write code whatever put that together uh
when that works which is not very often
it's like very magical

At about 52:54:

Fridman: I use it as a reading partner for reading books
it helps me think
help me think through ideas especially when the books are classic
so it's really well written about and it actually is is I
I find it often to be significantly better than even like Wikipedia on well-covered topics
it's somehow more balanced and more nuanced or maybe it's me
but it inspires me to think deeper than a Wikipedia article does
I'm not exactly sure what that is
you mentioned like this collaboration I'm not sure where the magic is if it's in here [gestures to his head]
or if it's in there [points toward the table]
or if it's somewhere in between

It's that magic-collaborative zone that interests me. While I've spent a great deal of time working with (plain old) ChatGPT, most of that time I've been doing research on how it behaves. But every once in awhile I'll play around just to mess around. And then I've seen sparks of magic. The interaction that generated AGI and Beyond: A Whale of a Tale certainly had the magic flowing, and it showed up here and there during the Green Giant Chronicles. I suspect those two cases are somewhat idiosyncratic. Nor am I sure that I can do this at will. But there's definitely something there, and its in the interaction.

I would guess that the magic varies from person to person as well. I wonder how many uses have had these kind of magical flow interactive states? I'd thinking finding that out would be tricky because they're likely to be idiosyncratic and elusive. If I were to research it, I'd probably start out with interviews, either face-to-face or through some online medium. That might lead to a questionnaire that could be used more broadly.

It'll be interesting to see how Altman ends up characterizing this flow state – which is what I'm calling it for the moment, a man-machine flow state. It's the human, of course, that's in flow. The machine is just being the machine. 

* * * * * 

I have a final comment, of an epistemological nature. As the post indicates, I'd already had a magical interaction or two with ChatGPT before I listened to this podcast. The first time it came up in the podcast, from Altman, OK, I noted it. And went on, playing solitaire with one part of my mind and listening in on the podcast with another part. But then it came up again, this time from Fridman. Wham! That's three, my threshold number for this kind of thing. Three people independently have the same or similar experience. Maybe there's something real there.

Clementines and acorn squash

Two thoughts about being human during the coming AI apocalypse

One problem doesn’t compute, the other does. Why?

Human specialness

Scott Aaronson worries: The Problem of Human Specialness in the Age of AI. I don’t get it. It doesn’t compute. Are we special now? Aaronson seems to think so; we’re special because we’re the smartest. I suppose that’s true. But you know, that doesn’t compute either. Not for me.

Back in 1990 David Hays and I published “The Evolution of Cognition,” where we observed:

A game of chess between a computer program and a human master is just as profoundly silly as a race between a horse-drawn stagecoach and a train. But the silliness is hard to see at the time. At the time it seems necessary to establish a purpose for humankind by asserting that we have capacities that it does not. It is truly difficult to give up the notion that one has to add “because . . .” to the assertion “I’m important.” But the evolution of technology will eventually invalidate any claim that follows “because.” Sooner or later we will create a technology capable of doing what, heretofore, only we could.

So that’s one thing. There’s another.

What to do with our time in a world with UBI

There’s lots of talk about universal basic income (UBI) these days, a lot of it coming from Silicon Valley, where they think that machines are going to take over most of the jobs, putting people out of work. What will those people do with their time?

We’ve got a society that’s been built on the idea that one’s job is one’s primary identity and source of a sense of self-worth. In that dispensation raising a family qualifies as a woman’s job. Keeping in mind that many jobs are what David Graber has called “bullshit jobs,” and so not an adequate anchor for self-worth, what happens in a world where many people are not required to work. What will people do with their time?

In that world, how do we connect with the larger world? For that matter, given the widespread existence of those bullshit jobs, how do we do it now?

See this post from 2021: Why are we as a culture addicted to work? [Because we have forgotten how to play.]

* * * * *

Let me repeat: The problem of human specialness does not compute. The problem of filling time in the absence of (nonbullshit) work does. Why?

An alternative to generative grammar formalisms

Geoffrey Keith Pullum, Theorizing about the Syntax of Human Language: A Radical Alternative to Generative Formalisms, Cadernos de Linguística 1(1):01-33, July 2020, DOI: 10.25189/2675-4916.2020.v1.n1.id279

Abstract: Linguists standardly assume that a grammar is a formal system that ‘generates’ a set of derivations. But this is not the only way to formalize grammars. I sketch a different basis for syntactic theory: model-theoretic syntax (MTS). It defines grammars as finite sets of statements that are true (or false) in certain kinds of structure (finite labeled graphs such as trees). Such statements provide a direct description of syntactic structure. Generative grammars do not do this; they are strikingly ill-suited to accounting for certain familiar properties of human languages, like the fact that ungrammaticality is a matter of degree. Many aspects of linguistic phenomena look radically different when viewed in MTS terms. I pay special attention to the fact that sentences containing invented nonsense words (items not in the lexicon) are nonetheless perceived as sentences. I also argue that the MTS view dissolves the overblown controversy about whether the set of sentences in a human language is always infinite: many languages (both Brazilian indigenous languages and others) appear not to employ arbitrarily iterative devices for embedding or coordination, but under an MTS description this does not define them as radically distinct in typological terms.

The AI Marketplace is cooling down

Anissa Gardizy and Aaron Holms, Amazon, Google Quietly Tamp Down Generative AI Expectations, The Information, March 12, 2024.The article begins:

In the past year, major technology firms have championed generative artificial intelligence as the next big thing, boosting the stock market to new highs. But behind the scenes, representatives of major cloud providers and other firms that sell the technology are tempering expectations with their salespeople, saying the hype about the technology has gotten ahead of what it can actually do for customers at a reasonable price.

Several executives, product managers and salespeople at the major cloud providers, such as Microsoft, Amazon Web Services and Google, also privately said most of their customers are being cautious or “deliberate” about increasing spending on new AI services, given the high price of running the software, its shortcomings in terms of accuracy and the difficulty of determining how much value they'll get out of it.

Sunday, March 17, 2024

Michael Moschen performs THE TRIANGLE

From 2007:

Master juggler Michael Moschen performs his incredibly famous, jaw-dropping piece where three balls and a triangle become a musical and visual work of art. From PBS Great Performances produced by Skip Blumberg. To book Michael Moschen visit WWW.BELENZON.COM or call 858-832-8380. Join the Michael Moschen mailing list at www.michaelmoschen.com/fans.html.

I've been quite critical of the MacArthur Foundation's Fellowship program, the so-called genius awards, and have argued at some length that it mostly awards the same-old same-old. Every once in awhile, however, the award someone of genuine talent and originality. Moschen is one of those. He's in the class of 1990.

Ride a bike

What we can learn from toddlers about well-being

Jancee Dunn, Who Has the Secret to Well-Being? The Answer May Surprise You. NYTimes, Mar. 17, 2024. From the article:

Try positive self-talk.

Young children tend to coach themselves out loud, a practice known as private speech (such as this popular clip from a 4-year-old snowboarder).

Toddlers aren’t shy about self-talk, Dr. Merali said, and you shouldn’t be, either. Research suggests that for adults, positive self-talk can help with problem-solving, learning, confidence and managing your emotions.

Take any opportunity to move.

Two-year-olds are active for almost five hours a day, according to a review of 24 studies. They move joyfully and instinctively, Dr. Merali said.

Adults can look for ways to move more, even if it’s just for a minute.

That's one thing that's a drag about my current living situation. There's not much room for me to pace the floor, which I love to do.

Ask questions.

Young kids are not afraid to pose questions, Dr. Merali said. One study found that they asked an average of 107 questions an hour. (This will not surprise their parents.) [...]

Adults have been socialized to hold back our questions because we’re often worried about what other people think, Dr. Merali said. But asking questions not only helps us to gain information, it’s also an important way to build relationships, he said. [...]

Fix your sleep schedule.

Toddlers thrive on routine, and having a schedule with consistent sleep and waking times will help you, too, said Alberto Ramos, a sleep neurologist and researcher with the University of Miami Miller School of Medicine.

If your schedule permits, and if you have the urge, napping also has a host of benefits, including sharper thinking and reaction times and improved memory.

Yes, I love naps, always have. And then there's laugh when you can:

One study found that young children laugh six times as much as adults. But we can seek ways to build playfulness and humor into our day.

Saturday, March 16, 2024

Today's lunch, The Little Grocery, Hoboken

Tight Like This: A Tale of High Adventure in Ancient Nubia [the fourth arena]

Yet another bump, and less than two years after the previous bump, In August of 2022. This time I'm preparing for the 3QD piece that's due of Mar. 24, for posting on the 25th.


I'm bumping this to the top of the queue. Why? Because I can, that's why. Because the world's got to change and it's got to change NOW. You know, now's the time! We've been mugged by the future and don't realize it. We're still waiting for jet packs when the time for jet packs has done come and gone. There isn't going to be any AGI (artificial general intelligence) coming down the pike, not in time to make a difference and maybe not ever. Get real and figure out how we're doing to deal with a network of self-driving cars that can learn. Anyhow, this is going to require major changes in how we think, and that's going to require a new mythology. I figure maybe old Golfotep and the Mystic Jewels have something to say about that.

I published this on The Valve some years ago and, as the introductory note says, I'd published it on Meanderings, now defunct, before that. No doubt I will republish it somewhere else sometime in the future. Or maybe someone will make it into a movie. Who knows.
Oh, and be sure to check out Tight Like This (this is Wynton Freakin' Marsalis, Armstrong link in text below) High drama & low-down hoochie coochie. 


Jivometrics

Back in the mid 1990s, about the time Mosaic was being unleashed on the internet, I met a fellow in the African-American Forum at AOL Online. Called himself Cuda Brown. We hit it off and began emailing privately. Before you know it we were collaborating on a website called Meanderings - which later became Gravity. Cuda did most of the work, including all of the HTML coding and the back-end database coding. I helped editorially, wrote some pieces, and did some art.

One day I sent Cuda an email containing a spur-of-the moment paragraph satirizing Afrocentrism, something much discussed at Meanderings. While Cuda and I were sympathetic, we were doubtful about the more inventive flavors of the brew. Thus I had improvised something about golf being invented by one Pharoah Golfotep: the primo white-shoe Anglo country club plus-fours game was invented in ancient Egypt. What could be sillier? Over the next two weeks, however, this little bit of satire jes grew and grew, like Topsy, and became something else.

So, we posted it in Meanderings, which may have become Gravity by that time, I forget. And people read it. In time, however, Gravity died. Since then I've been looking for a place to revive “Fore Play.” Well, here it is. Let's call it a cultural studies primer for the new millennium.

Some of it is a bit dated. Who remembers Dennis Rodman, much less his experimental tonsorial stylings? Back then Tiger Woods was more potential than achievement – but what potential! – while boom boxes have now bifurcated into iPods and beat boxes. "Muggles" hadn't become a Potteresque term of art for – well, just what exactly, non-magicals? (I've never read any of the books.) Back then I used it as the name of a Louis Armstrong blues and I'm sticking to it. It's also old New Orleans slang for something Bill Clinton didn't inhale.

Otherwise, it's pretty much now as it was then. Only back then the turn of the millennium was in the future. Now it's in the past.



Fore Play:

A Lesson in Jivometric Drummology



Jefferson Ribonucleic Parker IV

aka

Mr. Ribs


Tiger Woods is only the most recent in a long line of fine black golfers. In saying that I refer to players other than the moderns such as Charles Sifford, Jim Thorpe, Jim Dent, Lee Elder, Calvin Peete, and Renee Powell. Truth be told, the tradition of sepia swing masters started in ancient Egypt, where the game was invented. In that company Woods would be no more than a middling player.

By today's standards Tiger is ferociously talented and filled with promise for the future. Perhaps he'll become the best in post-modern times, the primo putter of the 21st Century, the first master to break par on the New Savanna. But those African drivers of ancient Egypt were giants the like of which haven't been seen in thousands of years.

Their stories, like so many stories, have been suppressed by the Europeans. Fortunately many of those stories have been collected by The Order of Mystic Jewels for the Propagation of Grace, Right Living, and Saturday Night through Historic Intervention by Any Means Necessary. The Jewels are dedicated to preserving the ancient stories and to intervening in history in ways variously clever and indirect. They are the chief source of that version of Afrocentric thinking known as Jivometric Drummology. In her classic study, Klactoveededstene: Riffing the Noumena, Ella Birks Roach defined the basic concept thus:
Jivometric Drummology: A philosophical system grounded in African and African-American musical practice. “Drummology” indicates that the governing logos is that of the drum, of rhythm, of hands and sticks coaxing sound from skin, of people joining together, each playing a simple rhythm, with the many simple rhythms melting into a single stream of infinite diversity. “Jivometric” is here because of the way it rolls off the tongue and tickles the ear; its meaning is secondary to its sound. Jivometrics is thus a principle of grace. When jivometrics is in play outer sonic auras join in the creation of tones played by no one, but heard by all. A treatise may have drummological ideas, but if the language lacks grace, then the treatise is not jivometric -- jiveturkey is all too often the appropriate term. In the most profound works of this school jivometrics and drummology are joined through agape

The Mystic Jewels, however, are not mere signifiers. They signify with a stern purpose. For example, Harriet Beecher Stowe was invented by the Mystic Jewels. They knew the abolitionists would never get beyond a lot of grand indignant talking so they figured a novel that stirred the imagination would be just the thing. A light-skinned sister named Eleanor Gough McKay changed her name to Harriet Beecher, married Calvin Stowe, and wrote Uncle Tom's Cabin. That book, as President Abraham Lincoln acknowledged, is what gave the North the guts to wage the Civil War.

Anyhow, my mother's father, Cassius Photon Gaillard, aka Slim, was a Mystic Jewel and took a special interest in the history of golf. The following story is based on information from his papers.

• • • • • • •


The Origin of Golf and the Lights in the Sky


Golf was invented by the ancient Egyptians – well, that's what we call them now, but they were really Nubians, and proud of it, too. Most of the details have been lost, but the general shape and thrust of the story has been preserved.

It began in the reign of Pharaoh Ramses Golfotep LVII of the 'N Baa Dynasty. One day Rams was hanging out with some of his friends in the Lark Meadow gazebo at his summer palace. As usual, they were playing bid whist and sipping Mount Gay and Coke, with a twist of lemon. As so often happens, they got to talkin' trash about their wives and girl friends. Ramses talked about how he particularly liked going into a special glade with his wife Cleopatra and a boom box loaded with some righteous jams. The best time was early evening when things were cooling down and the sun lit the sky with orange fire. They'd meander down this long narrow opening among the palms and get to a secluded spot ringed with patches of sand. The ground was firm and the grass kept closely cropped so they could dance freely. Inevitably the dancing would lead to a little fooling around, and that little fooling around generally led to more and before you know it Cleo was baking Ramses' sweet potato in her oven. That was some fine sweet potato pie they'd cook up. Yes indeed.

On the street [Hoboken]

Ellington's version of jazz history in 3 minutes & 45 seconds

Jazz scholar Lewis Porter has an interesting post featuring an all-but-forgotten composition in which Ellington runs down the history of jazz in three+ minutes. The only available recording is on a short film-clip, which Porter includes in his post.

Here's Porter's account of what happens:

A lot happens in this piece, which actually runs about 3 minutes and 45 seconds. First of all, it’s not a chronological history. It’s a kind of collage, with many short sections at different tempos that could, conceivably, be rearranged at will. It starts with a close-up of the clarinet going into a New Orleans piece—then it suddenly goes into a bebop line, followed by a tiny quote of Basie’s hit “One O’Clock Jump,” then a fast excerpt of that same number, followed by a little bit of stride piano by Duke. It’s from his own piece for Willie “The Lion” Smith, “The Second Portrait of the Lion.” At 1:50 all six saxophones play a ballad while the camera pans from left to right, as we saw above. Suddenly at 1:56 the whole band claps and sings “ah, ah, a-a-ah!” That marks the end of the first half.

The second half begins with Rouse playing, solo, a few phrases from the famous 1939 “Body and Soul” recording by Coleman Hawkins. Then the previously heard ballad continues, now featuring Glenn’s trombone. At 2:36 Duke plays some boogie-woogie, then goes into an abstract version of “Basin Street Blues” while Nance imitates Louis Armstrong with a handkerchief. But it’s a pantomime—he makes no sound! Then there’s a wild fast number, interrupted when we are suddenly taken back to the New Orleans piece that started it all—and finally, once again, “ah, ah, a-a-ah!”:

[FILM CLIP IS HERE]

What a journey! And what does it all mean? Today we might call this “post-modern” for the way it draws freely from such a range of styles, in a seemingly random, chopped-up fashion

This film also represents a step in Duke’s long-standing interest in creating a piece about jazz history. He credited Orson Welles for having gotten him thinking about this in the early 1940s, when Welles commissioned Duke to work on a history of jazz. Welles paid Duke in advance for that project, but it never happened. Soon afterward, in 1943, Ellington premiered his major suite, Black Brown and Beige. This was not a history of jazz—Duke’s subtitle was “A Tone Parallel to the History of the Negro in America"— but one could see these projects as related, since both involve creating musical works that offer a long historical view. He returned to jazz history in 1950 with the piece you just watched. The next year, 1951, he wrote his Controversial Suite with its two chronological sections, “Before My Time,” and “Later.” Then came A Drum is A Woman, recorded in 1956 and televised in 1957, which Ellington explicitly referred to as a history of jazz.

By all means, follow the link to watch the clip. It's marvelous.

Reckless abandon – the wisdom of Barry Harris [Master Class]

Friday, March 15, 2024

Power to the people [Hoboken]

How to power the AIs. It's a crazy new world.

Casey Handmer's Blog, How to Feed the AIs, March 12, 2024.

AIs use a LOT of power. Producing that power is one thing. Getting it to the AIs is another.

The challenge with grid development is that it’s a permitting and construction nightmare, and like many things in the West, has gotten extremely expensive and time consuming to execute. But the datacenters need power much sooner than conventional development can build new power plants and transmission lines – even if permitting, eminent domain, technology, and construction costs were a solved problem!

It is time to lift the constraint that a new data center must be attached to the grid. Instead, we can provide most of the power locally using “beyond the grid” renewable energy generation, backed up by batteries for energy storage overnight.

For example:

A 1 GW data center (containing roughly a million H100s!) would have a substantial footprint of 20,000 acres, almost all of that solar panels. The batteries for storage and data center itself would occupy only a few of those acres. This is in some sense analogous to a relatively compact city surrounded by extensive farmland to produce its food.

Of course future data centers will use more advanced and productive computers but their power consumption and heat generation will remain much the same. Radically more efficient panels, batteries, and computers will only increase the revenue per unit land used for this purpose.

The development cost of a 1 GW data center would be around $60b including the solar+battery power plant, and importantly the lead time and permitting complexity for each of the components: solar array, batteries, structures, racks, internet connection, and servers is about the same. This is in direct contrast to conventional on-grid power supply, whose capacity is nearly exhausted already.

Is this going to pave the Earth with solar?

My startup Terraform Industries looks to apply solar to produce synthetic fuel, consuming substantial amounts of land (though less than agriculture) in the process. Something like 2 billion acres, or 7% of Earth’s land surface area, would be sufficient to provide every man, woman, and child on Earth with US levels of oil and gas abundance and commensurate prosperity. It’s possible to imagine a future where people consume even more than that – widespread personal supersonic transport, for example – but ongoing conversion of land use away from intensive industrial agriculture toward inherently more productive solar synthetics is a clear net win for the environment.

Is this the case for solar AGI? It’s currently hard to imagine Nvidia and friends producing 100 billion or more H100s, but it’s also hard to imagine our collective demand for artificial intelligence will saturate. If we’re spending $60b on a 20,000 acre solar powered data center development, that’s about $3m/acre. Even if the land acquisition budget is only 3% of the budget at $100,000/acre, this is still substantially higher than essentially all land outside of major cities, including prime agricultural land. Is AI a solar panel maximizer, rather than a paperclip maximizer?

Will AIs prefer to live in space instead of on the Earth?

There's much more at the link, like this crazy-ass final paragraph:

It seems that AGI will create an irresistibly strong economic forcing function to pave the entire world with solar panels – including the oceans. We should probably think about how we want this to play out. At current rates of progress, we have about 20 years before paving is complete.

Friday Fotos: The saucer has landed [New park in Hoboken]