michaelgogins
Michael Gogins
Computer music, photography, some poetry and fiction, and something more or less like theology or philosophy of religion
515 posts
Don't wanna be here? Send us removal request.
Last Seen Blogs
darakna-tiaman-blog
• DARAKNA TIAMAN •
akalurkey-blog
The Food Addict
gureumz
pick your poison
artcorporation
Art Corporation
naughty-a-blog
n o v e l e l t y n a d i a
Text
More Notes on the Computer Music Playpen
I have finished maintenance on the VST3 plugin opcodes for Csound, Csound for Android, and some other things, and am re-focusing in composition.
One thing that happened as I was cleaning up the VST3 opcodes is that I discovered a very important thing. There are computer music programs that function as VST3 plugins and that significantly exceed the quality or power what Csound has so far done on it own, just for examples that I am using or plan to use:
The Valhalla reverbs by Sean Costello -- I think these actually derive from a reverb design that Sean did in the 1990s when he and I both were attending the Woof meetings at Columbia University. Sean's reverb design was ported first to Csound orchestra code, and then to C as a built-in opcode. It's the best and most widely used reverb in Csound, but it's not as good as the Valhalla reverbs, partly because the Valhalla reverbs can do a good job of preserving stereo.
Cardinal -- This is a fairly complete port of the VCV Rack "virtual Eurorack" patchable modular synthesiser not only to a VST3 plugin, but also to a WebAssembly module. This is exactly like sticking a very good Eurorack synthesizer right into Csound.
plugdata -- This is most of Pure Data, but with a slightly different and somewhat smoother user interface, as a VST3 plugin.
I also discovered that some popular digital audio workstations (DAWs), the workhorses of the popular music production industry, can embed algorithmic composition code written in a scripting language. For example, Reaper can host scripts written in Lua or Python, both of which are entirely capable of sophisticated algorithmic composition, and both of which have Csound APIs. And of course any of these DAWs can host Csound in the form of a Cabbage plugin.
All of this raises for me the question: What's the playpen? What's the most productive environment for me to compose in? Is it a DAW that now embeds my algorithms and my Csound instruments, or is it just code?
Right now the answer is not simply code, but specifically HTML5 code. And here is my experience and my reasons for not jumping to a DAW.
I don't want my pieces to break. I want my pieces to run in 20 years (assuming I am still around) just as they run today. Both HTML5 and Csound are "versionless" in the sense that they intend, and mostly succeed, in preserving complete backwards compatibility. There are Csound pieces from before 1990 that run just fine today -- that's over 33 years. But DAWs, so far, don't have such a good record in this department. I think many people find they have to keep porting older pieces to keep then running in newer software.
I'm always using a lot of resources, a lot of software, a lot of libraries. The HTML5 environment just makes this a great deal easier. Any piece of software that either is written in JavaScript or WebAssembly, or provides a JavaScript interface, can be used in a piece with a little but of JavaScript glue code. That includes Csound itself, my algorithmic composition software CsoundAC, the live coding system Strudel, and now Cardinal.
The Web browser itself contains a fantastic panoply of software, notably WebGL and WebAudio, so it's very easy to do visual music in the HTML5 environment.
2 notes
·
View notes
Text
Fractals, Roots, and Mapping
Long ago I published an article, "Iterated Function Systems Music" (Computer Music Journal 15.1, 1991). Later I published a related article, "…How I Became Obsessed with Finding a Mandelbrot Set for Sounds" (News of Music 13, 1992).
As my CMJ article showed, we already know from the Collage Theorem that an iterated function system can approximate as closely as desired any set, including a musical score or even a soundfile.
My musical motivation here is that, considering such an IFS as a Julia set, there is, near the point in the Mandelbrot set that generates a Julia set, a region that resembles that set in shape. This provides a new method of composing music, by exploring the Mandelbrot set to identify parameters that produce what looks to be an interesting score. I call this "parametric composition." This method has the potential to greatly simplify and speed up the composition of music that a composer does not simply imagine, and might find difficult or even impossible to imagine.
Over the years I have taken some time, now and then, to refine these ideas. I now have a working prototype of part of the idea -- the part about using a fractal function to represent (a slightly simplified form of) any possible musical score.
Today I found on the blog ofJohn Baez (a mathematical physicist, and a cousin of Joan) a link to an article by him and colleagues in Notices of the American Mathematical Society, "The Beauty of Roots". In this article the authors discuss how a map of the roots of Littlewood polynomials closely resembles the attractor, or dragon, created by mapping two contractive functions of the complex plane onto itself -- when the parameter of the contractive mapping is a root of the polynomial! You can play with these polynomials here, and with the duality here.
Anyway, the interest for me is the duality of generators with maps of generators, and Baez' article shows that this idea extends very deeply. There may be more than one way to derive a fractal function whose attractor is the graph of a score!
2 notes
·
View notes
Text
Talk/Performance at Luck Dragon
This coming Friday the 27th, at 6 PM, at Luck Dragon on 100 Main Street in Delhi, New York, I will present my browser-based computer music system cloud-5, play some music, and talk about algorithmic composition.
The slides for my talk are here and -- if you download the PDF! -- all links should be "live."
2 notes
·
View notes
Text
New beta release of cloud-5
This includes the piece that I will be performing at Brooklyn College tonight. Get it here.
2 notes
·
View notes
Text
Upcoming Performances
Also this: sound evening at Luck Dragon in Delhi, New York, a mere 16 miles (we drive that far just to buy groceries) down the road from our farm in the Catskills. Yes, computer music has finally found a home in Delaware County.
I will be one of the two presenters; I will demonstrate my cloud-5 system, perform a piece using cloud-5, discuss that a bit, play a straight ahead piece of what we sometimes still call "tape music," and discuss everything.
0 notes
Text
John Zorn Is Now Streaming
For most of his career, the American composer and performer John Zorn has resisted putting his music into streaming services. He is now celebrating his 70th year, and he has finally released what looks like all, or almost all, of his catalog to streaming. I have been listening to him now for several days on Amazon Music.
This is an immense gift. Zorn is one of the greatest composers in American history. And he is amazingly prolific, partly because much of what he does is to organize directed small group improvisations and simply record them. His crew of fellow performers is stellar, and most of them have been working together for decades.
If you're new to Zorn, listen to The Big Gundown and then just sample his works.
Highly, highly recommended.
Zorn does not himself, to the best of my knowledge, ever do computer music, electronic music, or much post-production (although he does compose for classical music ensembles, and he does work with people who have made important computer music, such as Ikue Mori). However, Zorn's compositional methods are relevant to computer musicians and algorithmic composers, especially with respect to live coding.
I plan to find out as much as I can about Zorn's methods (this might be a starting point), and adapt those that make sense in the context of computer music in my own work, for example in cloud-5.
0 notes
Text
Cosmology and Philosophy
This is my response to "Is There a Crisis in Cosmology?," by Adam Frank and Marcelo Gleiser, published in the New York Times on September 2, 2023.
TL;DR: I agree that there is a crisis in cosmology, but I do not agree that it is anything other than a normal crisis of the sort that appears in science when successful theories are applied to phenomena that they cannot explain. In other words, I don't think it's wise to take a crisis of this sort as an excuse to think that science must give way to philosophy, or even to religion, to deal with empirical facts. I rather think it means that science will end up finding explanations for the anomalous phenomena discussed in the article, such as "dark matter" and "dark gravity." But it could take a while! After all, it took 228 years after Newton published his Philosophiæ Naturalis Principia Mathematica for Albert Einstein to publish "Die Feldgleichungen der Gravitation".
In my view, the notion that the laws of physics evolve over time is a confused argument. If the laws appear to change, what causes this change? The only explanation is that a more fundamental law of physics caused the apparent, or emergent, laws of physics to change. The entire assumption of the scientific method is that the laws of physics are truly universal and apply in all places, at all times, to all objective phenomena.
It is entirely possible that a more fundamental law of physics describes a symmetry (most laws of physics in field theories are symmetries) that for one reason or another was broken, leading to new "emergent" or apparent laws of physics, early in our universe. This would be just like a phase change, as if the whole universe before our universe was liquid, but then flashed into steam, and that steam is our universe. But far from being any change in the way physics thinks, it is a more fundamental application of the same old way of thinking pioneered by Galileo and Newton.
The difficulties that science encounters in trying to model the physical world -- all of it, the whole universe -- from within the physical world are real. These difficulties most likely have nothing at all to do with philosophy or with consciousness as such. Rather, I believe, it means that the evolution of physical phenomena from the truly fundamental physical laws is chaotic, and that means that some well-formed statements about the physical world are simply undecidable. It's not that laws to do not apply, it means that the laws cannot be used to prove that some particular physical state actually follows from the laws. For example, there is a phenomenon in physics known as the "spectral gap." The gap is the energy difference between the ground state of a system and its first excited state. It has been proved that, in a toy physical universe with only 2 dimensions that is infinite in extent, the spectral gap is not computable. This does not mean that the gap is not real. It is. This does not mean that the gap does not have an actual value. It does. But it does, however, mean that the actual value cannot be computed.
The kinds of difficulties being encountered in cosmology today are nothing new and have nothing to do with the validity, or invalidity, or limits, of the scientific method. Ptolemy explained the motions of the planets very well, but his theory kept getting more and more complex, but then Copernicus came up with a better and simpler explanation. Newtonian gravity explains the motions of heavenly bodies very well, but not the retrograde motion of Mercury, but then Einsteinian gravity does explain it. Now there is dark matter and dark energy, and I am quite willing to bet that a new theory will explain these things very well without needing to change the scientific method or the fundamental assumptions baked into it.
The question hinted at by Frank and Gleiser is whether human consciousness is somehow required to complete the laws of physics. This is a subtle issue. In my view, the reality of human consciousness, agency, and moral responsibility is at least as fundamental as the laws of physics. After all, science and the laws of physics, as human activities, are based upon human sense perceptions, reason, and actions. And if we get stuck in doing physics for another 228 years, the only way we can make progress is by continuing to apply these informal powers. Only about 121 more years to go!
To question the reality and effectiveness of informal powers, is to question the reality and effectiveness of the laws of physics produced by them. But I believe it is a category mistake to think that consciousness is required to be effective in the same sense as the symmetries underlying fundamental theories. Think, if you will, like a theist (I am a theist). This would involve miracles. God would have to stick his pinkies through the curtain and fiddle with the dolls on his little stage. And that would imply, in turn, that God is not really God. That kind of a God would, in reality, be a part of Nature, not utterly beyond and above Nature. And we, if we are made in the image of God, as I think we are, must transcend the world in the same way. Far from implying that God did not really create the world from nothing, or that we do not really have freedom or the gift and burden of moral responsibility, it gives God, and us, the absolute freedom, beyond all time and space, that we do have.
0 notes
Text
The Causal Potency of Consciousness
The other day I read The causal potency of consciousness in the physical world on the arXiv, by Danko Georgiev.
I know little about this author, other than that he has several other preprints on the arXiv on this and related subjects. His affiliation is listed as the Institute for Advanced Study in Varna, but on the Web site of that organization, he appears as the only author, so that appears to be a shell for him. Some of the papers listed there do have other authors in addition to Georgiev.
Georgiev also is the author of the textbook Quantum Information and Consciousness, published by CRC Press and available from Routledge.
And Georgiev is the chief editor of the online, open access journal Quanta (not to be confused with the online magazine Quanta!). Some of the authors published in the open access journal Quanta appear in other peer-reviewed journals of academic science (e.g. Stan Gudder).
Whatever. Georgiev is not a professor, nor is he an institutional researcher, and he appears to be near the fringe.
But he is not on the fringe, or beyond it.
Georgiev's paper presents ideas that I myself had a year or so ago, when I was regularly answering questions on Quora and used that as an excuse to do some amateur philosophizing. And these thoughts of mine go way back to my twenties, when I was trying to relate Chaitin's ideas about Kolmogorov complexity to self-consciousness.
The paper is written in a clear and accessible style and makes several arguments:
Consciousness must provide a selective advantage to conscious animals, including human beings. I had also made this argument, but I am not 100% sure the argument does not beg the question. After all, it might be objected that something physical both causes consciousness and provides a selective advantage.
Consciousness, as a quantum process, observes itself and this is a quantum measurement. I had also made this argument, but again, I am not 100% sure that the argument does not beg the question.
Consciousness has no causal potency from the standpoint of classical physics. I agree.
The quantum state of an organism is capable of indeterminate behavior, which in conscious organisms can be identified with free will, which is required for any satisfying theory of moral responsibility. I also had made this argument, and I went a bit further and pointed out that a free choice must appear random from the standpoint of classical physics, yet is rationally motivated from the standpoint of consciousness. In my view, this overcomes the objection that random behavior cannot be free.
I also have made the argument that human self-consciousness involves a fixed point that is not possible in classical physics, and may not be possible even in quantum physics.
And I have argued that quantum indeterminacy means the phenomenal world is not computable.
I am still not happy with the clarity and completeness of my arguments, and I am still working on them. But I do not doubt that they are all wound up with the points above.
Yet it is very nice to see that these ideas are circulating and may receive the attention and further research I think they deserve.
3 notes
·
View notes
Text
NYCEMF!
Today is the first of the seven days of the New York City Electroacoustic Music Festival, from June 19 through June 25. Registration, tickets, and the complete program can be found at https://nycemf.org/. I have a piece of computer music, "snow-voice", algorithmically composed and synthesized, in Concert 15 at 8:00 PM June 21, in the Loreto Theater of the Sheen Center for Thought and Culture, at 18 Bleecker Street, New York City. If you like electronic music, computer music, soundscapes, or any of the sometimes illustrious composers and performers to be heard, come!
1 note
·
View note
Text
Homelessness/Housing
In California a few years ago, visiting friends in Albany just north of Berkeley, I walked to the Berkeley marina and just around the corners of some side streets I saw many parked motorhomes and recreational vehicles. People, obviously, were living in them.
Now the news is that Los Angeles also has this (to the tune of 40,000 people) and the mayor, Karen Bass, is saying she will do something about it. She mentions tents, and I saw those too in Berkeley, but what strikes me is the motorhomes.
Let's pause here a moment and think. One -- not the only one, but a major one -- of the underlying causes here is the sheer cost of housing. In real money, the cost of housing has almost doubled while, at the same, real income is stagnant. But motorhomes, recreational vehicles, camper vans, travel trailers (I saw all of these too) are much cheaper than zoned housing. For many reasons, not least of which is "not in my back yard," which ends up enforcing class and race segregation.
One might well ask, how come there are people who don't have houses to live, in but do have motorhomes to live in? I didn't do a controlled study but the obvious hypothesis is: simply because motorhomes are cheaper.
So I have a modest suggestion. There should be a national law permitting manufactured housing, including motorhomes and travel trailers, in residential neighborhoods and, perhaps, in specially zoned neighborhoods. Such a law would evoke howls of outrage: "Ugly! Dirty! Unsafe! Not in keeping with the character of the neighborhood!"
"Ugly" is a question of taste, but the law could deal with it by banning features widely felt to be ugly and requiring some widely felt to be attractive. "Dirty" can be dealt with by installing the requisite hookups, or even by having a public utility to come by periodically with a service truck for gas and sewage. "Unsafe" is simply not true. "Not in keeping with the character of the neighborhood" would, of course, usually be true.
There's no political capital right now for such a law, and the rich will always find some way of keeping this stuff out of their view, but something has to give.
Up to now, trailer parks are, indeed, usually kind of trashy. But there are increasingly attractive manufactured homes, tiny homes, container houses, and, yes, motorhomes.
So when overpopulation (yes, that's also a major cause of this) and climate change (up to a billion people will have to move in the next 50-100 years -- where to, pray tell, and where will they then live?) increase the pressure on housing prices even further --
The law should and probably will change to permit manufactured housing, if it meets certain standards, in more places including in many current residential zones.
Currently within 200 miles of my Zip code (13740) one may buy new or used motor homes within a price range of $50,000 to $200,000. Most of these things are small and, unfortunately, hideously ugly. But, they don't have to be ugly.
And, they are homes you can live in!
Manufactured housing can be bought for similar prices, is bigger, and is somewhat better looking. This, too, does not have to be ugly. There is a "designer" variety of manufactured housing. And it doesn't always cost a whole lot more than ordinary ugly old modular homes.
0 notes
Text
Guidance for Computer Musicians
Focus on what you love doing, and do it with the minimum of means.
This is something I have struggled with for decades. Today I went to the "Sound" event at Luck Dragon in Delhi, New York. Brian Crabtree spoke at length about his approach to running a small business supporting the arts (he makes the Monome controller, the Norns computer, and a few other things), and this phrase came to my mind.
Of course I have been pruning back my projects and my software stack now for several years. What initially totally fascinated me and what I loved was just taking fractals from Martin Gardner's Mathematical Recreations column in The Scientific American and trying to get music out of them. Most of these attempts were a page, at most a few pages, of code, BASIC or Fortran.
Brian identified many problems in the general world of software that I had encountered in my own work. These include bloat, semi-competence, too many cooks in the kitchen constantly re-inventing all the wheels, focus on money, and so on. His response is to focus on small, standardized, self-manufactured computers dedicated to his general approach to music. He pointed out that this simple platform make it easier to create a community.
My response is different, to focus on Csound in the Web browser as the platform, which also is standardized, has facilities for music, and is versionless and runs on all platforms.
Brian is much more interested than I am in fostering a collective, a community of people speaking from the same level, although of course in reality he is the wellspring.
I am much more interested in making the kind of music that I want to hear.
Or perhaps Brian just has more social skills than I do.
What I take away from these thoughts is that the computer is a double-edged sword. It is the engine of our current form of hyper-capitalism. But to keep that capitalist system running, it must nevertheless be based on inexpensive, easily programmed, general-purpose computers, operating systems, and compilers. Which can very easily be turned to other purposes. That will always be the case, even in a world of conformity, materialism, and surveillance.
If you do what you love, but with the minimum of means, you will not be spending time and money on tools you don't really need. You will not be following the fads of the programming world. You may not even have a boss. And even if you don't have much money, you will still be able to do what you love.
3 notes
·
View notes
Text
FTL
FTL, as most readers will know, is the science-fictional acronym for faster-than-light travel.
A very common idea is that special relativity forbids FTL because accelerating any mass to the speed of light requires an infinite amount of energy. Which is true.
However, I just watched this video talk by Sabine Hossenfelder suggesting that this common idea is, to put it bluntly, full of holes if extended to cover the real world.
I was expecting her to expound on warp drives, which are suggested by solutions of general relativity theory inspired by the apparent fact that our universe inflated space much faster than the speed of light for a very short time at the very beginning of the Big Bang.
But no, Hossenfelder's talk is very interesting in a very different way. She is a real physicist with publications, even though she is not a tenured professor, and her specialty is the phenomenology of quantum gravity, although she works in other areas as well.
I myself am no such specialist and so my understanding of her post, though it is aimed at non-specialists, has got to be limited. But the things I take away are:
Objections to FTL are most often motivated by special relativity, but our best theory of gravity is general relativity.
Hossenfelder is suspicious of the singularity represented by an infinite amount of energy being needed to accelerate a mass to the speed of light. She seems to think this is a big hint that not only special relativity, but also general relativity, are incomplete and misleading.
Current cosmology based on general relativity features a co-moving frame within our horizon, which Sabine thinks obviates arguments about temporal loops arising from FTL.
I found Hossenfelder's talk to be most impressive, and it has affected my own thinking about FTL.
As a one time aspiring science-fiction writer, I always wanted to base my writing on the best science, and to my mind, the best science pretty much ruled out FTL. But now, not only Hossenfelder's talk, but also continued attention by real physicists to warp drives and wormholes, make me aware that objections to FTL are based on assumptions that skip over a number of genuine gaps in our understanding.
I therefore give myself permission to go back to a future in which humanity has invented and uses FTL. It's highly speculative, but it can't be ruled out; therefore, it is a good premise for an optimistic future (or a pessimistic one -- perhaps UFOs are warp vessels spying on us to see if we need to be exterminated!).
4 notes
·
View notes
Text
New York City
Today, Heidi and I took the NYC Ferry from Wall Street to Rockaway Beach and back. I spent all of the trip out and most of the trip back on top of the ferry boat, looking at the view.
It was a partly cloudy day, breezy, indeed cool.
What struck me above anything else was the incomprehensible scale of the town. In the decades since I first visited NYC in 1972, the skyline of Manhattan now rises above the Empire State Building both downtown and in midtown, and clusters of high-rises also have appeared in Jersey City and in downtown Brooklyn.
The skyline in my photograph is 15 miles from the ferry, which is just off the neighborhood of Manhattan Beach. The left cluster of buildings is downtown Manhattan plus Jersey City. The right cluster of buildings is midtown Manhattan plus downtown Brooklyn. The visible skyline is about 6 miles across.
But there is more. The business of the city, including the harbor from JFK Airport to Highlands in New Jersey, with the Port of Elizabeth clearly visible, to the ships riding at anchor out in the open ocean off Long Island, to things not visible such as suburbs in Eastern and Central New Jersey, the western part of Long Island, the Hudson Valley, and southern Connecticut, lies within a radius of about 80 miles and is the home of anywhere from 19 to 22 million people, depending on how you count. Just the harbor and rivers are vast.
I know there are bigger cities. I have looked down on Sao Paulo from a jetliner and been similarly flabbergasted by a seemingly endless forest of white high-rises.
But what I am focusing on and trying to explain here is the feeling I had traveling through this landscape and seascape. I felt like an ant who has climbed up from the nest and on to a rock and suddenly can see more than a few inches. I could truly see where I live and I could truly begin to comprehend what it is. This was quite a bit more impressive than the already impressive views from jetliners landing or taking off, or from cars driving along the Brooklyn-Queens Expressway, because this view from the ferry was long, slow, and uninterrupted.
The entire skyline in the picture was built by the hands of men and a few women in about 120 years.
It is still growing -- fast.
0 notes
Text
Xenakis
It's not that I go back and listen again to Iannis Xenakis.
Rather, it's that when I did listen to it, even decades ago, it went in so deep that I am still hearing it.
And I'm still trying to use it.
1 note
·
View note
Text
Artificial Intelligence and What's to Come...
This is my current working summary of thoughts regarding artificial intelligence and its potential dangers.
What controls what ultimately happens is not "intelligence" or "consciousness" (both of which I take to be both very real and very important), but evolutionary theory.
In other words, what's important is whether AIs are reproducing in the biological sense and, if so, what if any is their selective advantage over human beings. The best summary that I know of how natural selection actually works is John Baez' work on the Fundamental Theorem of Natural Selection. It implies that a high variance of fitness and high speed of reproduction in AIs could give them a serious selective advantage -- but only if they reproduce autonomously. Right now, the only way AIs can reproduce is if people copy them. Right now, this is far from autonomous -- we can just shut them down if we don't like them! And right now, AI fitness is simply how well AIs do what we want.
In the future we might get "smart computer viruses" that fool us into copying them. This is one of the main things the doomsters worry about, and I agree with them that at some point that could be a real threat. Until such time as there are literally self-reproducing machines, von Neumann machines with replicators (this is the other main thing doomsters worry about), any AIs will need to keep us around in order to reproduce themselves, and their fitness will simply be how well they serve us. But that might be cold comfort, because how well AIs serve us (in terms of our Darwinian fitness) might be quite different from what we think it is. After all, we've improved the Darwinian fitness of cattle quite a bit by keeping them and breeding them so that we can milk and eat them, kind of like in the Matrix films. I repeat, none of this requires AIs to be conscious or to have goals.
To the best of my knowledge, nothing in any current AI architectures involves any knowledge of external reality, any consciousness, or any means of judging truth that is not encoded in high-dimensional correlations between tokens (words, phrases, and meaningful fragments of words) on the Internet. However, the huge discovery of AI is that these correlations can be exploited in AMAZINGLY POWERFUL WAYS.
As long as there is no truth-testing by AIs, everything produced from them has to be truth-tested by some, hopefully expert, human individual or group. Probably because of this, there's no indication of any ability of AIs to produce what scientists call "major results" or artists call "great works." If there were, I'm sure it would be big news, and if there hasn't been, I'm sure it's not for lack of trying.
So, what are some of the potential dangers of AIs?
Nation states, with the help of experts, will probably try to weaponize AIs. I do not know how to estimate this risk. But, if I were responsible for my nation's defense, the precautionary principle would make me assign high priority, big funds, and major talents to it. I sure wouldn't want to wake up one day and find all the computers in all my weapon systems had been taken over by an enemy sending out super-smart computer viruses!
Human subcultures, political parties, businesses, and even individuals will use AIs to make propaganda. This has already started happening, and I have no idea of how far it will go, or what the consequences might be.
Individual human beings may find it increasingly difficult to know what is true. This is already hard enough. Social media have injured democracy around the world, and AI could kill it.
Individuals may find their screens even more addictive than they already are. One wonders, for example, about the impact of sexbots run by AIs.
I've been fooling around online with ChatGPT and DALL-e, and I plan to do more of that. At first, I could trip ChatGPT up in a few minutes, by asking it to prove theorems, or by getting it to contradict itself. But this has definitely gotten harder.
2 notes
·
View notes
Text
Just an Argument...
This is another in my ongoing series of posts expressing a single philosophical argument in the style of Just the Arguments by Michael Bruce and Steven Barbona, a book I highly recommend.
My simple argument today is: Artificial general intelligence is not possible in a computer (a Turing machine). Here I define artificial general intelligence as scoring in the normal range, or higher, on a standard IQ test, being conscious, and having moral responsibility. I also assume the logical equivalence of digital computers and quantum computers, which has already been proved.
I have no doubt that in some sense, a computer can score in the normal range, or higher, on a standard IQ test. This is clear from the excellent performance in answering questions of current "chatbots" such as ChatGPT.
My argument then comes down to consciousness and moral responsibility. Actually, because there are people who are conscious and intelligent yet appear to completely lack moral responsibility, i.e. they are psychopaths, my argument comes down to consciousness.
The critique of self-consciousness as self-representation goes all the way back to Aristotle, and recurs in the history of philosophy, up through Sartre's critique of Husserl's transcendental ego and MacKay's critique of the logical indeterminacy of self-knowledge.
My argument is that self-consciousness involves a fixed point in which a conscious subject takes itself as an object, that is, a conscious subject is conscious of being conscious. The key point is that self-consciousness does not involve or depend on an infinite regress of self-reference. It is true that the phenomenology of self-consciousness involves a subtle shifting back and forth from being conscious of oneself as object and simply being conscious of oneself (as subject), or for that matter implicitly being conscious of oneself as a condition for any kind of consciousness (in the form of pre-reflective self-awareness), but nevertheless in both phases there is a perfect coincidence of subject and object in self-consciousness. This is not possible in a computer. A computer can model self-consciousness as self-reference, but this self-reference cannot include itself.
One possible perspective on this difference between self-consciousness and computers is that computers operate step by discrete step, whereas self-consciousness without self-reference involves a fixed point that is possible only in a continuum. Even quantum computers, which are analog computers that do operate in a continuum, operate step by discrete step in the sense of implementing a set of discrete logic gates.
My argument is not a proof. It is a conditional proof. My argument is that, if self-consciousness cannot be reduced to self-representation (a view shared by Aristotle, most Buddhist thinkers, Sartre, and MacKay), then computers cannot be self-conscious.
1 note
·
View note
Text
Just an Argument...
This is the first of what is planned as an ongoing series of posts expressing a single philosophical argument in the style of Just the Arguments by Michael Bruce and Steven Barbona, a book I highly recommend.
My simple argument today is: We can be almost completely certain that humanity can create artificial general intelligence. Here, I define artificial general intelligence as being able to score in the normal or high range on a standard IQ test, being conscious, and having a sense of moral responsibility.
Human beings are living organisms, endowed with general intelligence, who reproduce sexually.
Scientists have already created fully synthetic living organisms that divide and reproduce normally. These artificial organisms, however, reproduce asexually. Also, only the genes of these artificial organisms were synthesized. These synthetic genes were inserted in place of the natural genes in host micro-organisms.
Of course, the human genome is far more complex than this very simple bacterial genome.
To produce artificial general intelligence the following tasks must be performed:
Synthesis of a complete male human genome. This would have to include considerable supplemental molecular structure that does not appear in bacteria, such as histones.
Synthesis of a complete female human genome in the same manner.
Insertion of the synthetic male genome into a sperm cell.
Insertion of the synthetic female genome into a human egg cell.
Fertilization of the egg cell with the sperm cell.
The egg should then begin to divide and grow, and be implanted in the wall of a woman's uterus.
The fetus grows and is born in the normal way.
The infant is raised in the normal way.
If the child seems as conscious as any other child, learns to speak and read, to read and write, scores in the normal range or higher on an IQ test, and behaves in a morally responsible way, then artificial general intelligence has been proved possible.
The weakest links in this chain of tasks are (1) and (2). Tasks (3) and (4) resemble somatic nuclear transfers that has been done many times in the laboratory to clone mammals. The succeeding tasks have all been performed many times in in vitro fertilization.
To repeat, the weakest link is simply synthesizing the fantastically complex and tiny human genome.
However, since this has been done for microorganisms, there is no obvious point where Nature says "you can synthesize a germ, but you can't synthesis an egg or sperm." That is, there is no barrier in principle.
Therefore, it is almost completely certain that, in principle, human beings can create artificial general intelligence in the fullest sense.
Please note, this argument says nothing about artificial general intelligence implemented using Turing machines, or about possible limits or side effects of synthetic biological artificial general intelligence, or how long it might take to learn how to actually do this.
In particular, just as with synthetic mycoplasmas, we have no idea what a lot of the genes in our cells are actually up to. Attempting this feat would, no doubt, teach us a great deal about that. The knowledge gained in this way might even shed light on whether artificial general intelligence in Turing machines is possible, or not, and if so what kinds of structures it requires.
1 note
·
View note