Welcome to one of the most active flamenco sites on the Internet. Guests can read most posts but if you want to participate click here to register.
This site is dedicated to the memory of Paco de Lucía, Ron Mitchell, Guy Williams, Linda Elvira, Philip John Lee, Craig Eros, Ben Woods, David Serva and Tom Blackshear who went ahead of us.
We receive 12,200 visitors a month from 200 countries and 1.7 million page impressions a year. To advertise on this site please contact us.
Amazing what happens when a culture loses the last vestiges of understanding that conceptualization, like any other skill, requires knowledge of the right and the wrong of the process. It can be done well, but it can also be done poorly.
These days just about everyone but the biologists is doing it poorly. The biologists are better not because they think about epistemology, but because working with categories, i.e. making connections and differentiating correctly, is in-your-face crucial to their field.
I had a chat with my buddy who works for google AI and he confirmed, AI is BS.
I haven't looked at your videos but can agree with the above without doing so. "AI" is just a very misleading marketing term. There is no intelligence in the AI-labelled language models, for example. They are essentially pattern-matching algorithms relying on very large human-created datasets with some weights and rules on how to pick the next statistically most plausible word; true or false does not enter into that.
Those TwoSetViolin videos helped get me through the pandemic, lol. Their feud with the bass player Davie504 was epic!
These guys are excellent content creators. The Davie504 therapy endings to some of his videos are hilarious, so perfectly idiotic. Thanks for reminding me of them :)
AI exists, and is employed in daliy use by much of the population.
Like most neologisms made from pre-existing words, "artificial intelligence" is open to serious misinterpretation. Mathematicians are used to this, so are physicists. I have followed AI, sometimes closely, sometimes rather loosely, since the early 1960s. In early days "artificial intelligence" was routinely ridiculed as "genuine stupidity." But in recent years a lot of progress has been made employing machines in such fields as speech recognition, image sorting, and so on.
Although I use, nearly daily, the computerized speech recognition which often answers the telephone, and Google Translate, Google Search and other AI applications, i remain conscious of the limitations of even such impressive apps as ChatGPT, which has passed the Bar Exam in at least one state. ChatGPT still comes up with fantasies or outright falsehoods. What's impressive about it is its ability to come up with long stretches of coherent, believable text using such easily described concepts--unrelated to the usual descriptions of human intelligence.
Do these same concepts underlie much of human intelligence? ¿Quién sabe?
Machines remain incapable of full human intelligence. I wonder whether they are capable even of the problem solving capabilities of apes or ravens.
But examples of practical AI exhibit profound progress since the early 1960s. As a grad student I worked at the University of Texas Linguistics Research Center. It spent $millions per year trying to translate languages via computer. My good friend and colleague David Senechalle (RIP) called it the "Linguistics Berserk Center." He used to say, "If we had a computer with infinite storage which could do everything at once, we still wouldn't know how to program it."
Eventually people gave up on the deterministic generative grammar approach employed by the Linguistics Berserk Center. Translating from English to Russian and back again produced such hilarious results as "The spirit is willing but the flesh is weak"==>"The soul is strong but the meat is rancid."
These days I regularly rely upon Google Translate to interpret swathes of Japanese or Thai characters into English. I do it because Google does such a good job with languages I know. I will regularly post a Google translation from Spanish to English, maybe editing only a word or two, or altering a word-for-word rendering into something more idiomatic.
Google's technique is accurately described as AI, in the technical sense of the phrase. The app is trained on large bodies of translated text, and works accurately on texts it has never seen before.
But if I want a brief translation from Russian, I still ask Larisa. In this task, I trust her English better than Google's. And I can ask her about nuance.
So AI exists, and is widely employed practically, but the word "intelligence" in the phrase "artificial intelligence" doesn't mean the same thing as when it applies to a human, an ape or a raven.
So AI exists, and is widely employed practically, but the word "intelligence" in the phrase "artificial intelligence" doesn't mean the same thing as when it applies to a human, an ape or a raven.
So this is the main point I guess, the word “intelligence” now has to have more than one meaning. Perhaps an “Artificial Intellectual” is what the true smart robot of the future could be distinguished as? An ape has intelligence but can’t be an “intellectual” entity. I realized recently that there is truly a difference in the term “time” when used by a physicist vs a layperson or musician. For a musician or layperson time is a CONSTANT of nature, as proven by a true “groove” of rhythm where the future is set and no one could play in rhythm if this was somehow interpretive or malleable. So what happened several minutes before the Big Bang is not meaningless at all, it is a legit concern because that minute is a different concept of “time”. It is the “count off” before the downbeat of the first measure, for example. Where as for physics time is interwoven with space and malleable, where c is the constant by contrast. A minute before time or space is “created” has no meaning. So a never-to-be resolved conflict between two usages of the same word.
Also the concept of “time travel”. Ok, we understand that toward the future is seemingly possible because our forward direction of time slows relative to the background at high accelerations. But it requires a “bubble” of isolation, another “universe” which separates us and our time experience from the outside universe, such that our clocks can tick at different rates. But the reverse? Not ever possible, because it requires this same type of bubble to “jump” backward, again maintaining our slow or “normal” forward direction inside the bubble. If WE were to travel backwards literally, it also means growing younger, losing information and neural pathways, etc, and traverse the same spatial movement slower than c in the reverse direction. Not useful anymore than the old slow rewind function on the VCR. The practical case of jumping to a backward time spot in our bubble means we arrive were space was AT THAT TIME, meaning, no where near the dang Earth.
ChatGPT does a pretty good job, scouring the web and other sources for it’s answers:
Tell me about the soleares flamenco song.
quote:
Soleares is one of the most fundamental and iconic forms of traditional flamenco song, or "cante," in the flamenco genre. It's characterized by its deep emotional expression and complex rhythmic patterns. Here's more about the soleares flamenco song:
1. Structure: Soleares typically follow a 12-beat rhythmic pattern, which is divided into three groups of four beats each. The accent is often on the 12th beat, creating a distinctive, syncopated rhythm.
2. Lyrics: Soleares lyrics often delve into themes of love, longing, and pain. They are known for their deep emotional intensity, and the singer (cantaor or cantaora) delivers them with passion and authenticity.
3. Melody: The melody in soleares can be haunting and mournful, reflecting the deep emotions conveyed in the lyrics. The singer's melodic lines often feature melismas, which are elaborate vocal ornaments.
4. Versatility: Soleares can be sung alone or as part of a larger suite of flamenco forms, such as seguiriyas or bulerías. It serves as a foundation for various other palos (flamenco forms) and can be combined with different rhythms and variations.
5. Guitar Accompaniment: The soleares is commonly accompanied by the flamenco guitar, which plays intricate and expressive melodies and chords to complement the singer's performance. The guitarist often uses techniques like rasgueado and picado to add depth to the music.
6. Dance: Soleares can also be performed as a dance, with the dancer (bailaor or bailaora) interpreting the song's emotional content through their movements. Soleares is often included in the repertoire of professional flamenco dancers.
7. Regional Variations: Different regions of Spain have their own variations of soleares, each with its own unique characteristics in terms of lyrics, melody, and rhythm. Some well-known variations include Soleá de Alcalá, Soleá de Triana, and Soleá por Bulerías.
8. Historical Significance: Soleares has a long history in the evolution of flamenco and is considered one of the foundational forms of the genre. It has been influential in shaping other palos and remains a critical component of traditional flamenco performances.
In summary, soleares is a deeply emotional and significant form of flamenco song, known for its complex rhythms, passionate lyrics, and its ability to convey intense feelings of love and sorrow. It continues to be an integral part of the flamenco repertoire, cherished for its authenticity and expressive power.
I use it a lot in my line of business (programming) and my wife sells art derived and inspired by AI. It certainly exists and very useful. It has been very successful in analysing radiographic and MRI images for more accurate and faster diagnoses (which I originally studied at university in London). Machine learning is a subset using neural networks and deep pattern analysis and is very interesting to me as I develop cloud-based IoT services.
Example of my wife’s AI work:
Images are resized automatically to a maximum width of 800px
Ok, we understand that toward the future is seemingly possible because our forward direction of time slows relative to the background at high accelerations. But it requires a “bubble” of isolation, another “universe” which separates us and our time experience from the outside universe, such that our clocks can tick at different rates.
Travel fast in Earth orbit to experience time dilation through velocity. GPS satellite clocks tick at a different rate than on Earth and need a calibration to offset.
ORIGINAL: Richard Jernigan So AI exists, and is widely employed practically, but the word "intelligence" in the phrase "artificial intelligence" doesn't mean the same thing as when it applies to a human, an ape or a raven.
The implied premise that this is fine is what I was talking about in my previous comment. It's not fine, it's arbitrary and cognitively confusing at best. It also isn't always a carefully re-contextualized specialized usage. In many instances, and probably the original instances, its usage reflects ignorance about what intelligence is.
Concepts (the ideas behind our words) are only useful as tools for thinking about the real world insofar as they group things in the real world that are relevantly the same. We use concepts for thinking because we are limited in the number of things we can hold in immediate awareness. Concepts vastly increase our scope, but only if they are used according to their purpose and our nature.
One of the ground rules implied by the above is that if you arbitrarily add a new kind of thing to your concept, something that is not relevantly the same as the rest of what is grouped in the concept, you break the concept--you convert it from a cognitive tool into a kind of cognitive cancer that eats away the validity of any thinking you do using the concept.
The current confusion about AI, including among people actively involved in developing it, is a perfect example.
Artificial Intelligence (AI) possesses a form of intelligence, but it's important to understand that it's different from human intelligence. AI is designed to perform specific tasks and make decisions based on data and algorithms. Its "intelligence" is more accurately described as machine intelligence, which includes the following key characteristics:
1. **Learning**: AI systems can learn from data and improve their performance over time through techniques like machine learning and deep learning.
2. **Problem Solving**: AI can solve complex problems and make decisions by analyzing vast amounts of data quickly, which humans might find challenging or time-consuming.
3. **Automation**: AI can automate repetitive tasks, making processes more efficient and reducing the need for human intervention.
4. **Pattern Recognition**: AI excels at recognizing patterns and trends in data, which can be valuable in various applications, such as image recognition and natural language processing.
5. **Scalability**: AI systems can handle large volumes of data and tasks simultaneously, which is beyond human capacity.
However, AI lacks some aspects of human intelligence, such as common-sense reasoning, understanding context, and the ability to handle complex, ambiguous situations with the same flexibility as humans. AI operates based on predefined algorithms and data, and it doesn't possess consciousness, emotions, or subjective experiences.
In summary, AI exhibits a form of intelligence tailored to specific tasks and applications, but it's distinct from human intelligence and lacks certain human-like cognitive abilities.
Concepts (the ideas behind our words) are only useful as tools for thinking about the real world insofar as they group things in the real world that are relevantly the same. We use concepts for thinking because we are limited in the number of things we can hold in immediate awareness. Concepts vastly increase our scope, but only if they are used according to their purpose and our nature.
One of the ground rules implied by the above is that if you arbitrarily add a new kind of thing to your concept, something that is not relevantly the same as the rest of what is grouped in the concept, you break the concept--you convert it from a cognitive tool into a kind of cognitive cancer that eats away the validity of any thinking you do using the concept.
Artificial Intelligence (AI) possesses a form of intelligence, but it's important to understand that it's different from human intelligence. AI is designed to perform specific tasks and make decisions based on data and algorithms. Its "intelligence" is more accurately described as machine intelligence, which includes the following key characteristics:
1. **Learning**: AI systems can learn from data and improve their performance over time through techniques like machine learning and deep learning.
2. **Problem Solving**: AI can solve complex problems and make decisions by analyzing vast amounts of data quickly, which humans might find challenging or time-consuming.
3. **Automation**: AI can automate repetitive tasks, making processes more efficient and reducing the need for human intervention.
4. **Pattern Recognition**: AI excels at recognizing patterns and trends in data, which can be valuable in various applications, such as image recognition and natural language processing.
5. **Scalability**: AI systems can handle large volumes of data and tasks simultaneously, which is beyond human capacity.
However, AI lacks some aspects of human intelligence, such as common-sense reasoning, understanding context, and the ability to handle complex, ambiguous situations with the same flexibility as humans. AI operates based on predefined algorithms and data, and it doesn't possess consciousness, emotions, or subjective experiences.
In summary, AI exhibits a form of intelligence tailored to specific tasks and applications, but it's distinct from human intelligence and lacks certain human-like cognitive abilities.
The fact AI knows that it is not truly intelligent, is the most intelligent thing it has ever spit out, therefore showing it is intelligent. But we can be honest and admit it got that info from some human. About solea, only number 8 I agree with. AI, even as a simple machine filter, should be able to parse out all the nonsense and define the correct formal structure, melodies, rhythm patterns, etc. The fact it has not yet means it is simply regurgitating typical misconceived talking points of aficionados. Very human like.
AI, even as a simple machine filter, should be able to parse out all the nonsense and define the correct formal structure, melodies, rhythm patterns, etc. The fact it has not yet means it is simply regurgitating typical misconceived talking points of aficionados.
In other words, AI is subject to the old law of "Garbage In, Garbage Out." In any system, the quality of output is determined by the quality of the input. If that continues to hold true for AI, in my opinion you cannot call it truly "intelligent." It is just a more sophisticated "system" lacking the ability to conceptualize on its own, via "thought experiments" for example.
Bill
_____________________________
And the end of the fight is a tombstone white, With the name of the late deceased, And the epitaph drear, "A fool lies here, Who tried to hustle the East."
ORIGINAL: Richard Jernigan So AI exists, and is widely employed practically, but the word "intelligence" in the phrase "artificial intelligence" doesn't mean the same thing as when it applies to a human, an ape or a raven.
One of the ground rules implied by the above is that if you arbitrarily add a new kind of thing to your concept, something that is not relevantly the same as the rest of what is grouped in the concept, you break the concept--you convert it from a cognitive tool into a kind of cognitive cancer that eats away the validity of any thinking you do using the concept.
The current confusion about AI, including among people actively involved in developing it, is a perfect example.
Trained as a mathematician by one of the greats (R.L. Moore) and a student of physics by inclination, I am accustomed to naming novel and relatively distinct concepts by words already in the vocabulary, imparting to the words a new meaning. Writers of dictionaries have long recognized that a word may have more than one meaning, listing them with arabic numerals. As long as one distinguishes the neologism from the colloquial, confusion can be avoided.
Examples include the words "connected" and "continuous," both employed in novel and precise ways as core concepts in topology and analysis. The words "regular" and "normal" are employed in so many different senses in mathematics that one must be competent in specialized disciplines to have any idea what they may mean.
In physics, "momentum" and "energy" are key concepts, precisely defined in terms of a body's mass and velocity. Yet dictionaries consistently give imprecise definitions of the terms as they may be used in everyday speech.
This habit of mathematicians and physicists has been common at least since the 17th century, with little resulting error or confusion. But I will concede that it poses a danger for the non-scientist, including journalists who may report on science or technology. A distinctive term for AI might be useful, or even needed to avoid this kind of error.
But the founders of the discipline were thoughtful enough to label it artificial intelligence, not some possibly more confusing term like ''computer intelligence."
Here's an application of AI that is rather striking.
It is implemented entirely through machine learning, text and voice generation.
A machine or machines mimicking the distinctive voices of Werner Herzog and Zlavoj Zizek carry on a highly credible ersatz conversation between these two well known figures.
It seems to me that a naive person's theory of mind would be fairly likely to attribute "intelligence"--in the sense originally employed--to whatever is carrying on the "conversation."
the instigator of the "conversation" says he implemented it as a warning, and gives some details of how he went about it, using readily available apps.
If I start from the definition of artificial intelligence as a non-biological entity that posses the ability to solve problems and is able to express the solutions to those problems, then yes obviously AI exists. I guess there's no question about it, and practically AI as a field of study has always been about that. And though I don't usually like having to deal with re-contextualized concepts, yes AI happens to fall into that category. A better term would probably be something like Statistical Learning, or whatever.
I guess then that the philosophical aspect of AI, together with its connection and relationship with biological forms of intelligence, is much more interesting and open for discussion. Add in consciousness and things get much more complex and unknown. Like, animals are both conscious and intelligent, but arguably there exist non-conscious biological entities which are also intelligent, such as some plants. AI solves problem without consciousness, but how far can it go? What is consciousness anyway? How about Artificial Consciousness?
Another thing is how AI models work. ChatGPT only predicts the word that is most likely to appear next in a sentence, but the results it achieves with that simple behavior are astounding, practically no one expected such good results. It will (almost certainly :)) not going to take over the world, but it's still pretty impressive for what it is under the hood. Same fort the human brain way anyway, just a huge bunch of gray matter getting electric shocks all the time :)
Anyway, it was fun thinking about and writing all this, though it requires more thought and everything but it's too late and gotta go to work tomorrow.
If I start from the definition of artificial intelligence as a non-biological entity that posses the ability to solve problems and is able to express the solutions to those problems,
This definition captures any calculator, never mind a computer, but we don't call those algorithm-executing tools 'AI' because that would be laughable. Calling the dumb (but fast) pattern-matching algorithms 'intelligence' is no different; they are just much better obfuscated with relentless marketing.