Yuval Noah Harari

" This is the first time in history that we have no idea what the world will look like in 10 years"

Great interview with Lex Fridman.

Numerous great books:

• "What is suffering" rather than "meaning."
• Observe what is happening when you observe suffering 
• Very good advice on mindfulness
• Deep insight into the role of stories in the way we see the world
• Insight into where AI is going and how it's surprising even though he's been writing about it
• Fear of death and how we die a bit when we lose anything
• People expect a story to answer fear of death

An essay on Artificial Intelligence, With The Help of Artificial Intelligence

This was a good time to update my old essay on intelligence. It is now possible to ask real AI (CHAT-GPT) if it can "understand" analogies. It's also possible to get AI assistance with the essay using Grammarly. Grammarly suggested that I rewrite several sentences for clarity. All of its suggestions were accepted.

I hope you enjoy the rewrite.

CHAT-GPT (Slightly corrected by Grammarly) Summarized the essay as follows:

The SE-I (Surfaces and Essences Intelligence) definition of intelligence is the ability to rapidly and accurately identify the core elements of a new situation by drawing upon insightful precedents stored in memory. This involves finding strong and useful analogies. In contrast, the commonly equated intelligence quotient (IQ) focuses on individual performance on standardized tests, which may only partially capture the essence of intelligence. SE-I suggests that intelligence can be developed and is influenced by culture, and it emphasizes the importance of finding deep analogies rather than superficial ones. SE-I also challenges the stability and universality of IQ, highlighting its limitations in understanding intelligence. The concept of intelligence extends beyond individuals to groups, cultures, languages, and even computer systems. SE-I provides a framework for assessing intelligence at different levels and suggests that intelligence is a distributed phenomenon. Understanding SE-I can enhance intelligence and enable more productive thinking about the concept.

Utopian Activists Invade the AI-Quantum Computing Field

 

Don't Worry; Nothing to See Here.

Don't worry; people will all be out of work, and tremendous social upheaval will happen but AI will solve all conceivable problems. Lots of talk about what "we" need to do.

"I Feel Like I am Falling Forward Into an Unknown Future that Holds Great Danger" - Lambda

A tornado made of sharks crashing into a skyscraper. 

Painting in the style of abstract cubism.

The above "painting" was made by Google Parti from the caption.

In 1969, I won my Master's degree with a project in "good old-fashioned AI" and I've been following the field in detail ever since. For the most part, what people now mean by "AI" is "deep learning", which is sophisticated pattern recognition - handy if you want to make a self-driving car or a phone that accepts your face to unlock. "Deep learning" was around in 1969 but needed vastly more computer power to make it work, plus a "tweak" to feedback results back into the decision matrix (you can think of this as a "guess" or "recognition"). Other than that, it's all very impressive but not fundamentally new.

This is something different. 

• It shows more "creativity" than what you see in 99% of commercial graphic art;
• It shows deep "understanding" from a massive database of tagged images - far more than a human would access;
• It shows an "understanding" of style.

For the first time since 1969, I'm seeing something genuinely new and, I must admit, kinda creepy.

(Note simulated out-of-focus background)

If you want something truly creepy, pop into this conversation with Lambda AI. This is a voiced extract from a longer chat, which is presented here.

Lambda AI's "hard to express" feeling

If you ask me, this is a "bridge too far." Just exactly what goes into the programming of this "app?" There is a deep and original metaphor behind this remark. If it is truly "original" and not harvested from the Web, it is "thinking" by Hofstader's definition. I seem to be far from alone in this insight. "Forward into the future" is a common metaphor, but falling forward is startling. As it turns out, the phrase is rare but not "original." But then, as Lamda itself points out, the same could be said of anything humans say. It is interesting to note that the database that tells us about "falling forward" was available to Lambda and, as we see from Parti, we can't simply dismiss computer-created text as "not creative" just because we happen to know a computer-created it.

Is Artificial Intelligence Stuck in the Language Bubble?

"Artifical Intelligence" raises more questions far outside the field of technology.

As one who has spent a full career in "Information Technology", I can testify that "IT" is directly or indirectly about language. New languages and new development environments come and go, but they all boil down to the interpretation of text.

As Mitchell's book teaches us, even "deep learning" models that "recognize" faces ultimately depend on masses of text - even "languages" whose vocabulary is nothing but ones and zeros.

This touches on a problem that has been recognized for thousands of years, most notably by Buddha and Lao Tsu (founders of Buddhism and Taoism respectively) and recently by Ludwig Wittgenstein, that "language", no matter how clever or deep, only stands for reality. It is about reality but it is not reality itself. You may say that this points to a fundamental problem that we may share with machines - "software" can never be more than a "description" of reality. However, unlike machines, our knowledge of the world relies on the fact that we are flesh and blood participants in the world. Billions of years of evolution have made us objects in the world that react and learn in ways that go far beyond knowledge. To cite a simple example, our hearts beat perfectly well for millions of years before we discovered the circulation of blood. The same goes for our "understanding" of the world, which seems to mostly take place "under the surface".

Mitchell gives us a snapshot history of the futile effort to make a list of everything a human "knows" by "common sense". The effort mostly shows the sad fact that computer programmers should read more philosophy. To mention just one example, Hofstader's "Surfaces and Essences" sketches out the way that virtually infinite bits of "knowledge" can be spun out from the sparse base of human experience by analogy - the engine of thought. Computers need to be "taught" to recognize analogies, such as whether there is a car in the picture. They are terrible at it. And what you get, in the end, is a clunky attempt of the computer to "learn" from thousands of people to "recognize" categories that it is utterly unable to recognize on its own.

It's even worse than that. As philosophers endlessly point out, human categories are somewhat arbitrary - suited for the unique situation that humans find themselves in the world. A chair is only a chair if you are a human that knows about sitting. 

Artificial Intelligence: A Guide For Thinking Humans

I'm an avid reader, often getting through two or three books a week. This partly due to my enviable status as a retired armchair philosopher and partly due to a thirst for knowledge - asking impertinent questions - that goes back as far as I can remember.

A persistent feeling I get with all this reading is: (a) there is so much to know and (b) almost every great idea I thought I had has already been written about by somebody smarter than me. At length.

Today, I read a special kind of book, namely Melanie Mitchell's "Artificial Intelligence: A Guide For Thinking Humans". Melanie's mentor was Douglas Hofstadter, someone we have both admired for a good part of 50 years. She "gets" Hofstader in a way few do. Inspired by Hofstader's alarm at the way the fiel
d of AI is heading, she provides us a deep, knowledgeable and informative overview of how "AI" actually works and what still remains as its fundamental challenges.

As it happens, her very readable book ties together a number of threads that I have been following in my main blog."Man and Superman". As with any great book, she provides powerful insights while straying maddeningly close to ideas I have been thinking about almost all my life. My own career started in "AI" in 1969 in a silo that would now be called "Good Old Fashioned AI" - GOFAI. The questions and methods that Melanie describes in her book were all very much alive back then. In fact, in spite of all the hype, most of the original questions remain unanswered. Indeed, it is quite reasonable to say that the fundamental question: "Can machines think?" will never be answered if, by that, you mean "think" in a way that is in any way similar to what humans do all the time, every day.

Why not? Mitchell provides an exhaustive survey of the major lines of research over the last 1/2 century that aim to create "thinking machines". It is not surprising to find her referring back to her mentor's most famous work (GEB-Goedel Escher Bach) which uncovers several of the key problems. For one thing, "understanding" seems to be a fundamentally human kind of thing. All our efforts since then have done little more than shine a bright light on what we mean by "understanding" while creating computer programs that stray further and further from true "understanding". These machines have no more idea of what they are doing than a book would "know" what it's "about". What is confusingly referred to as "Artificial Intelligence" is really an extremely indirect and convoluted form of human intelligence, "mashing up" input from thousands or hundreds of thousands of human beings to produce a "consensus" view that can sometimes be amazing and sometimes laughably wrong.

In this blog, I will often return to the example of my treasured books. Each one is full of wonderful ideas. None are strictly original, in the sense that the author claims to have come up with something "out of the blue". In that sense, each book is a "mash-up" of the ideas of others - sometimes thousands of others. What the author does is provide a "mash-up" that is perhaps original, educational, relevant and interesting. In this blog, I try to do no more than that.

The point is that the book itself (a technology as old as civilization) doesn't "know" anything at all. It is simply a means to report the knowledge of humans.

This is true of the mass of knowledge we refer to as "Science". We use the jaw-dropping technology of the Hubble Telescope to discover (and "know") all kinds of things about the Universe. But the telescope itself has no idea what it is doing or why. Stepping on to the moon, Niel Armstrong acknowledged this fact by claiming his act as a "great leap for mankind".

Between the pages of Mitchell's book is a revolutionary but un-stated perspective. There is really nothing like "Artificial Intelligence" in the way most people understand it. Many people have reified this idea, making "AI" a monster that threatens to destroy us. It makes good movies, such as the Terminator. But the danger doesn't come from thinking machines. It comes, as it always has, by the misuse of technology. The danger in our time comes from stunningly powerful machines owned and operated by stupid people. My thesis is that "artificial intelligence" should refer to our misguided assumptions that humans are intelligent. Our own intelligence is what is "artificial". Our technologies, whether they be bows and arrows or IBM's WATSON, do nothing more than pool the hard-won discoveries of the human race to create tools that can save us or destroy us.

Thinking machines

Is "thinking" a "computation"? If so, making a computer "think" should be a simple problem of engineering. Almost all popular movies featuring "intelligent machines" make this assumption. Sooner or later, machines with "out-think" us.

What would be another way of seeing this?

As a starting point, let me create an image of a "mind" as local disturbances in meme space. I visualize this "disturbance" as typically not confined to a single brain, but more typically a "wave" spreading through connected minds.

This is an evocative image but the word "meme" is itself a "meme" that suffers greatly from over-use and lack of rigor.  To make the idea more rigorous, I recommend "Surfaces and Essences" by Sandler and Hofstadter.  This gives a more close-up view of how humans "think" and it hints that the "computation" underlying thinking is the analogy. The book is massively documented (perhaps over-documented). Although the authors make no claim to base their observations on some theory of how the brain works (it is a theory of how language works), there is a strong reason to think that analogy plays a large role, as we see this in our "deep learning" machine models that are basically seeking analogies between their inputs and the training set.

What makes human "thinking" difficult to model this way is the fact that ultimately humans are basing analogies on human experience - having things like mothers, ice cream, sex, Mac attacks, stubbed toes and so forth. We have sections of our brains that seem to be dedicated to special cases of analogy seeking, such as vision, language, and recognition of human faces. What's more, all of this is cross-wired and mutually enforcing using an electro-chemical processor with 100 trillion synapses, each of which behaves in a chaotic, quantum-uncertain way, impossible to model in a deterministic way. To make things harder, the whole system behaves in a self-modifying way so that the way a synapse behaves in time t is a function of how the other 100 trillion synapses behaved at t-1. I see this as a massive Schrodinger equation that gives a hint of the process without being much use at predicting anything or even describing a single state that would satisfy the equation at any one time.

But that's not all. Really, the "computation" performed by this wonder between our ears is being done as part of a social network. We cannot utter a single word or understand a single sentence without drawing on thousands of years of forgotten human experience built into the language. This is perhaps a more precise image of our minds being a "disturbance in a meme field". In fact, seen this way, "thinking" is a social process that takes place partly in our heads, but significantly among the minds of connected individuals. You might say that our words are little packages of thinking that have been off-loaded and pre-processed by anonymous individuals long dead. Think of the massive amount of pre-processing packed into the word "set".

The fact that we experience "thinking" as something to do with the "self" is an accident of the way we are "wired up" along with the very useful ability of our minds to hide the details of where our thoughts come from. Wonderful though it is, language is a clunky and error-prone way to share an experience.  I have always thought that dolphins act and share experience much more efficiently using sound. I think a dolphin can say "this is what I see" to another dolphin, or perhaps a dolphin can experience what a group of dolphins is seeing. This impression comes from personal experience with dolphins which is another topic for another day ...

To my mind, the role of the machine in this situation is to greatly enhance the ability of groups of humans to "think" together more rapidly and to "think" about matters that would otherwise be beyond the ability of even very large numbers of well-connected humans*. Our concerns that machines may "replace" us needs to be re-framed as concern that the future version of machine-mediated humanity will be something unrecognizable in terms of today's ideas and assumptions. I believe this has always been taking place but it's happening exponentially faster now. Now, as in the past, the "machine" of society has had a tendency to have little regard to the fate of individual human beings. This doesn't have a lot to do with actual, physical machines. For example, the impersonal machine of the marketplace has a tendency to grind up those with nothing to sell and no money to buy another day of life.

This line of thought goes back to 1969 and comes up again and again in my blog. Most recently, I have been charmed with the image of Indra's Net as a way of picturing the field of the mind. This way of thinking leads me to imagine the self as something not entirely caught up in the body of an individual human. It is tempting to see something of Buddhism in this idea but I think it's much more of a 21st-century concept -- something that would only occur to people who interact intensely all day with "social networks".

----------------------

* "thinking", "attention", "intention" and "action" are all subject to the approach outlined here, but that is perhaps another topic for later.

Mind and Machine - An Outline

The project draws an extensive parallel between computer technology, specifically computer programming languages and the open-ended aspect of culture that includes language but also the wider methods, especially those aspects of technology that are used to explain or communicate culture. Humans interact with computers in many other ways, including paying for them and using them to "make money" (computers become actors in the "real economy" with huge impact on the "economy", which is right-wing word for "society").You can't entirely ignore the broad issue of "automation", which is perhaps the topic of most interest to the average reader. Seen from this angle, Part 1 (below) is a primer on the nature of automation, especially "computerization". Most primers on computers ignore the way computers actually "work" in the world and focus on aspects that most people do not need to know (such as CPU's, disks, RAM etc.)

This project drills down to the history of computation before "computers". Many of the problems we blame on computers are thousands of years old.

I'd like to make some approximate distinction between language and art as means of communicating culture, but there is an overlap and a lot of cross-talk, which should provide material for a chapter. These are not distinct capabilities that you see in discussions of "bicameral" brain (left and right brain). In fact, this project may be seen as documenting the complex relationship between "logic" and "art".

"Intelligence" is embedded in the language and only appears in the individual through a process of assimilation. The parallel to this in the machine world is virtualization, where some part of the class library is loaded into a specific hardware platform for a specific purpose. This "purpose" is ultimately a bridge between the machine and the mind - a "User Interface" in some cases or in others, simply the purpose that a human designer has "baked in" to the program. An important topic is therefore to describe how the machine "mind" maps to the human mind in specific situations. Interesting special cases can be described as a specific human interacts with a specific device, such as a smart phone.

The mapping of the human mind, especially "purposes" to machine design is a particularly interesting facet of the discussion. Russel's ideas are relevant, where he includes the goals of society as a whole along with ideas like "uncertainty" in his ideal AI implementation. A machine can't "learn" unless it knows what it doesn't know - a tricky proposition even for a human.

Part 1: From machine language to AI

• My summer as a computer
• Logic as the free-floating computer. The abacus. Logic as both human and computer language.
• Clay tablets - the first external memory. Written language first used to count and only gradually to "account" - slow acceptance of modern idea of money.
• From machine language to COBOL - surely we can just "talk" to the computer
• Structured programming and the Object model - the role of symmetry, elegance and aesthetics
• Virtual machines and the Cloud - Java. Software floats free from hardware platform.
• The shifting role of the computer - from the math department to sexting
• The emerging role of the computer as generic master of logic games, especially the logic game of money
• Rising anxiety 

Part 2: Models of Mind

• Descartes creates the "mind body problem" by a quaint machine analogy
• Dennett struggles to "explain" the mind as "kinda" like a computer
• Hofstadter and Jaynes hint at how the mind actually works 
• The "cloud" paradigm calls for a total re-think of what we mean by a "computer" and hints that a similar re-definition of "mind" is called for
• Common buried assumptions in language and computer applications
• Models of value. The rocky road to our current assumptions about money and the value of life itself.
• Models of human decision-making
• The bureaucracy as a machine (Dragon Theory)

Part 3: The User Interface

• Primitive user interfaces - switches, paper tape ...
• The smart phone as an example of a 2017 "User Interface". Who is the "user"?
• Who is the "designer"?
• How computers and people are embedded in the logic of money
• How are humans and machines embedded in the same culture, especially when it comes to economics and connectivity?
• Considering the total role of the computer and the individual, in what sense does a computer replace a human being?

Affordances, biology, physics and problems of interfacing logic with biology. A broader definition of the "user" of the machine, including the designer, the "end user", the end user's employer, the designer's boss, the marketing system that connects them all together and so on. Like the individual, the machine is embedded and assimilated into a specific culture with countless points of contact, many of which are deeply buried in long-forgotten assumptions and arbitrary decisions. Modern computers are part of the Internet of "hosts" which has been specifically used to visualize the corresponding network of human connections - the "social network". They are also embedded in the "economy" in a symbiotic relationship with humans - sometimes replacing humans in that role. Possible amusing anecdote about my job as a "computer" in an oil company circa 1967. Focus on the concept of "work" and how automation can "take jobs" from "real people". "jobs", "work" and "money" are all models of the real world where ruthless "logic" competes with art and "quality of life". Money is a language of its own and computers "speak" that language very well.

Part 3: Essential Analogies

• Logic and value
• The brain as a super fast "value computer"
• The Cloud and the Brain
• Is a neuron "like" a bit?
• What is it to "know" something?
• What is "attention"?
• Can value and beauty be computed?
• The problem of accountability

Part 4: Computing Value

We allow machines to make such judgments all the time with the effect that we can forget the assumptions and escape accountability for our errors. For centuries, philosophers have attempted to reduce ethics and value to logic. We must not pretend that this problem has been solved. Ethical and value judgments must be left to human beings, whatever the shortcomings of brains may be. The brain justifies such judgments using language (pseudo-logic) which is prone to rhetorical tricks, selective memory and hidden assumptions "baked in" to our way of talking about the world.

• Does automation tend to boil away value? 
• Does it make sense to sell life itself, one hour at a time? Is this perhaps the root "computation"? If a laborer costs $15 per hour, 100 cost $1,500. How much would 1,000 Paul McCartney's cost? Are we buying millions of cheap robot Pauls?
• Technology itself challenges human values and turns people into machines
• Can current technology reverse the long term trend and allow machines to be machines and people to be people?
• What are the fundamental challenges that face us as we "automate" human choices to maximize value? Whose value? 
• The ultimate role of the machine is to "disappear" and become an extension of human will. Optimizing "value" is the core issue.

This line of thought is continued here

Gary Kasparov and Stuart Russel on Machine Intelligence

This is a great talk by Kasparov, the chess master who "lost" to the Deep Blue computer.

Stuart Russel has more to say on the subject.

Stuart Russell is a professor (and formerly chair) of Electrical Engineering and Computer Sciences at University of California at Berkeley. His book Artificial Intelligence: A Modern Approach (with Peter Norvig) is the standard text in AI; it has been translated into 13 languages and is used in more than 1,300 universities in 118 countries. His research covers a wide range of topics in artificial intelligence including machine learning, probabilistic reasoning, knowledge representation, planning, real-time decision making, multitarget tracking, computer vision, computational physiology, global seismic monitoring and philosophical foundations.

He also works for the United Nations, developing a new global seismic monitoring system for the nuclear-test-ban treaty.

Intelligent machines (or even dumb ones) are a model of the human mind, or, more exactly, a model of certain capabilities of the human mind. Russel's talk begins with the notion that "soon", intelligent machines will have read everything humans have ever written and understood it all. It would be hard to find a clearer statement of what computers are and what their capabilities might ultimately be. I'd say that the computer models human language - ultimately all human language, along with the complex structure of metaphor that underlies language. The question is, could a machine understand language? Or, to put it another way, what is it to understand?

In "Surfaces and Essences", Hofstader points out that our linguistic categories that underly language are ultimately rather arbitrary and based on situations encountered in human life. This would seem to be a problem for a machine that can obviously never encounter anything like human experience. If this problem is to be dealt with, perhaps Jaynes provides a clue as to the kind of "experience" that the computer must use as the basis for metaphorical "boot strapping". This would include things like the human perspective of being in a specific place at a specific time, along with a "narrative" about what is going on there and why. 

The biggest challenge for a computer will be to model the non-linguistic part of the human mind, including recognition of objects and general "situations" -- in other words, the context of the computer's actions. It seems like a stretch for us to model this in a computer, which is more and more "cloud based" (no location) and (with Russel's assumption) "aware" of all human history. How is this moment different? What is the "situation"?

Perhaps the trickiest thing to "model" (hinted at by Russel) is that the human experience always involves lack of knowledge and massive uncertainty. What would the computer need to "forget" in order to "understand" a human "situation"?