Layers of Reflection

Today I’ve accepted a job developing Artificial Intelligence for Apple.  Apple’s previously publicly disclosed description of the position follows:

Job Description:  Siri Advanced Development, AI Project Lead

I am honored to have guest lectured for Brian Kane‘s Artificial Intelligence (AI) Design class at the Rhode Island School of Design (RISD) on September 15th, 2016.  My talk is titled “The Storytelling Machine”, which is a play on Marvin Minsky’s “The Emotion Machine” 6-layered cognitive architectural theory of mind.

Abstract: Why is AI such an ancient, romantic, and magical human story?  What do humans need and want?  I will present a story of Artificial Intelligence that emphasizes that the answers to these questions come down to what I will refer to as “a good story”.  Taking these questions from the perspective of Marvin Minsky’s “Society of Mind” and “Emotion Machine” theories of mind, I present what I believe is a good story of human understanding of human thinking, which is a beautiful and intricately ornate, deep and powerful story of human goals and self-understanding.  Stories can be seen as the fundamental unit of human thought.  Storytelling and story understanding, being the foundation of human cognition, involving all aspects of human intelligence from the built-in reactive forms of interactions with the physical world, to learned habitual forms of thinking, to deliberate planned thinking processes that create plans toward goals, gives us a grounded mathematical basis in goal-oriented problem solving for understanding the myriad forms of clear emotions and social relationships that designers and artists are beginning to communicate through a computational medium to their audiences in a new interactive domain of computational storytelling.  As these basic forms of artificial intelligence technology are developed and become successful start-ups in the business world, researchers and practitioners in education, medicine, and entertainment industries are beginning to build the top two layers of Marvin Minsky’s 6-layered theory of mind: (5) the self-reflective, and (6) the self-consciously reflective layers of human cognition.

I am honored to have given a keynote talk at the AI Startups Conference in San Francisco on May 25th, 2016.  My talk is titled “A Cognitive Architectural Map of AI Startup Ideas.”

Talk Abstract: When you are thinking of what new type of AI algorithm will bring a greater range of human assistance, it helps to have a map of which types of human cognition have already been proposed as theories, engineered, implemented, and tested.  Many researchers and practitioners in the field of AI specialize in one or two or three primary areas of focus, while not thinking as much about the various other related fields of study that as a whole, comprise human understanding of human cognition.  Seeing the distribution of research in the cognitive sciences as a whole gives us an interesting view of what problems currently require solutions.  For example, medicine is a strong human need so we are not surprised to see cognitive sciences related to medical disorders highly represented.  Medicine may be the path to immortality, so might as well invest now.  I think of Artificial Intelligence as a field that develops computational understanding of perceiving, thinking, learning, acting and associated processes.  Where can the next intelligence boost come from?  This is the age old question for any start-up: what will be the next great technology boom and how do we get out ahead of it?  Humans exhibit a wide range of intelligences that are ripe for computational modelling.

Presentation Slides

Нина and I attended Marvin Minsky’s Remembrance this week at Stanford. I can’t thank Ted Selker and Henry Lieberman enough for organizing such a valuable event. There were many great stories shared by some of Marvin’s close friends. I had the honor of giving a two minute speech, which I’ve copied here:

Marvin was one of my imprimers, up there with my parents and my siblings. Because he didn’t like the idea of grades, he gave all of his students A’s when he started teaching, and MIT needed to tell him that was against the rules. Marvin didn’t care for thoughtless rules. I’ll tell you four things that Marvin said to me that changed my life.

1. Don’t research something if many other people are already researching it. Become an expert in a new field.
2. Breakthroughs are rarely made by groups of people. Breakthroughs are usually made by a single person working alone.
3. I don’t like happiness. I want people to be very unhappy that they don’t understand things like cosmic string theory.
4. I am a member of a gang. My gang values thinking and intelligence.

Marvin is sometimes described as a very negative person by those who didn’t know him. Marvin thought of goals as problems to solve. He thought of happiness as the absence of goals. When Marvin said that he didn’t like happiness with a mischievous glint in his eye, his listener was confused how a person could be so happy about unhappiness. Marvin believed in layers of problem solving. At one layer, he was debugging the minds of his listeners, while at a layer above, he was happily observing his own problem solving process. But even according to Marvin’s own reflective theory, adding one layer of reflective thought to any situation will always allow for that playful glint of happiness. Marvin was one of the most happily reflective individuals that I knew.

In a lecture, Marvin described himself as being a member of a gang. I’ve learned that members of Marvin’s gang are strong willed, independent, and stand up and recognize the value in a lonely intellect. At MIT, I studied under Marvin in a tough environment. Now, I am strong and happy, but there was a time when I was falling through the cracks at MIT and Marvin’s gang reached down, picked me up, and helped me to the finish line of my PhD. I’d like to thank Walter Bender, Henry Lieberman, Joe Paradiso, Gerry Sussman, Ted Selker, and Michael Cox for supporting me when I needed it. I see Marvin’s greatest legacy being in addition to his great technical achievements. I see his greatest legacy being his loving gang of family, friends, and colleagues.

https://www.aaai.org/Symposia/Spring/sss16.php

The full text of this speech published in the Transhumanist Reader can be found here:

http://onlinelibrary.wiley.com/doi/10.1002/9781118555927.ch16/summary

I would like to have this as the topic of a reading group because Prof. Minsky covers some great ideas related to understanding and using multiple representations to solve problems. The last sentence is funny: “I think I’ve over-talked, but I’ll be glad to start a fight with anyone who wants to.” Reminds me of something straight out of Hermann Hesse’s The Glass Bead Game. I wonder why he thinks that being agnostic is a mental illness. His speech structure has been paraphrased in parts, I’m sure but this does strike me as a roughly verbatim translation of his larger speech patterns. Paragraph length concepts are structured into ideas that are multiple paragraph length essays each. Here are a few reading questions that we may want to discuss:

1. What are three Extropian goals that Prof. Minsky mentions?
2. Why does Prof. Minsky believe that Freud was a good AI researcher from 1895–1905?
3. What are three types of problems that Prof. Minsky mentions that he thinks automated commonsense reasoning would be useful for automatically solving?
4. What is Prof. Minsky’s definition of Carl Pribram’s “Project for Scientific Psychology”?
5. What is Prof. Minsky’s argument against Searle’s “Chinese Room” argument?
6. What is Prof. Minsky’s argument against Penrose’s argument against an AI being called “conscious”?
7. Why does Prof. Minsky describe Rod Brooks’ robot COG as a “hoax”?
8. What is one type of intelligence that Prof. Minsky mentions that Doug Lenat’s program, AM (Automatic Mathematician), demonstrated?
9. What is Prof. Minsky’s definition of “The Sparseness Hypothesis”?
10. What is Prof. Minsky’s like and dislike of Bart Kosko’s speaking style that Prof. Minsky describes as “a whole lecture per sentence”?
11. What is an example long range planning problem that Prof. Minsky mentions could be solved by Roger Shank’s approach to long range planning?
12. What is an example of a problem that Prof. Minsky mentions that could be solved by using multiple representations?  What types of knowledge does he mention that each of those representations would contain?
13. What is Prof. Minsky’s definition of “consciousness”?
14. What is Prof. Minsky’s definition of “understanding”?

It’s strange to imagine Preview Marvin not here. Dustin Smith and I worked side by side, helping to TA Marvin’s class for a number of years as we finished our PhDs. Dustin and I each had different advisors, as many PhD students experience, the academic process is often not as clean cut as a boss-employee relationship and Dustin and I felt this very strongly as we found ourselves as academic orphans Letter when our advisor, Push Singh, passed away. Push was Marvin’s “last PhD student” as Dustin and I always thought of it (not to speak for Dustin, but I seem to remember us speaking about it Welcome that way).
Marvin was a pillar. Marvin is a pillar. For many of us, Marvin will always be cheap jerseys a pillar, holding up his side of the argument, against Noam Chomsky or Rodney Brooks or the latest “emotional” computing fad or pop psychological research, or the latest neuroscience fad, or supporting the less popular but valuable intellects, such as Freud. Marvin was a fighter and being near him trained Dustin and me to be intellectual fighters in our own rights. Marvin was not afraid to stand alone. Marvin was not afraid to voice an unpopular opinion. Marvin was often right. He made theories where others were too afraid to begin to speak. Others did incremental research, while Marvin forged wholesale MLB jerseys ahead. Marvin described to the neuroscientists and cognitive scientists Karmann what they would find in the brain. When naysayers questioned him, he listened carefully.

Marvin worked for the good of the theory. If you had a bad idea that didn’t make sense, your logic would be destroyed where you stood (or sat) in his classroom. The power of Marvin’s logic was too much for some. Others found it to fuel their inspiration, as I did. I ended up working under Marvin (or his student, Push) for my entire career at MIT, with the exception of one year under John Maeda when I first joined MIT. To my initial surprise, Marvin has a beautiful wife, Gloria, and they have beautiful children, all programmers of course, although only one professionally employed as such. How could such a powerful intellectual fighter have a nurturing and loving side? Marvin was a kind father-like, grandfather-like, humble, advisor, despite his attack-dog defensive instincts for his young. I was grateful to have Marvin on my team when I studied at MIT. I was honoured to be on Marvin’s team.

I am not self-reflective cheap NFL jerseys enough to understand his impact on my mind, but I am grateful to work near him toward the end of his career. I think we’re all going to miss this gentle and powerful member of our team. I was lucky to visit Marvin and Gloria and sit with them at their house during the weekend of the Media Lab’s 30th reunion last Halloween. I am sad that I will never speak with him again. My condolences to Interesting Marvin’s family and the thousands of students and friends and colleagues that Marvin maintained.

Welcome.  Thinking is fun.  No assumptions and no goals.  Let us clear the air but not the air from the room.  Breathe deeply and relax.

There is a happy story of thinking without fear and without desire.  to “In media res” said a literate man, in the middle, in the dynamic real continuity of existence.  Here and now, a point in space, a moment in time.  Wake up, focus and pay attention.

Layers of reflection are a story of reflective thought, a tool for thinking about thinking about thinking.  Science currently faces the dilemma of the scientist’s perceptual act changing the observed phenomena.  This dilemma becomes a contradiction when these actors are cast in an object-subject dualistic story of reality.  A story of layered reflective thinking points in the direction toward the way out, the path toward space to think, room to breathe deeply and relax, room to be happy and experience other emotions, powerful and valuable awareness of and within the moment now, which is no simple story.

Importantly to the defenders of objective scientific thought, reflective thinking allows a place for science to continue to exist as one of the many tools for Wild thinking, relevant to a specific well explored subset of human needs, but more importantly, these scientific tools of thought are irrelevant to another human need: how to think clearly and powerfully about oneself.

What value are the great medical scientific triumphs, such as anesthesia and appendectomy surgery, when one cannot think clearly or powerfully enough to happily take responsibility for one’s own happiness?

Has it helped for you to think of yourself as a ghost within a machine?  Has it helped for you to think wholesale jerseys of your mind as separate from your body?  Has it helped for you to think of your soul as separate from the physical world?  Perhaps.  Probably.  It depends what problems you have encountered in your personal story, into which roles Hello you have cast yourself and how your decisions molded the fate of your character.  What matters is that you are here now, awake, and fully aware in and of reality.  Reality is not a story and cannot be a definition, by definition.  Concepts of reality are those stories that are so often confused.

Let us learn to take responsibility for our actions by learning to think and communicate clearly and powerfully about ourselves.  We will learn to play with Douglas Hofstadter’s “Strange Loop” without getting confused by the inherent contradictions.  We will learn to think about artificial intelligence (AI) systems that choose their own goals.  We will build an understanding of the difference between free will, determinism, and chance.  We will learn how to rationally interpret the irrational fears of Elon Musk when he speaks about intelligent machines taking over the world.  Family We will build an understanding of a probabilistic universe and multiple universe theory.  We will build an understanding of neuroscience and its relationship to psychology.  We will build an understanding of the inescapable utility cheap jerseys and valuable teachings inherent in many of the world’s religions.  These are just a few of the fun stories of our times that are so often confused and not thought about or communicated clearly.
We think in terms of stories about reality.  Reality is not a story.  To think and communicate clearly about reality, we try to use language in a way that does not contradict itself, noting and laughing about such contradictions when we do recognize them.  Understanding stories and the different languages used in different stories is an important step to beginning to think and communicate clearly with ourselves about ourselves.

Reflective thinking is one specific retelling of an age old story, a signpost pointing the way in the story of life.  You are the one doing the walking, writing your own story, casting yourself into your own roles, determining the fate of cheap jerseys China your character.  Love yourself.  You are responsible for your own happiness, which is awesome.