On the Physics of Thought

 Closer to understanding of the thinking process.

ACM TECHNEWS

How the Brain Prepares to Think

By Texas Advanced Computing Center,  June 23, 2022

The University of Texas Southwestern Medical Center's Jose Rizo-Rey and colleagues used the Texas Advanced Computing Center's Frontera supercomputer to probe the physics of thought activation in the brain.

The researchers have generated all-atom molecular dynamics simulations to explore the nature of the primed state of synaptic vesicles, indicating specialized proteins are "spring-loaded" and awaiting calcium ions to induce fusion.

The models only simulate the first few microseconds of the fusion process, but Rizo-Rey posits that fusion should occur in that time.   Rizo-Rey said, "If I see how it's starting, the lipids starting to mix, then I'll ask for 5 million hours [the maximum time available] on Frontera" to record the spring-loaded proteins' trigger and the fusion/transmission process.

From Texas Advanced Computing Center  ...    

Said University of Texas professor Jose Rizo-Rey, "This country was very successful because of basic research. Translation is important, but if you don't have the basic science, you have nothing to translate." ... 

Comparing AI Reasoning with Human Thinking

Comparison is always useful.   To mimic or check useful contexts. 

Comparing AI Reasoning with Human Thinking

IEEE Spectrum, Charles Q. Choi, April 27, 2022

Researchers at the Massachusetts Institute of Technology (MIT) and IBM Research have developed a method for comparing the reasoning of artificial intelligence (AI) software with that of human thinking, in order to better understand the AI's decision-making. The Shared Interest technique compares saliency analyses of an AI decision with human-annotated databases. It classifies the AI's reasoning as one of eight patterns, ranging from the AI being completely distracted (making incorrect predictions and not aligning with human reasoning) to making correct predictions and being completely human-aligned. Said MIT's Angie Boggust, "Providing human users with tools to interrogate and understand their machine-learning models is crucial to ensuring machine-learning models can be safely deployed in the real world."  .... ' 

AI in a Jar, a Thought Experiment

Thoughtful,  with link to comments 

The AI in a Jar,    By TechTalks  April 15, 2022

The "brain in a jar" is a thought experiment of a disembodied human brain living in a jar of sustenance. The thought experiment explores human conceptions of reality, mind, and consciousness.

This article explores a metaphysical argument against artificial intelligence (AI) on the grounds that a disembodied AI, or a "brain" without a body, is incompatible with the nature of intelligence ... 

Illustration of a brain with gears inside a jar.

The brain in a jar asks whether thinking requires a thinker, but the possibility of AI primarily revolves around what is necessary to make a computer or program intelligent.

Full Article  

Book: Full Spectrum Thinking

Good thoughts by current and longtime colleague.    

CURRENT BOOK: Full-Spectrum Thinking: How to Escape Boxes in a Post-Categorical Future

Full-Spectrum Thinking:  How to Escape Boxes in a Post-Categorical Future    By Bob Johansen

Download Sample Chapter

Leading futurist Bob Johansen shows how a new way of thinking, enhanced by new technologies, will help leaders break free of limiting labels and see new gradients of possibility in a chaotic world.

The future will get even more perplexing over the next decade, and we are not ready. The dilemma is that we're restricted by rigid categorical thinking that freezes people and organizations in neatly defined boxes that often are inaccurate or obsolete. Categories lead us toward certainty but away from clarity, and categorical thinking moves us away from understanding the bigger picture. Sticking with this old way of thinking and seeing isn't just foolish, it's dangerous.  .... '

Danel Kahneman vs Deep Learning

From Linkedin, by Ajit Joakar

(You need Linkedin membership to see this introduction) 

Artificial Intelligence #18: Of Daniel Kahneman and Deep Learning

Published on August 24, 2021

Course Director: Artificial Intelligence: Cloud and Edge Implementations - University of Oxford

 Welcome to Artificial Intelligence #18

 This week, I am still on work / holiday in Germany

 In this newsletter, we cover an important topic based on the book thinking fast and slow

 The main thesis of Daniel Kahneman’s landmark book thinking fast and slow is the dichotomy between two modes of thought: "System 1" (fast, instinctive and emotional) and  "System 2" (slower, more deliberative, and more logical).

The ever insightful Kahneman also points out to system 1 v.s. system 2 as one of the challenges of deep learning in an interview with Lex Fridman

 To summarise Daniel Kahneman's view on deep learning.  What is happening in deep learning is more like a system one product. Deep learning matches patterns and anticipate what's going to happen so it's highly predictive

However, deep learning doesn't have the ability to reason, manage temporal causality and represent meaning

Deep learning has made rapid progress but there are many problems where we need ability for reasoning. These shortcomings are also pointed out by Gary Marcus and other critics of AI.These are important limitations to deep learning and AI today.

. Solving these challenges could be a trigger for deep learning to evolve  ... .'

From Computational to Generalized Thinking

Interesting view that I am not quite understanding.   Shall we think as computers, the way the exist today or should we break through from this approach?   Thought provoking, but very technical.

HCDA: From Computational Thinking to a Generalized Thinking Paradigm   By Yuhang Liu, Xian-He Sun, Yang Wang, Yungang Bao

Communications of the ACM, May 2021, Vol. 64 No. 5, Pages 66-75   10.1145/3418291

In 2006, Jeannette M. Wing45 proposed the concept of "computational thinking," which has produced significant worldwide impacts on the education, research, and development of computer science. After more than a decade, we reexamine computational thinking, and find that a more general-thinking paradigm is urgently needed to address new challenges.

A couple of recent commentaries12,41 regarding computational thinking attracted our interests and inspired us to reflect further. More than that, we want to summarize and generalize the rationale of our solutions, for instance, the Labeled von Neumann Architecture (LvNA)1,28 and the Layered Performance Matching (LPM) methodology.26,27

Nurtured by Moore's Law, the number of transistors available on a single chip increases exponentially. Meanwhile, due to architectural innovations, transistors are organized more effectively and utilized more vigorously. As a result of the combined efforts, computers have witnessed a significant performance advancement during their 70-year history. However, the new age, characterized by the slowdown of Moore's Law and Dennard scaling,44 and by the rise of big data applications, brings serious challenges that computer scientists must face.

The scaling of on-chip transistors impacts microprocessor performance significantly. However, further improvements to transistor density and power become more difficult due to the limits of semiconductor physics.44 As a result, architectural innovations become increasingly crucial for performance breakthroughs, and the epoch we are entering is "a new golden age for computer architecture."14

The rise of big data has caused an unprecedented shift, where the memory system, rather than the computational core, plays a more vital role. Accordingly, the memory access limitation described by Sun-Ni's Law38 is becoming a performance killer for many applications. Thus, data-centric innovations of computer system design are urgently needed to address the issues of data storage and access. Both the emergence of big data and the slowdown of Moore's Law have changed the landscape of computer systems and require us to examine past solutions to pave a new path for future innovations and for addressing new challenges.  ... " 

HCDA: ... . A Framework of the Four Thinking Patterns ("H" represents "Historical thinking," "A" represents "Architectural thinking," "D" represents "Data-centric thinking," and "C" represents "Computational thinking"). ... "

AI Improves the Way Humans Think

After reading this a few times I wondered.   Because we can use an AI to explore ideas more quickly? Things that wold have taken more time to examine?

Mind meld: Artificial intelligence is improving the way humans think

When AIs and humans work together they discover superior solutions to the world’s problems that would elude either working alone. Together, they will change the very process of thinking

By Douglas Heaven

IKE other human champions facing a machine opponent, Grzegorz “MaNa” Komincz rated his chances. “A realistic goal would be 4-1 in my favour,” he told an interviewer before the match.

One of the world’s best players of video game StarCraft II, Komincz was at the height of a successful esports career. Artificial intelligence company DeepMind invited him to face its latest AI, a StarCraft II-playing bot called AlphaStar, on 19 December 2018.

Komincz was expected to be a tough opponent. He wasn’t. After being thrashed 5-0, he was less cocky. “I wasn’t expecting the AI to be that good,” he said. “I felt like I was learning something.”

It was just the latest in a series of unexpected victories for machines that stretch back to chess champion Garry Kasparov’s 1997 defeat by IBM’s Deep Blue. In 2017, another of DeepMind’s AIs, AlphaGo Master, beat the world number one Go player a decade before most researchers predicted it would be possible. The company’s AIs then mastered chess and StarCraft – a game played with dozens of different pieces with hundreds of moves a minute.

But this isn’t just a case of humans being humbled by superhuman AI. The real story is that each win gives us a glimpse of how AIs will make us superhuman too. That’s because thinking is set to become a double act. Working together, humans and AIs will bounce ideas back and forth, each guiding the other to better solutions than would be possible alone.  .... '

(Requires subscription, sorry,  but I liked the premise)

Read more: https://www.newscientist.com/article/mg24332440-700-mind-meld-artificial-intelligence-is-improving-the-way-humans-think/#ixzz5yOTMjYzB

Machines Thinking are like Translation

Intriguing idea. Includes video.

A New Approach to Understanding How Machines Think
From Quanta Magazine   Link to full article. 

Been Kim and colleagues at Google Brain developed a system she calls a translator for humans that permits them to ask questions of an artificial intelligence. 

Google Brain research scientist Been Kim is developing a way to ask a machine learning system how much a specific, high-level concept went into its decision-making process.

If a doctor told that you needed surgery, you would want to know why — and you'd expect the explanation to make sense to you, even if you'd never gone to medical school. Been Kim, a research scientist at Google Brain, believes that we should expect nothing less from artificial intelligence. As a specialist in "interpretable" machine learning, she wants to build AI software that can explain itself to anyone.

Since its ascendance roughly a decade ago, the neural-network technology behind artificial intelligence has transformed everything from email to drug discovery with its increasingly powerful ability to learn from and identify patterns in data. But that power has come with an uncanny caveat: The very complexity that lets modern deep-learning networks successfully teach themselves how to drive cars and spot insurance fraud also makes their inner workings nearly impossible to make sense of, even by AI experts. If a neural network is trained to identify patients at risk for conditions like liver cancer and schizophrenia — as a system called "Deep Patient" was in 2015, at Mount Sinai Hospital in New York — there's no way to discern exactly which features in the data the network is paying attention to. That "knowledge" is smeared across many layers of artificial neurons, each with hundreds or thousands of connections.  ... "

Long Now Meets Land Art

Some long ago encounters and conversations with folks that founded the Long Now Foundation reminded me again of Art in desert spaces.   See also their recent advances in clock technology.   Long term thinking is very rare, and should be appreciated.

The Long Now Foundation
Fostering long-term thinking through projects like the 10,000 year clock; Rosetta Disk; Revive & Restore; Seminars, videos, podcast ... 

James Turrell, Earthworks, and Monuments of Deep Time    A History of Land Art in the American West .... "

Thinking Machines

Like the broadening of what we sometimes call AI.  I like the idea of 'Cognitive' because it promises less.  Would also like to add other classifiers, say 'Process Intelligence'?

Thinking Machines Going Mainstream 
in SIGNAL Magazine  By George I. Seffers

Experts predict cognitive computing will eventually become normalized as a routine behavioral component in any newer systems. "It will be an expectation of the users that this assistive, interactive, iterative role that it plays within decision making becomes the norm," says Cognitive Computing Consortium co-founder Sue Feldman. She believes more interactive technology will be "able to return answers or graphs or whatever is necessary in an iterative manner with the people who are using it," while also being contextual. Meanwhile, consortium co-founder Hadley Reynolds expects an acceleration in the blurring of boundaries between technological devices and other objects, including apparel and everyday appliances, to the point where they vanish completely. Feldman and Reynolds agree big data will continue to fuel advances in cognitive computing, with a major possibility of a new profession stemming from cognitive computing and artificial intelligence. The consortium co-founders alternately cite ethics and trust as the most pressing remaining challenges for cognitive computing. ... " 

Learning by Thinking

Can we make machines work this way?   In a sense we do. we train and re-train.  And test against reality.  But is the semantic language or architecture correct to make this work efficiently?  Not yet.

In the Edge: 

Learning By Thinking
A Conversation With Tania Lombrozo 

Sometimes you think you understand something, and when you try to explain it to somebody else, you realize that maybe you gained some new insight that you didn't have before. Maybe you realize you didn't understand it as well as you thought you did. What I think is interesting about this process is that it’s a process of learning by thinking. When you're explaining to yourself or to somebody else without them providing feedback, insofar as you gain new insight or understanding, it isn't driven by that new information that they've provided. In some way, you've rearranged what was already in your head in order to get new insight. .... " 

Superforecasting and Straight Thinking

In this week's NYT book review, a piece on  “Superforecasting: The Art and Science of Prediction,” by the psychologist Philip E. Tetlock.   I have mentioned the work here a number of times.   This overview does a good job presenting key elements.   Also a review of “Mindware: Tools for Smart Thinking,” by the psychologist Richard E. Nisbett.  I have not read the latter,  but based on the review it is a look at principles of thinking that are well known, but could serve as an introduction for students at all ages.