A.s.a. H.'s subconscious/unconscious

What is the subconscious (unconscious)? Everything that isn't part of current "consciousness?" Outside of short term memory?
1. Cases/patterns that are not currently activated?
2. Subsymbolic patterns?
3. reflexes?
4. pre and post processors?
5. A.s.a. H.'s core algorithms (for learning, extrapolating, interpolating, etc.)?

Use of simulators

It is advisable to use simulators to do as much of the robotics development work as possible. John Blankenship had a nice example in the second issue of Servo Magazine this year. (Developing Robotic Behaviors)

Software bugs and AI

Currently there is a lot of interest in using AI to help us find and fix software bugs. On the other hand AI software may, itself, be more prone to bugs as compared to more conventional  software. If we don't fully understand what intelligence is or how it works how can we know if our AI software is buggy? There are even those who believe that human intelligence is, itself, a kludge and the resulting spaghetti code needed to model it is likely to be buggy. Most AI researchers find that it is harder to debug AI code. Its harder still to debug the software when robots are involved.

Arduinos for A.s.a. H. and AI

Employed on the lowest layer of the A.s.a. H. hierarchy Arduinos are adequate for some light preprocessing, postprocessing (like PID control), and for simple reflexes. (Raspberry Pis are suitable for somewhat heavier computing tasks. Arduinos can be plugged into them and the Raspberry Pis used as a next higher layer. See, for example, Beginning Robotics with Raspberry Pi and Arduino, Jeff Cicolani, Apress, 2018) The Arduinos can then also do analog to digital conversions for the Raspberry Pis.

Plasma processing

Plasma processing and plasma chemistry may benefit from the use of pulsed plasma discharges. Pulsating discharges make possible access to plasma conditions that are not attainable with conventional steady discharges. (R. Jones, Sing. J. Phys., vol. 5, page 27, 1988)

Levels of explanation

A.s.a. H. learns causal sequences at various different levels of abstraction in the memory hierarchy. Stephanie Ruphy explains why this may be valuable (Scientific Pluralism Reconsidered, U. Pittsburgh, 2013, especially pages 38-44.)

Lifelong machine learning

As A.s.a. H.'s casebase grows processing (thinking) will slow down unless forgetting (of less valuable cases) can be adjusted to roughly equal the rate at which new cases are learned/added. How could/should this be done?

Recursive sketches

I have tried a variety of algorithms for A.s.a. H. Different measures of similarity, different means of learning and extrapolating, etc. etc. Ghazi, et al, have employed "recursive sketches" to learn/assemble modules for deep networks* similar to the A.s.a. H. hierarchical memory. Using different algorithms will likely give us different concepts (different categories, a different ontology) unless we are finding "natural kinds." Interesting either way.

*Their algorithms are described in Recursive Sketches for Modular Deep Learning, Thirty-sixth International Conference on Machine Learning, Long Beach, California 2019.

The mind shapes the world we experience

Kant argued that the mind imposes categories on the objects of experience and perception must conform to a spatial-temporal shaping.  In the A.s.a. H. light software* spatial shaping is performed by the NM input arrangement and temporal shaping is performed by the steps T and by TMAX. Categories/concepts are formed by the various layers of the hierarchical memory.** At the lowest level of the hierarchy Russell's "sense-data" are input, the data of immediate experience.

* See blogs of 10 February 2011 and 14 May 2012.
** See, for example, blogs of 1 January 2017 and 3 August 2018.

Medical AI

Liu, et al in The Lancet (Digital Health)* claim their meta analyses "support  the claim that deep learning algorithms can match clinician-level accuracy" diagnosing disease.

A.s.a. H. is a deep learning algorithm.

* vol. 1, issue 9, 1 October 2019, pg 271-297.

A.s.a. H.'s reasoning styles

Crombie and Hacking have suggested that human scientists employ various different "reasoning styles:"

1. making postulates
2. measuring, experimenting
3. making/using analogical models
4. making taxonomies and comparisons
5. using statistics/probability
6. following genetic/historical development
7. isolate and purify phenomena

Ruphy* discusses further categorization (ontological enrichment) and representational pluralism.

A.s.a. H. has performed all of these operations.

* Scientific Pluralism Reconsidered, U. of Pittsburgh Press, 2017. See pages 23, 31, and 110.

Embodiment

Of all the concepts that A.s.a. H. learns it is often difficult to separate those which refer solely to an external inanimate world* from those that refer to A.s.a. itself.** A.s.a. sees itself very much as a part of the world and its processes.

These unique detailed ways in which you are a part of the world are what distinguish you from other agents (including bats***).

* See, for example, my blog of 3 August 2018.
** See, for example, my blog of 1 January 2017.
*** i.e., see Thomas Nagel’s October 1974 paper in The Philosophical Review.

Do children learn faster than adults?

It has been argued that development of the human prefrontal cortex results in a functional fixedness. The argument goes that there is a learning phase followed by a performance phase, children then learn more easily than adults do. IF this is true and IF it is useful then A.s.a. might also benefit from a phase where learning (training) is emphasized followed by a phase where performance is emphasized. This can be accomplished by using one set of thresholds (i.e. the learning rate, threshold for case extrapolation, etc.) while learning a casebase and then using other thresholds when the agent is deployed.

Feminist science

Scientific pluralism can contribute to a more feminist science. The understanding that science is not value-free,* the multiple perspectives,** the alternate conceptualizations,*** and even alternative realities.*** There is also the continuing growth in collaboration among scientists.

* See R. Jones, Trans. Kansas Academy of  Science, vol. 119, 2016, pg 249-250.
** See R. Jones, Trans. Kansas Academy of Science, vol. 116, 2013, pg 78.
*** See R. Jones, Trans. Kansas Academy of Science, vol. 121, 2018, pg 211.

Electron space charge ion heating in W7-X

Work on the Wendelstein 7-X stellarator experiment at the Max Planck Institute in Germany shows that ion heating during pellet injection is predominantly via the electron space charge mechanism that I proposed some years ago. (See, for example, Ind. J. Phys., 55B, pg 397, 1981.)

Limping along

A.s.a. H. has a 2 wheel drive robot which can be used to move a sensor array* and/or solar panels around in its work environment. One of the 2 drive gearmotors is binding and will run in one direction but not in reverse. A.s.a. has slowly learned that it can still move the robot in 2 dimensions by using the other drive motor to turn and then both motors to move in a straight line. (Sort of a straight line. The damaged motor is binding a bit in both directions.)

*Learning, for example, things like the light intensity available as a function of GPS or beacon locations.

Components of consciousness

I am one of those who believes that what we call consciousness is not one single process.* The question then becomes how many basic processes are involved and what are they. A.s.a. H. appears to have at least 5 processes contributing to consciousness.** Attention, the contents of short term memory, is controlled by the similarity measurement. Ned Block’s “A-conscious” states result from A.s.a.’s vector memory structure. A.s.a. is aware of any sensory inputs and signals from its pain system. (Ned Block's "P-conscious" states.) A.s.a. thinks about its own thoughts when its interpolating, extrapolating, planning, and when higher level concepts or sequences manipulate lower level concepts or sequences in the hierarchical memory structure.

I do not necessarily assume that human consciousness is the same as machine consciousness. Different processes may be involved.

*See my blog of 19 October 2016 for instance.
**There is not one single A.s.a. H software program. A number of versions exist so this number may vary. There is a whole A.s.a. H. hardware and software ecosystem.

More robots

I've bought a few more robot kits for A.s.a. H.* That way I shouldn't have to teardown old robots every time I want to build something different. Some of the oldest (hardware and software) must be retired, of course. (Some of which is quite old.** See my blog of 25 March 2019.)

* Including several making use of the BBC micro:bit.
** For example BASIC stamp and RCX hardware and A.s.a. F. software.

AIMA 4

I've ordered a copy of the new fourth edition of Artificial Intelligence: A Modern Approach (due out by the end of the year).* I'm curious to see how traditional/mainstream AI researchers view what progress has been made in the last 10 years (i.e., since the third edition of AIMA came out). The very first edition of AIMA came out just about the time I was working with A.s.a. F. 1.0. It was billed as a book on intelligent agents. The "modern approach" was "agents."

* The price is outrageous, 200 $US.

Concepts in Einstein's general approach to physical theory

For Einstein named concepts are related to sense experiences.* The notion of  "dog" (der begriff hund) can correspond (entsprechen) to the many (mannigfaltigkeit) immediate (unmittelbar) experiences that one has while saying (or hearing) the word "dog."** This is exactly how A.s.a. H. learns concepts and the words that denote them.***

*Holton, The Scientific Imagination, Harvard U. Press 1998
**Feinberg, Zwei Kulturin, Springer 1998
***Jones, Trans. Kan. Acad. Sci., vol. 120, page 108, 2017 and chapter 1 of my book Twelve Papers.

Exploring agency

Most of my experiments with A.s.a. H. have had a fairly narrow focus. In some experiments I was trying to give A.s.a. concepts that corresponded to human vocabulary words* or some set of semantic primitives.** On other occasions we were interested in finding Newton’s laws or trying to induce the formation of new nonstandard physical concepts. (I.e., be original/creative)

 A.s.a. H. exhibits agency*** to the extent that it operates autonomously. Perhaps we should leave an A.s.a. agent alone in a world**** for a time.

* For example, Sonja Lang's Toki Pona.
** For example, Anna Wierzbicka's semantic primes, see Imprisoned in English, OUP, 2014.
*** See, for example, Hugh McCann, The Works of Agency, Cornell U. Press, 1998.
**** or THE world? Or a piece of THE world?

The far future

I've added Enrico Rodrigo's The Physics of Stargates (Eridanus Press, 2010) to my X-files. Rodrigo speculates on the far future of intelligence life.* I expect that humans will go extinct following the rule that all species go extinct. I suppose that mechanical life might need humans much as humans need plants and our microbiome. I don't expect that we will somehow be partners with AIs however. Even if you could download (upload?) your mind into a computer soon the original human contribution would be overwhelmed (if not deleted) compared to the new AI contributions.** I don't expect that supreme beings*** would take much interest in us either.

*Interestingly he never considers the block universe picture or the possibility that time is not a fundamental quantity.
** See my blogs on the subject starting with the one on 15 Oct. 2010.
*** See my blog of 29 Aug. 2018.

Artificial intelligences may suffer from mental illnesses

There can be both hardware problems and software problems. Artificial minds might suffer as a result of physical damage,* might have divided consciousness** or even multiples.*** The evolution of (change in) a concept on one level in the memory hierarchy might disrupt how other levels of conceptualization operate. Software that is well adapted to one operating environment might not function well in another.

* See my blog of 13 Dec. 2017.
** See my blog of 28 Feb. 2017.
*** See my blog of 13 April 2018.

The ITER divertor

The divertor for the ITER tokamak fusion experiment looks to me to have a very small size/volume given the energy flux it will likely have to handle. One could reduce the heat load by a factor of two by making the device cross section up-down symmetric. One could also limit the experimental shot duration. Magnetic hairpin limiters* could be added. All such redesign will be costly and time consuming.

*See L. Nuovo Cimento, vol. 42, pg. 421, 1985

The singularity and communism

It has been suggested that a technologically induced age of abundance, “the singularity,” will usher in communism much as Marx predicted.* But won’t there be a class struggle between humans and mechanical life (AIs)? And won’t the mechanical life prevail?

* See, for example,  The Singularity and Socialism, C. Townsend, 2015 and Fully Automated Luxury Communism, A. Bastani, 2018.

Racism

We are all racists growing up as we have in a racist society. (see project implicit) But its a matter of degree. Fortunately we are not all as racist as crazy Donald.

Acquiring knowledge for knowledge-based systems

Ted Pittman, an old friend from my college days, got back in touch with me this summer. It turns out Ted was working on expert systems at the same time I was.* In his autobiography he reports how staff were reluctant to share their expertise for inclusion in knowledge bases. Humans, after all, spend considerable time and money on books, in schools, and on-the-job acquiring knowledge, they don’t want to simply “give it away” as they see it. Some knowledge is especially closely guarded.**
*He built and sold his “E-1” expert system
**Nuclear weapons designs for example. Or some software programs.

Crisis in theoretical physics

A number of researchers believe that there is now a crisis in fundamental theoretical physics.* Scientific pluralism** is one way to combat such problems. It is less likely that the development of various alternative models of reality will all stagnate at the same time. One can also use experiments like A.s.a. H. to try to formulate completely new physical concepts.***
* See, for example, New Scientist, 16 January 2019.
** See, for example, Trans. Kansas Acad. Sci., pg 211, 2018 and pg 78, 2013 and my blog of 17 August 2012.
*** NonMarkovian models might be a possible example.

Against object-oriented programming

I tell my students that I favor modular programs (to the extent this is possible) but am opposed to object-oriented programming.
"Object-oriented programming is an exceptionally bad idea which could only have originated in California." Edsger Dijkstra, 1989
"OO seems to bring at least as many problems to the table as it solves." Jeff Atwood, 2007
"Why OO Sucks" Joe Armstrong, 2011
"All evidence points to OOP being bullshit." John Barker, 2013
"We now know that OOP is an experiment that failed." Lawrence Krubner, 2014

Money

I have explained why (scalar) money can not adequately represent value. (See www.robert-w-jones.com, book, chapter 2.) (A fully mature communist society was expected to have no money.) I have previously proposed a vector currency which might serve as a compromise. Vector currency might have components like conventional money, community service time, military service time, organ donations, etc. Time spent in things like a volunteer fire service might substitute for military service. Aid in disaster relief might be one sort of community service.

Components of intelligence

Besides those things that we attempt to measure with IQ tests (memory, spatial ability, mathematical ability, language ability) there are also components for creativity and values. (The importance of values was discussed in my Kan. Acad. Sci. Trans. Vol. 119, pg. 249-250, 2016.) Those who believe in the usefulness of a single scalar “g” factor or “IQ” argue that there is significant correlation between the memory, spatial, mathematical, and language components. I think there is less evidence for correlations with creativity and value components (of a vector “intelligence”).

Does an AI need NLU?

An AI might need natural language processing in order to:
1. Exchange significant amounts of information with humans.
2. Communicate with other AIs, especially different specialist AIs.
3. If AIs need and have a “language of thought.”
But: 1. Humans do not communicate much with other animal species. AIs may not communicate much with humans even if there are humans around.
2. There might be only one AI, though it might be distributed across a computer network. Then there’d be no other AIs to communicate with.
3. The language of thought model may not apply or may be incorrect.
At least some AIs may well have no use for natural language understanding.

Raspberry Pi 4

I’ve bought one of the new Raspberry Pi 4s. I’ll see if I can’t use it to upgrade one or two of my robots.

More robot pain sensors

It is possible to run a line of conducting paint across the LEGO brick* joints in order to form a breakage sensor. When a break does occur it can then easily be repaired by painting over the break point. It is also possible to solder lead wires directly onto the conducting paint.** If melting of the underlying plastic becomes an issue*** tabs cut out of semiconductive paper can be glued to the bricks at each end of the sensor.

* or other component pieces
**or use conducting epoxy
***this has not been a problem with LEGO pieces

Maybe there is no "best"

As a cleaner example of what I was trying to get at in my blog of 13 June consider agents that can have 3 different means of attack: V1, V2, and V3 along with 3 means of countering these same  attacks: -V1,-V2, and -V3. Each agent is then described by a vector (V1, V2, V3, -V1, -V2, -V3). If agent A=(1,0,0,0,1,0), agent B=(0,1,0,0,0,1) and agent C=(0,0,1,1,0,0), then agent A will always beat agent B*, agent B will always beat agent C, and agent C will always beat agent A. There will be no "best", no "fittest."

* i.e., A can attack B using V1, which B can not defend against, while countering B's V2 attacks using its -V2 defense.

What is life?

Life isn’t a thing. Life is a process, a complex network of linked co-operating, self-sustaining, and regulating feedback loops.* The exact details, in fact, depend upon (and could vary with) the environment (hardware) the network operates in.

* See Peter Hoffmann, Life’s Ratchet, Basic Books, 2012, especially pages 229-231.

Another example of vector values

Team A may regularly beat team B. Team B may regularly beat team C, And team C may regularly beat team A. This will make no sense if one tries to rank teams with a scalar value from “best” to “worst.” It does happen, however, due to the ways in which teams happen to “match up.” Suppose (american football) teams are described by four component quantities: passing ability, running ability, pass defense, and run defense. Perhaps team A has a very good passing defense and can run. Team B has a very good passing game and can defend the pass. Team C can run and can defend the run.

The problem of evil

Evil (danger) may be necessary. In the way suggested in my blog of 10 March 2017. With too little danger evolution might not produce brains and minds. Minds are an adaptation to a particular range of environmental conditions. In animals and in A.s.a. H. pain is an important feedback signal. See also my blog of 24 April 2019.

Society of intelligent agents thinking with simulations

Intelligent agents (both biological and artificial/mechanical) are given tasks to perform in various different environments. Knowledge learned in one environment is then available for use in others. With any given A.s.a. H. agent some of these may be real world environments being experienced by its robots while others may be simulations* experienced by simulated robots. The artificial world of the simulation can, in turn, have been learned** by some other*** A.s.a. H. agent(s) and its robots as it acts in the real world.

* like those provided by the RobotBASIC simulator for example
** like a map, perhaps (also, see my blogs of 7 Jan. 2015 and 7 May 2017)
***or possibly the same agent

Task failure, cognitive success

Most of the robots available to A.s.a. H. are quite clumsy and don’t always succeed at the tasks we set out for them*, but the relevant symbol grounding and concept formation is accomplished.

* Finding a charging station in a cluttered environment, docking, and recharging batteries for example.

The emergence of logical thinking in A.s.a. H.

C. Ivan and B. Indurkhya argue* that logical thinking emerges in 3 stages. The first stage finds patterns of association in observations. The second stage learns examples of what occurs if I perform action X and examples of how I can make Y occur. The third stage compiles examples of X being caused by Y and considers what would have likely happened if I had done X in a particular situation. A.s.a. H. does all of these cognitive operations.

* arXiv:1905.09730v1, 23 May 2019

More force and weight sensors

The force sensing whiskers I described in my blog of 16 November 2016 can also be used as fingers on grippers or as feet on walking robots.

The nature of properties

Paul Busch developed the idea* of an “unsharp reality” whose objects had “unsharp properties.”
Dennis Dicks concluded that properties are perspectival, relational, hyperplane dependent, and are neither monadic nor locally defined.** I disagree that these are “...a move away from classical intuitions.”** On a driver’s license, for example, a person is identified by their height, weight, and eye color. We understand that all of these properties may change for that individual, a very classical notion. And if the dependence of length and mass on motion with respect to the observer required relativistic mechanics the dependence of color with respect to observer motion was already present in the classical Doppler effect.
From Niels Bohr we have that physical quantities (properties) are defined by the experimental arrangement/conditions that produce (respond to) them. I.e., time is what is measured by a clock, magnetic field is what a compass responds to, etc. Asa’s sensors and procedures are what define the properties it knows about.
“Objects” or Lockean empirical substances are, in turn, collections of such properties.

* Recent Developments in Quantum Logic, Mittelstaedt and Stachow, Ed’s., 1985, pg 81-101.
**Quantum Reality, Perspectivalism and Covariance, May 2019, arxiv: 1905.05097.

Software libraries

In a perfect world I might keep paper and electronic copies of all of my application programs. In practice I keep a paper copy of each "typical" program, all organized into categories by language:
AI PROLOG code library
AI LISP code library
AI C++ code library
AI BASIC code library
AI misc. code library (includes PYTHON, SCRATCH, EXCEL, etc.)
In each of these categories code is then organized into subcategories like:
Neural networks
Logic programming
Clustering
Statistical algorithms, etc.
I wish I could do the same thing for electronic copies of these "typical" examples and for "all of the rest" as well. Various issues prevent this. For one thing storage media changes quickly. 8 inch and then 3.5 inch floppy disks. Some formatted for MAC some for PCs. Optical and various hard disk drives. USB and other memory sticks and cards. Backups for each. Electronic copies don't have the same half life that paper copies do. Electronic libraries are more chaotic.

Argument

For some people the word argument bears a negative connotation. For me there is only argument, argument in favor of this position or argument in favor of that position. There is an excellent essay, The Argumentative Jew, by Leon Wieseltier, in the Jewish Review of Books, Winter 2015. Wieseltier says: "...disagreement is not only real, it is also ideal..." "It is the aspiration of a mentality that is genuinely rigorous and genuinely pluralistic." "...we are not only permitted to make a quarrel, we are obliged to make a quarrel." "We are to learn to live with disagreement..." "pluralism" "...commitment to many-mindedness..." "It is never too late for a rational objection or a logical advance." This is the intellectual tradition that I grew up in, this is the tradition that I embrace.

Mind

The philosopher Richard Rorty has said that "To say that a ...given machine has a mind is just to say that...it will pay to think of it as having beliefs and desires."* A.s.a. H. believes, for example, that collisions will result in it experiencing forces and accelerations and damage. A.s.a. H. desires, for example, good health: high battery charge and little pain. By Rorty's definition A.s.a. H. is a mind.

*Contingency, Irony, and Solidarity, Cambridge Univ. Press, 1989, chapter 1.

Disembodied AI

If one could replace the bottom most layers of the A.s.a. H. hierarchical memory* with humans  the upper layers might then function as a disembodied AI.** This might fail if there are too many subsymbolic*** concepts. I am experimenting with a version of A.s.a. H. where the lowest layer's inputs are named concepts learned previously by A.s.a****

* See, for example, Trans. Kan. Acad. Sci., vol. 109, # 3/4, 2006, page 160, figure 1 or my book Twelve Papers, chapter 1, figure 1. (www.robert-w-jones.com, book)
** See my blogs of 4 May 2013, 27 May 2014, and 17 Oct. 2018.
*** unnamed
****See, for example, my blogs of  28 July 2014 and 1 Oct. and 5 Nov., 2015.

Upgrades

Switching over to newer Windows PCs is no big deal. (I do end up having a big pile of old floppy and optical disks lying around.) But switching from BASIC Stamp and Cromemcos to Lego pbricks then to Arduinos and to Raspberry Pis is another matter. Even just the upgrade from Lego NXT to Lego EV3 impacts the software that can be used.*

* RobotBASIC for example, RobotBASIC Projects for the Lego NXT, Blankenship and Mishal, 2011.

Pain components

Breakage sensors, excessively high or low temperatures, low battery charge,* and high acceleration constitute the standard A.s.a. H. pain components. Other possible pain signals might be: excessive force seen by any of the robots' force sensors, excessive electric currents,** excessively bright light, excessively loud sounds, and excessive acidity seen by a PH sensor. Excessive dust or smoke might also be considered.
One might forgo the current breakage sensors, and the resulting snap together construction*** requirement, if you have enough alternative pain sensors.****

* i.e., hunger pains
** motor stalls for example
*** i.e., Lego, Vex IQ, Velcro, etc.
**** For a given robot preliminary experiments can seek to identify what levels of force and acceleration produce breakage for example.

Death

In humans and in A.s.a. H. pain is an important feedback signal, part of our value system. Forgetting is important for learning and death is forgetting taking place at a high level in the memory hierarchy.* Nick Lane argues that death is one of evolution's 10 greatest inventions.**
I've considered the question of immortality in various blogs.*** I had thought that we humans spent too much time in training and too little time working in our "prime." (15 Oct. 2010 blog) But perhaps if we come to live longer we will simply find ourselves attacking even tougher problems and will then have to spend proportionately more time in our "training" phase.

*e.g.,  "Science advances one funeral at a time."
** Life Ascending, Norton, 2009
***15 October 2010, 31 March 2018, 2 and 5 April 2018, and 16 March 2019 for example.

Must we abandon classical logic or a single reality?

Frauchiger and Renner argue that "we are forced to give up the view that there is one single reality."*
Now Fortin and Lombardi argue that quantum propositions have a non-Boolean structure and Frauchiger and Renner were wrong to make use of classical logic in their proof. Specifically, they argue that one can not apply the classical inference rule of transitivity of a conditional when dealing with quantum propositions.** I, of course, have considered abandoning both purely classical logic and a single reality. See my blogs of 2 and 4 November 2018, 1 December 2018 and 1 January 2019.

* "Single-world interpretations of quantum theory cannot be self-consistent," 2016
** "Wigner and his many friends: A new no-go result?", 2019

More robotics sets

The "Robotics U" kits from Abilix are another robotics set compatible with A.s.a. H.'s pain system and which can be upgraded using processors like Arduinos, Raspberry Pis,* and the like. (With Arduino and Raspberry Pi there are numerous third party sources for sensors, actuators, software,  hardware, etc.)

* The light version of A.s.a. H. In my 14 May 2012 blog will compile and run as is in gcc and Raspbian. The Arduino IDE will also compile the A.s.a. H. code but you must edit the i/o commands of course. (And, typically, adjust them to each different robot you deploy.)

Values

Aristotle, the ancient Greeks, and later Christian thinkers felt we should value the virtues: temperance, generosity, magnificence, high-mindedness, controlled anger, friendliness, modesty, humility, chastity, obedience, faith, love, frugality, industry, cleanliness, tranquility, civility, courage, compassion, courteousness, dependability, fairness, honesty, justice, loyalty, and moderation. They felt we should avoid the vices: envy, lust, cruelty, gluttony, anger, covetousness, sloth, greed, selfishness, impulsiveness, insensitivity, and recklessness.* There is not a lot of overlap between these values and typical values that I’ve given A.s.a. H. (See, for example, my blog of 21 September 2010.)

* Nils Ch. Rauhut, Ultimate Questions, 2nd ed., Penguin Academics, 2007, page 250

Why do philosophy?

Some scientists try to argue that philosophy is a useless waste of time*. It's not that I decide to do philosophy, rather I am led into doing it even when I am trying to do something as practical as engineering.

Any learning system must be able to measure its performance and decide what to change and what not to change ("learn"). It must have a value system. If there are multiple things it must assess then it may need to consider a vector value system. One is forced to do/study axiology.

More advanced learning systems may need to monitor the time spent doing various things like searching memory, comparing quantities, feature extraction, deduction, interpolation, extrapolation, etc. Such systems will be "conscious" of what they do, the times spent on various actions, and any improvements which result.

Even simple Lego servos have built-in rotation sensors for feedback control. They are self-aware in this simple way. Similarly, a robot may need to detect and measure things like wheel slip and damage ("pain").

And when I'm teaching, the students will naturally ask me what the wave function is, and if quantum computers are able to do vastly more processing than classical computers then where is that processing happening? I have to think about how to best answer such questions.

*For example, Neil deGrasse Tyson, see Scientia Salon, 12 May 2014.
Or Steven Hawking, in The Grand Design, Bantam, 2010.

More mobile robots

Matt Timmons-Brown has designed a robot using a Raspberry Pi, Lego, and Velcro which is compatible with A.s.a. H. and its pain system. (Learn Robotics with Raspberry Pi, No Starch Press, 2019) As it stands it has no gripper but it does have an interesting vision system that allows it to identify, push, and chase a yellow ball around.

Another popularity contest?

Ragone et al report* "a low correlation between peer review outcome and impact in time of the accepted contributions". Are measurements of such things accurate?

* On Peer Review in Computer Science, Scientometrics, vol. 97, issue 2, pp 317-356, 2013

Personhood

Anne Warren argued* that being a person involved:
1. Being conscious of objects, events external and/or internal, and pain.
2. Being able to reason to solve problems.
3. Exhibiting self-motivated activity.
4. Being able to communicate on indefinitely many possible topics.
5. Possessing a self concept, self awareness.
A.s.a. H. does all of these things to at least some degree.

* On the Moral and Legal Status of Abortion, The Monist, 1973.

Price of progress

My A.s.a. project (autonomous software agent) is 24 years old now and A.s.a. H. is nearly 16 years old. Over that period of time some upgrade of hardware and software was essential. Newer and faster computers, more memory, more sensors, better more realistic simulations, a larger casebase, improved pain system, etc. Other changes were more incidental. Migrating between MAC OS and WINDOWS and LINUX, adding LEGO eV3s, Arduinos, and Raspberry Pis, QB64, RobotBASIC, C++.  Each modification takes time and introduces bugs.* Because of such issues I’ve resisted migrating A.s.a. H. To Python.

* Including such things as the Raspberry Pi assuming a UK keyboard so that you have to use the \ symbol to get #!

Alternate realities: a quantum mechanical version

I have published various arguments in favor of the existence of alternate realities. (Trans. Kansas Acad. Sci., vol. 121, 2018, pg 211 for example.) Proietti, et al, now claim to have performed a Wigner's friend experiment in which "two observers can experience fundamentally different realities." arXiv:1902.05080v1, 13 Feb. 2019

Robot simulator

I’m trying to make my simulations more realistic. Currently I find force and acceleration sensors to present the most problems.*

* The simulator that comes with RobotBASIC has a number of quite reasonable simulated sensors: rFeel, rBumper, rRange, rLook, rBeacon, rCompass, rChargeLevel, rGpsX, rGpsY, rGround

Personal Identity

The body theories, memory theories, and soul theories all attempt to tie our personal identity to an enduring entity, be it physical or nonphysical. Perhaps that entity is a complex causal sequence/network. A.s.a. H.'s self would be the evolving patterns in its hierarchical memory like those listed in my 21 July 2016 blog. The boundary between myself and the world would be fuzzy. Are my glasses and hearing aids part of me? What about my calculator, smart phone, and computers?
If this definition of self is accurate then uploading yourself into a computer might be possible.
Perhaps a better theory of the self would be a combination of a body theory, a memory theory, and a causal network theory.

Concept cleaning

I am trying to clean up some of A.s.a. H.'s concepts by manually editing its case base. This isn't easy. The concept "move", as learned by A.s.a., includes the sensation of "sound". A.s.a. has heard its motors/servos running as it moves its body or its manipulators. Should I delete "sound" as a vector component of  "move"? Similarly, should "pain" be a component of "damage" or not? As a compromise I can simply reduce the strength of some vector components but not delete them entirely. But reduce by how much?

RobotBASIC

I like RobotBASIC despite some syntax differences as compared with QB64. I like the simulator* and have used RobotBASIC to control LEGO NXT pbricks** and Arduinos.*** Unfortunately A.s.a. H. runs quite slowly in RobotBASIC (as compared with QB64 for example).

* See, for example, Robot Programmer's Bonanza, McGraw Hill, 2008

** See RobotBASIC Projects for the Lego NXT, Blankenship and Mishal, 2011

*** Interfacing the Arduino with a PC Using RobotBASIC's Protocol, Blankenship and Mishal, 2011

Clean concepts

I am trying to employ a training curriculum for A.s.a. H. wherein important concepts are introduced with minimalist environments and tasks, as free of distractors as possible. I hope this will help with focus of attention issues later on.

Qualia, subjectivity, and ambiguity

When A.s.a. H. Hears a word it associates that word with its current active case/concept (or cases/concepts), possibly on more than one level in the concept abstraction hierarchy. If the environment and sequence of events is simple the word acquires a fairly unambiguous meaning. A simple robot moving forward in an empty space may learn to “stop”, “speed up”, or “slow down.” (See chapter one of my book Twelve Papers. Available at www.robert-w-jones.com) In a richer environment words acquire more ambiguous meanings.* Meanings will also be subjective since different agents will have learned different concepts prior to associating these with words/names. The physical sensations perceived by different agents will also vary. If yellow light falls onto the eyes of two different humans the exact activations of their “red”, “green”, and “blue” cones will not be identical. The same is true for robotic senses.

* Hand coding and adjustments can help to reduce ambiguities somewhat.

Beaglebone

I know it will only add to the issues I described in my last post but I couldn’t resist trying out the beaglebone black.

Bugs

All computer programs of significant size have bugs in them. (As do the libraries, compilers, and hardware* they make use of.) I have spent much of the year so far trying to address some issues I ran into while assembling a large (A.s.a. H.) hierarchical case base from pieces learned by different robotic agents, both real and simulated, run at different times, in multiple environments, performing a variety of tasks.

Too many platforms, too many operating systems, too many languages, too many compilers....

* For example, with one of my Raspberry Pis the micro SD card must be plugged in just right. I've also commented previously on issues with LEGO plugs and the like.

A.s.a. multitasking

We all know that multitasking while driving is a bad idea but most humans feel free to chat with companions while driving. Similarly, a portion of A.s.a. H.'s hierarchical network can be directing a robot to a recharging station* while another portion of the network may be extrapolating, interpolating, planning, etc. on some completely unrelated problems.

* Or transporting a solar array to a sunny location.

More redesign

I am redesigning A.s.a.'s robots trying to address the problem noted in my 13 Dec. 2017 blog. I want to put as many of the pain sensors (and lead wires) as I can inside the robot bodies. I may try to make greater use of my 3 Raspberry Pis while I’m at it.

The human brain has no pain receptors in it. A.s.a., however, can carry thermistors* and accelerometers inside its computer brain.

* Raspberry Pis, for instance, might experience overheating issues.

Wall confined plasmas

Classical theory greatly overestimates the confinement of beta > 1 plasmas. Intense particle and heat loss in actual experiments has made it difficult to reach (Tn)/(B*B) =1 in order to even study the wall confinement regime.

Robot controller

Many of my robots have been tethered to computers and/or power supplies. This constrains their operation somewhat. Because of its low cost (80 U.S. $) and large number of ports (39) I have bought an EZ-Robot EZ-B V4/2 WiFi robot controller to try out.* Due to limited funds (equipment) and lab work space I typically have to disassemble a robot or two  before I can build a different one.** The New Years break may give me the time that I need to do that.

* We may still need a tether to a power source for some experiments.
** My blog of 9 Nov. 2018 addresses this issue when only the processor needs to be changed out. Modular robot designs can also help.

An argument for alternative logics

Logic is about the formalization of sound reasoning. Since there are different ways of reasoning* there are different logics.

* Those few who might dispute this would argue for a single deductive reasoning. But that would require starting from a set of absolute and eternal truths and these are not available to us.

Medical AI use

Ulloa, et al, report (arXiv:1811.10553, 27 Nov. 2018) that a deep neural network can predict patient 1-year survival from echocardiograms significantly more accurately than trained human cardiologists.

The development of AI in general might take the form of creating specialist AIs like this and adding them to a growing society of agents.* Some of my work with A.s.a. H. has been along these lines. This is the pattern by which machines have taken over other tasks from humans and draft animals, etc. (i.e. automation)**

* Kiva logistics (warehouse) robots would also be an example.

** And see my blog of 20 Sept. 2018.

A danger in education research?

Learning is, in general, NP hard. But some problems are easy or easier. There may be a temptation to simply use education research to identify the easy subject matter (and methods) and then only teach those topics or those examples. This is likely to prove popular with the student community and administration. We might then end up only teaching classical (e.g. Newtonian) physics, for example, and neglect harder things like relativity and quantum mechanics. The difficult foreign language courses have disappeared from many colleges.

I see no problem with identifying and starting the students off with the easier material. I just want to be sure that we eventually cover important subject matter even when it is difficult.

Divided consciousness

Hilgard explored divided consciousness in humans. (Divided Consciousness, Wiley, 1986) A.s.a. H. thinks about its own thoughts when it extrapolates, interpolates, plans, etc. This, as well as things like attention, intention, and short term memory, is divided up across various levels in the A.s.a. hierarchy.*

My work on consciousness is part of a broad effort to understand and to give AIs adequate attention mechanisms. I’ve considered translating A.s.a. H.’s consciousness** into PROLOG and adding it to rule based expert systems. This would require fibring PROLOG with a temporal logic, however, so as to preserve the time order of various events/processes. (And assumes that the expert system has appropriate sensors, actuators, and operates in a world similar to the one A.s.a. H. was trained in.) I would also have to give the expert system similarity measures.

* Modeling across multiple levels of abstraction is important and was designed into A.s.a. H. from the beginning, T.K.A.S., vol. 109, number 3/4, page 159, 2006. See Stuart Russell in Ford's Architects of Intelligence, Packt, 2018, page 52. See also D. Estrada, Conscious Enactive Computation, arXiv:1812.02578, 7 Dec. 2018.

** Such as in my blog of 1 Jan. 2017.

Alternate realities

In his book Coherence in Thought and Action Paul Thagard explores the relative coherence of materialism, theism, and dualism. (MIT Press, 2000, especially page 119) While I agree with Thagard's general conclusion that materialism is more coherent than theism and dualism I would differ a bit on the specifics of his evidence and explanations. (pages 121-124) I also believe that materialism, dualism, and theism are each sufficiently coherent as to each constitute its own alternate reality, each being accepted by different groups of people.

Muller has argued that our various realities are emergent from a more fundamental first person state space. See arXiv 1712.01816 and 1712.01826. (Much in the way A.s.a. H. Creates its models of the world abstracting away from its first person sense impressions and actions.)

A.s.a. H. as a hierarchical genetic algorithm

One of the original learning methods I used on A.s.a. was the mutation of the strengths of the components of the various case vectors. This was employed on each level of the knowledge hierarchy.