Tuesday, September 29, 2015

Pain sensors for Lego NXT robots

Fine wires can be strung between Lego NXT robot components (e.g., head, body, arms, etc.) or even between each individual brick.  When the pieces separate (break) the wires break and this provides a pain signal to the Lego brain brick or Asa H mind software.

Saturday, September 26, 2015

Open house

Today is the Emporia State "family day" open house. The state of Kansas wants us to have 10 physics majors per year so there is a lot of emphasis on recruiting.  For me at least physics, and science in general, is about ideas.  These public events tend instead to be more like carnivals. By and large my colleagues do not seem to share my values. They want to say physics is fun.  I want to say physics is important.

Some things are easy to learn.  But learning is, in general, NP hard.  You can't just learn the easy stuff.  You can't expect it to always be fun. University should not be considered a part of the entertainment industry.

Maybe the english department would do things more to my liking.

Monday, September 21, 2015

Ignoring input

I might learn nothing from reading a book written in Italian.  A specialist expert may learn nothing from a book written about some other specialty.

Asa H can be given specialist knowledge by loading a casebase learned previously. It may be that Asa should ignore input which is outside of its area of specialization. (To do otherwise may just fill the memory with useless information and slow down Asa's cognitive processing.)  This can be accomplished by having Asa ignore input when the similarity measure is sufficiently low. Whole data files can be detected and skipped in this way.

Sunday, September 20, 2015

Unassimilated memories, library use

There are many things which I haven't committed to memory but which I know where/how to find, the mass of a uranium atom for instance or my brother-in-law's telephone number.  I want to give my AI Asa H the ability to use an external electronic library.  Such an auxiliary memory would then expand Asa's knowledge without slowing down its regular search.
What we want is more than a robot librarian.  We want to retrieve individual facts, not whole books. I am looking at the universal decimal classification system for one thing.

Monday, September 14, 2015

Why no larger space program?

If the Russians had put a man on the moon first.  If venera had found life on venus. If Viking had found life on mars. If the space shuttle had reduced launch costs by more than an order of magnitude. If some economically important space processing had been developed on the ISS.
If seti had heard intelligent signals from space. If any of these had happened then we might be sending humans to mars today. But they haven't happened, at least not yet.

Saturday, September 12, 2015


Intelligence requires the definition and use of values.  (See Anderson's ACT-R model or my Asa H) During the course of doing science an intelligence promotes and develops a number of values.  (Things like: the importance of evidence, testability, consistency, skepticism, logic, etc.) I have watched a simple value system as it develops in an artificial mind using my Asa H software. (See my blog of  25 Sept. 2013 for instance.) The values that we develop while doing science are then applied to the other areas of our lives, religion and politics for example. This helps us to create a single unified world view.

Friday, September 11, 2015

Sensitivity analysis

Sensitivity analysis is used in Asa H and other AIs for/during things like feature discovery, generalization, forgetting, abstraction, etc. How sensitively utility depends upon a given input or group of inputs is one important measure.  In many tasks and environments that Asa H has explored there are far more input variables than output variables (actions).  For that reason sensitivity of output variables to variations in input variables and clusters of input variables (features) is also a valuable measure.

Thursday, September 10, 2015

The ideal?

On the move I could work on documents on my smartphone.  (With a USB attachable physical keyboard sometimes.)  When I arrive at home or at the office I plug the phone into a dock which, in turn, charges my phone, may provide additional processing power, physical keyboard, monitors, external bulk storage, printer, scanner, etc.  Files can be transferred to and from the phone as desired. Each element can be upgraded independently.

Wednesday, September 9, 2015


In some strong forms of "embodied cognition" researchers suggest that intelligence emerges from the interaction of brain, body, and world.  I am not a believer in such a strong form of embodied cognition.

I have previously argued, however, that computer simulations can not totally replace real world experiments because simulations contain only the physics we understand and have modeled while real experiments also contain those aspects of reality that we do not yet fully understand. (You don't have to look very far, no one can predict how a piece of paper will flutter to the ground when you drop it this time.)

Similarly, a simulator can only present a software bot with that portion of reality that we understand and have mathematically modeled.  A real physical robot will also experience those features of the world that we have not yet been able to model.

Clearly then, my AI Asa H should get some experience operating real physical bodies.  Just how much experience is needed I can't  say yet. Intelligence is a matter of degree in any case. Some useful AIs might never have contact with the real world.

Tuesday, September 8, 2015

One standard operating system?

Over the last few months (year?) I have been mostly using 2 operating systems, iOS on smartphones and tablets and Windows (mostly 7) on laptops and desktops.  This has involved little communication between the two.  Could one single operating system be better? 

Android on desktop PCs has gotten off to a slow start.  I thought I'd give Windows 10 a try once they got the worst of the bugs sorted out. (I use Windows a lot because it's the standard for my employer and because there is so much software for Windows out there.) I'm not sure if  I've waited long enough but I now have and am playing with Windows 10 on an HP Pavilion x360 two in one. (Running Lisp and Asa H 2.0 for example.)

Friday, September 4, 2015

I, robot

I've been watching how a concept of selfhood develops in my AI Asa H 2.0.  (See, for example, blogs of 4 March 2015 and 28 April 2015)  If there is a strong association with some vocalization from a trainer/teacher it is easy to identify/name the concepts that are developed, e.g.:

Health=(recharge, damage, hear sound "health")
Collision=(sense near, bump, decelerate, hear sound "collision")
Push=(move to, touch, feel contact force, hear sound "push")

But in many other cases the categories may be difficult to name.  The casebase will be just a sequence of numbers in a computer file.