Saturday, January 27, 2018

More Thinking about AI


I've done something to my FireFox, to where the Twitter stuff doesn't come through all fancily formatted, like in the post below. Or maybe your browser doesn't render it either. I'm still seeing the tweet content at least.  Even that might go away, right?

The hope is our committed records will have some permanence but that's all presuming it makes sense to keep all the server farms running.  A sense of heritage might fuel that, but will that be sufficient?  They say server farms are already eating 2% of the grid's energy.  Good thing President Trump is committing through Twitter, for Twitter.  Historians will want to keep that set of records.

Not that Trump is the only tweeter the curators treasure, just he paves the way for Federal funding or something similar to keep universities training up new breeds of engineer, ready to tackle the challenge of keeping history plugged in.

There's a technical challenge in not overwhelming ourselves with more information than we're prepared to handle, even power-wise.

Going back to recent themes...

The invocation of Kant in connection with Abbott:  was the latter breaking any rules in conceiving of beasts of fewer dimension?  We suppose that three dimensions "encompass" two, one, zero and none.

That volume goes up as a 3rd power seems ipso facto the key argument for dim 3 terminology, whereas 3rd powering as a rate of growth (or shrinkage) gets shoved on to Frequency in Synergetics, which is about subdividing.  We have 3rd powering against a backdrop of a Fourness in pure shape, all change rates aside.  That was the 4D.  The F took the 3rd powering.

These shifts in meaning mostly serve a practical objective of making it OK for a cube of face diagonals 2, to have a volume of 3.  That works out when the shape of 3rd powering is either of the duo-tet's tetrahedrons.  With the cube-based model of 3rd powering, you have an edge of pow(2, 1/2) and therefore a cube of volume 2.828427..., not 3.  From this ratio comes S3.

I watched the video below today, as well as the one above. The below one is about brain power and what the chemistry might be, whereas the one above is about electronic circuits learning from feedback to ape, match or exceed human abilities.  AlphaGo.

Lets hope Google stays in business and our blogs live on.  Ditto Facebook and all that.  Scientific literacy means having the ability to grapple with issues while maintaining a cool head and a solutions focused mindset, even in the face of problems that seem insoluble.

A tetrahedron inscribed in any parallelepiped, any hexahedron with oppositely parallel faces, is going to have one third said hexa's volume.  In the case of the cube, the inscribed regular tetrahedron is complemented by four regions comprising the remaining 2/3rds i.e. 1/4 of 2/3 or 2/12 = 1/6.

The regular octahedron that complements the regular tetrahedron to fill space, has volume ratio 4:1, meaning the 1/8th octahedron corners that pack out from the reg-tet, each have volume 1/2. Four corners have volume 2, adding reg-tet gives 3, the cube's overall volume.  One half is one sixth of three.

These simple fractions work best when we give ourselves permission to have such a cube of volume 3, yet with edges that would normally not give that.  The shift is in making 3rd powering a Frequency operation inside an initially "4D" framework, the reg-tet itself, Unit Volume, edges 2R.

Once the logical path is established, go ahead and throw away the ladder and go back to "space is 3D" with an XYZ orientation.  When in Rome.  You've got your touch stones, a way to your favorite garden, but go ahead and surround it with a more conventional brick and mortar wall.

Rather than fight tooth and nail for anything, we're motivated to continue our investigation. When did we come up with height, width and breadth as the names of three dimensions and to what extent do each of these partake of self nature?  A somewhat esoteric question perhaps.

The neural nets that program neural nets get credit for refining their art.  I'm talking about the humans, however quantum minded.  They needed a way to make trial and error count for something. Figure out a feedback loop that continues to fine tune in the face of consistent feedback regarding performance.  As long as the game holds still for awhile...  fortunately chess and Go do.

The sense that AI is "winning" is an over-collapse of an either-or logic, whereas the cell-silicon hybrid is on both sides of the fence.  We encounter our ancestors in the codes they embedded, as silicon learns to echo our sense of music, even logic.  Mutual recognition. A mix meets a mix.

We want to think what we consider thinking has held constant, with maybe machines catching up, but that's not it.  Machines have already changed what we consider thinking.  We needed that word ("thinking") to stay up to date, and so we've somewhat lost our sense of what it meant -- through the ages. Not a constant. What was thinking before electronics closed so many gaps? What was thinking before reading?

Wandering back around...

Keeping a large lawn watered is a chore, but humans have undertaken such duties long term and considered them a privilege to perform.  I'm thinking about servers again, and the spinning water wheels, the hydro-dams. When humans insist on sticking their nose in one another's business, is when many problems arise.  Yet closeness is a fact of life.

Sometimes we're not prying into others' business so much as shielding ourselves from becoming inundated in too many details about affairs we can do nothing about.  The marriage of focus and attention with what's necessary work would need to have divine grace behind it, or some other extra human principle, as we have no idea how to go about consciously designing a reality to work that way.

Probably there's a sense of keeping a safe distance from what might get ugly, that pervades many disciplines.  Altercations you don't want to have to witness tend to drive behavior.  The train wrecks may never happen, and perhaps our avoidance maneuvers were actually constructive?