Beyond Turing?

One question that been bubbling up is the Turing-Church theorem, and its potential implied limitation on the possibility of machine intelligence (i.e. the limits of logic and computation). The prospect of interactive computation escaping this limitation seems like a particularly promising thought, one which aligns nicely with some of the thinking Walter and I have been doing on what we are referring to as the “bidirectionality” of abstraction in the mind.

Here Jeff Hawkins (founder of Palm, Treo, and now Numenta) suggest in the title of the below lecture that his HTM (Hierarchical Temporal Memory) machines can “Computing beyond Turing.” One has to be a bit skeptical of whether we are really getting beyond Turing here. As the final questioner points out, this whole thing is running on a Turing machine, after all. The name itself jumps out – can a fundamentally hierarchical system ever go beyond Turing, or are we once again turning the world into so many trees (in Christopher Alexander’s sense of the word)? In fact, as Hawkins admits near the end of the lecture, his system is in fact less general than the Turing machine – a hierarchical system can only solve hierarchical problems – it is all nails to a hammer. Certainly the human mind has the capacity for a whole range of non-hierarchical “leaps”. It does open an interesting set of questions about the “inherently hierarchical organization of the world” as no doubt many key problems (at least one that can be monetized in a straightforward manner) do tend towards hierarchical characteristics, but again, we must wonder to what extent this is due our own projections into the world (and the projections of our long history of rationalization machines – from money to maps to machines). But in any case, Hawkins certainly brings some interesting approaches, and I will definitely explore the software in so far as it is available.

One side note, just noticed as I enter into this world of Neuroscience + AI research – the characters are quite an interesting bunch. Brilliant no doubt, and all quite polyglot, as you would expect. But particular as well. From “Theme Park” to AI…sure why not? From Palm Pilot to AI… what else would you do…? It does beg the question: is AI research the new-age, post-dot com success mid-life crisis dream job de jour? And for that matter, what is the impact of the business-driven approaches that underpin many of these efforts at realizing AI. How does this monetization of the technology impact was is produced?

Beyond the obvious TED Talkers, it will be interesting to dig into some people who are operating at a bit more of a theoretical level, whatever that might mean.

A few more videos that may (or may not) shed some light on the Turing incompleteness limitations:

 

 

 

G.E.B.

Along with the Glass Bead Game, Douglas Hofstadter’s classic book Godel, Escher, Bach represents a key point of departure for this thinking about the fundamental properties of mapping, both as a mental construct, and as understood in and through the city. In particular, the possibility suggested by this book that any formal system of sufficient complexity has the hypothetical capacity to exhibit some form of  self-awareness (self-mapping). This runs counter to the suggestion that Godel and Turing’s incompleteness theorems undermine the possibility of machine sentience. For Hofstadter, the secrete of the “recursive I” arises precisely out of the inescapable incompleteness of the system (which is itself rooted in the recursive nature of that system), it is not some magical, unique property of the brain. I need to revisit the concepts here, but my hunch is that this kind of representation of self-contradiction is necessary but insufficient cause of sentience. A formal system understood in itself, outside of interactivity with a larger environment, can never make the jump out of itself.  Recent thinking in interactive computation point in this direction.

Here is a series of courses organized for high school students by some MIT undergrads, really well done and thoughtful:

and a couple of lectures by Hofstadter himself, including this gem from the Stanford Singularity Conference in 2011:

and as mentioned by Douglas’ talk, David Cope’s Emmy algorithmic music generator:

 

Glass Bead Games

I had the chance to read this great interview of Keller Easterling and Benedict Singleton by the editors of The Glass Bead, a fantastic site in itself, and a great reference for the “mapspace as gamespace” project now in conception – this page in particular is a great articulation of the idea. I am familiar with Easterling and her work on the spatial software of the city, but Singleton was new to me. I will be digging more into his work. His unpacking of plot, plotting, the plot, plot twists, etc. here is great:

If you trace the conceptual history of ‘plot’, you find that before around 1500, the term refers solely to a marked-out site, an area of land. Over the next century or so, the term’s meanings proliferate to the point where their connections are no longer immediately obvious: drawings, narratives, and seditious plans are all called plots. The underlying logic that guided this development illuminates an alternative conception of design in a very striking way.

Plot’s initial, spatial meaning, the demarcation of an area, transferred into the language of the workshop. One plots out a design on paper before acting on other, more expensive materials. So a constructive sense of plot arises, relating to diagrams, maps and charts. And within a few decades, this graphical ‘plotting’ was adopted into the lexicon of the early modern theater, where its artisanal meaning deepened into a narrative sense: plotting as the arrangement of people and things over time, so as to tell a story.

Up to this point, ‘plot’ shares a substantial similarity to ‘plan’. Both words couple the idea of a spatial arrangement with a schedule of unfolding action. Plot’s connection to territory (and the politics of its division), cartography, and stories make it, perhaps, the richer word. But most interesting is that, on the back of its theatrical use, plot acquired a further, specificallysubversive, sense, which planning does not possess: plotting as the subtle orchestrations of an unseen director, manipulating the course of events from behind the scenes.

So ‘plot’ encodes a particular form of creativity, too, which can be glossed as the production of a plot twist. This is the point at which one plot is subverted by another one, just as the routines of the bank, the placement of cameras, the structure of the vault and the peccadilloes of the manager become the raw material of the heist. Put another way, plotting is always re-plotting: discerning the contours of an unfolding situation and locating the opportunities it presents for ‘leverage’–points in space and time at which an action can generate an effect disproportionate to the physical effort put into it. A plot, we might say, is a plan invested with this kind of underdog intelligence.

In a kind of closing of the circle, ‘site’ (the original meaning of plot) remains critical to this idea of the creative twist or what we might call the kick–the moment where one plot is derailed by another. Rather than conjuring an image of how the world should be and then trying to force it into being, plotting takes a site’s particular structure, its fixity or at least predictability, as the platform for new and potentially unlicensed operations. Recovering the full sense of plotting, as an intervention that starts from a point of comparative weakness and proceeds through guile and ingenuity, forges a deep conceptual link between the creation of artifacts and political intrigues, dissident stratagems, and other ruses.

some vids:

Interactive Computing

A nice little rabbit hole here. What happens to Turing incompleteness (and the argument that it renders machine intelligence impossible) once machines start interacting directly with the world? Finite State Machine (and non-determinant FSM’s) also seem quite useful going forward.

From Wikipedia:

The famous Church-Turing thesis attempts to define computation and computability in terms of Turing machines. However the Turing machine model only provides an answer to the question of what computability of functions means and, with interactive tasks not always being reducible to functions, it fails to capture our broader intuition of computation and computability. While this fact was admitted by Alan Turing himself, it was not until recently that the theoretical computer science community realized the necessity to define adequate mathematical models of interactive computation. Among the currently studied mathematical models of computation that attempt to capture interaction are Japaridze’s hard- and easy-play machines elaborated within the framework of computability logic, Goldin’s persistent Turing machines, and Gurevich’sabstract state machines. Peter Wegner has additionally done a great deal of work on this area of computer science.

Douglas Engelbart, father of interactive computing.

Theory of Computation course at Portland State (FSM, etc.):

http://web.cecs.pdx.edu/~harry/TheoryOfComp/index.html

 

The Chinese Room

John Searle offers a contrasting point of view to the previous post and its “solve AI and you solve the problems of the world AI” optimism. His Chinese Room thought experiment is a powerful critique of strong AI. There is something wonderfully Borges-esque about Searle’s imagery in this, something akin to the Library of Babel, as one mechanically shifts from input, to database, to code, to output, and the baroque infinitude of such an undertaking. But what about boredom in this, the human (conscious) impulse to introduce humor, playfulness, creativity into the most repetitive tasks?

In this video of a recent talk at Google, Searle expounds on the Chinese Room, as well as his theories on epidemiological- vs. ontological – subjectivity and its critique of the possibility of machine consciousness as a category error. I will need to think about this a bit more, though we might ask, as Hassabis points out, if the question of machine consciousness (at least one predicated on the human/mammalian brain and experience of consciousness) is even an appropriate question for the development of AI.

Searle, a key philosopher in the analytical tradition, is also famous for his not-so-friendly exchange with Derrida over speech-act theory.

Demis Hassabis + DeepMind

Some very interesting emerging work from the leader of DeepMind, now part of Google, and its research into AI, Deep Learning. His incredible background goes from chess master at the age of 13, design of the video game Theme Park at 17, a PhD in neuroscience, and finally the founding of DeepMind and its purchase by Google in 2014. DeepMind recently made the news for its AlphaGo program, which beat the European Go champion. A couple recent lectures:

 

Thinking Skin

Some great links on the biology, cognition, and behavior of octopus, cuttlefish, and other cephalopods, including the incredible morphology of their arms and skin, and attendant development of their highly distributed neurological system (same number of neurons as a cat, but 2/3 distributed throughout the arms) and associated behavior (mimicry, camouflage, mesmerization, and other nomadic/”soft” tactics).

CuttlefishSkin2

Marine biologist describing the mechanisms of octopus skin coloration. VERY USEFUL

Great video of octopus and the beer bottle – squeezing and adaptive coloration. VERY USEFUL

Overview of octopus morphology and intelligence. USEFUL, esp. at 3:40

Great overview of octopus neurology.

Arduino Basics (Automated Inflation)

(The first foray into Arduino programming. Pretty much following built-in examples for driving DC motors and reading analog sensors (here a IR distance sensor). The fan here was nowhere near powerful enough to inflate even a small garbage bag, but I think the principle is useful, and hopefully this is the start of a series of experiments.

And a couple useful links for future development:

http://www.instructables.com/id/RE-Inflatable-Vest/?ALLSTEPS

http://www.instructables.com/id/Use-Arduino-with-TIP120-transistor-to-control-moto/

 

Hutong Agency

This is a project was developed by student Dongni Lu as part of the Fluid Spaces studio by Profs Sheng Qiang and myself at Tianjin University School of Architecture in the Spring of 2012. It explores the evolving relationship between the production of space and the organization of social activity in the hutong alleyways of Tianjin, China.  This project simulates the process by which these alleyway streets are shaped through the encroachment and decay of house boundary walls into the public space of the street. The patterns of occupation of the public space and the highly localized street topology serve to define one-another through this historic process of accretion. Overall the system seeks a dynamic equilibrium between the conflicting desires for the expansion of private space vs. the need for the constitution of a functional public domain for circulation, social gather, commerce, and play.

hutongAgency

Dynamic Bus Routing

Examples of the simulation of a dynamic bus routing system running on the streets of Chicago. Work completed by Profs. Thomas Kearns + Jordan Kanter and students Adam Weissert, Haidong Fei, and Li Gong as part of the Urban Data Model Prototype Studio at IIT in 2013-2014.

image135
whiteboard systems diagram: interaction of bus, rider, and street network objects in simulation
image136
bus routing diagram: indicating the process of describing catchment zones based on an updated voronoi diagram of current bus route to determine which bus in the system to assign a new pick-up to and thus rerouting the buses
image137
process for generating riders (pickup location + destination) based on analysis of demographic data

 

image138
process of assigning weighted lengths for each edge in the street graph based on various city data and spatial inputs: number of street trees, sidewalk widths, pothole complaints, expected ridership, etc.