Pragmatics in Praxis

Image

This morning, I read a New Yorker article on A.I. entitled “Why Can’t My Computer Understand Me?”  It’s worth a read.  The article’s protagonist, Hector Levesque, denounces the Turing Test as too easy to scam.

I agree…with the proviso that, in the development of useful expert systems, we’ve reached a historic plateau in which, for business purposes, a useful metric is:  “Time to Turing-Complete” (TTTC).

My thinking on general AI still orbits a praxis-to-pragmatics approach, as opposed to development of highly specific algorithms that remain in the realm of mere semiotics or semantics:  (e.g. Explicit / Latent Semantic Analysis, Cluster analysis, Inverse Word-Frequency Analysis, HMM, etc.; e.g. Google search, Google Knowledge Graph, Evi, Siri, Wolfram Alpha(?), etc.)

However, lately I’ve been pondering a radical pragmatic expansion of Dedre Gentner’s “ad hoc categories.”  A popular stock example of an ad hoc category would be “Things you’d grab from your house in a fire.”  (Of course, life is always even more ad hoc:  “Things you’d grab from your house if there was a fire in the kitchen and you knew you had at least two minutes, but probably not five.”)

The radical pragmatic expansion is prompted by meditation on the social.

In every social system we engage, we generate an entire Gestalt, ad hoc, fabric of meaning (e.g. shared meanings, shared allusions, private codes, inside jokes, et al).  It’s as if there’s a pragmatic “terroir” to our everyday actions (e.g. My girlfriend appreciates the subtle inflections of what it means for me to do dishes these days, given my current projects.  On another level of granularity, every time I do dishes, I use an ad hoc cognitive map of which regularly-used bowls in our apartment fit inside other bowls).  In a social context, ad hoc categories are the rule, not the exception.  We live a social tapestry of ad hoc categories, an ad hoc cognitive tapestry.

To get what I mean by “pragmatics”, a concept as simple as J.L. Austin’s “performative utterance” suffices as an initial springboard: “By saying X, I hereby do Y.”  E.g. “By saying ‘I do,’ I hereby commit myself.” But Austin cared about “how to do things with words.”  Praxis approaches pragmatics from the action side rather than  the semantics side.  Thus, I envision a sort of socially-aware “performative activity” / “performative agency”:  when J does X in context Y, it means Z to M.  How to signify things with actions.

For General AI, then, one requires:

– Machine Learning
– Basic self-awareness (can represent and manipulate its own code) **not strictly necessary, but super cool…and perhaps easier to code.
– Social awareness & social self-awareness (awareness of oneself as a social agent among other social agents)
– Event ontology – Event matrix, Causality matrix, Pragmatic matrix (notion that every event derives meaning from social fabric)
– Rules for principled norm-keeping & norm-breaking
– Multi-modal & cross-modal representation paradigms (requires at least two sensors…e.g. audio, visual, text)
– Socially engaged experience
– Abstraction to rules from particular experiences, integrated with a
– Categorical ecology (continually updated “ontology”) derived from the social realm (others in this situation, do X, mean Y, etc.).

For the AI envisioned by the New Yorker article (let’s call it “Alligator-AI”) you need much less (for an initial prototype):

– Machine Learning
– A general pragmatic ontology (including all relevant facts about, say, an alligator…like its body plan)
– Precise grammatical parsing (proliferate potential grammatical models, then use a semantics parser / neural net to narrow down to a frame)
– The ability to invoke an answer-frame appropriate to the question-frame (Alligators can’t run 100M hurdles. Gazelles, on the other hand….)

…or we could just rest on our laurels with the accomplishment of AI in Twitterbots with the same satisfaction as if we’d just built the Great Pyramid.

Advertisements

Slices & Traces

Slices & Traces
In graduate school, I once heard a medieval scholar remark that we now knew what Thomas Aquinas did on nearly every day of his life.  While such a feat is perhaps the wet dream of a medievalist, technology is reaching the point where the same may soon be true of me or you.

Historians compile numerous traces (any historical artifact that says “Thomas was here”) into slices (e.g. a biography).  In the digital age, what fascinates me is that numerous ready-made slices of our virtual lives may be compiled easily from databases that archive massive amounts of our personal digital traces.

I recently had the opportunity to experiment with Stanford’s Muse Project, which provides various analyses and visualizations based on personal email history and browser history (sentiment analysis, social group change over time, et al.)  Pros:  the program provides an interesting slice of one’s virtual self.  Cons:  the slice of my personal history recorded in my email database feels partial and one-sided.[1]  For another example, consider Facebook’s recently released “timeline.”  The history embedded in your Facebook timeline is yet another  slice of your personal history.  Each slice tells its own story, albeit an incomplete story.  A slice is just a slice.

What if, like an fMRI, we were able to capture and compile slice upon slice?[2]  Would the slices add up to a complete picture?[3]  What if one were to aggregate and integrate all the slices of one’s virtual life?  What if you had the tools to capture & integrate your own personal data from email history and browser history and add that to your data from social networks (Facebook, Twitter, LinkedIn), dating sites (eHarmony, Match, OKCupid), bookmarking sites (Stumbleupon, del.i.ci.ous, digg), music sites (Pandora, Grooveshark), movie sites (Netflix, Blockbuster), video sites (Youtube, Vimeo, Dailymotion), commerce sites (Amazon, ebay), banking sites (Mint, Quicken), location services (FourSquare, GPS), SMS history, and blog corpus?  What if, to that already rich textual and social data, one added perceptual data capture via webcams, haptics, and EEG/GSR?  What if one were to sift, analyze and integrate the data using textual algorithms (corpus linguistics, LSA, ESA, sentiment analysis), social algorithms (network & influence analysis), and perceptual algorithms — replete with visual recognition (facial, gesture, object, movement), audio recognition (voice, music, sound), and touch recognition (texture, heat, pressure)?  (See Figure 1)

Slicing & Tracing
Such a system of integrated personal data, collected en masse (even if anonymized), would prove invaluable to social scientists, historians, marketers, Big Brothers, and researchers of all ilks.  Although we’d never achieve Rankean history “wie es eigenlich gewesen ist,” (as it actually happened) through such a system, it represents a potential tool (among other tools we’re developing) that will soon get us closer to historical realism (or even hyper-realism).  What I’d like to discuss today is not the fine-grain detail we may someday achieve by integrating slices and traces.  Instead, today I want to talk about the slicing and tracing.

Suppose you mummify your information…all of your information.[4]  You’re still just a data-fossil in a museum exhibit a millennium from now (and if everyone gets mummified, probably a poorly-visited exhibit).  But your data doesn’t even make it to the museum without first undergoing some form of condensation and selection.[5]  I don’t care how much you love your grandpa, you’re not going use your entire life to watch a second-by-second video of his entire life.

Before the digital age, condensation and selection happened naturally in places like family photo-albums and dinner-table stories.  These human-sized brain-morsels could be chewed and digested comfortably.  In the digital age, a deluge of data makes you cross-eyed and bloated while historians babble about Kim Kardashian and advertisers hypnotize you with french fries.  As we speak, historiography is being asked to develop some frighteningly powerful tools to condense uncompressed information, select salient aspects, and present us with soundbites (Think Robin Williams in The Final Cut).  Too much data is the first challenge facing next-gen story-telling gurus.

But too much information (TMI) is merely the prima facie challenge.  The real challenge, as I see it, is not TMI but too little intelligence.  I’ve often said that “after the Information Age comes the Intelligence Age.”  I want to see a generation of “intelligence scientists” rise up to replace today’s “information scientists.”  Would you rather preserve your intelligence (creativity, intuition) or your information?[6]  What would that even look like?

____________

FIGURE 1:

In the spirit of Aristotle and Nietzsche, I’ve nicknamed the data-integration algorithm-hub “VirtuAlly.”

FOOTNOTES:

[1] Also, the sentiment analysis engine in Muse is amateurish.

[2] The current discussion assumes that the capture, aggregation and integration of data would be for private and personal use only.  With increasing sousveillance, each of us may be able to compile an increasingly complete picture of our personal histories.  As technologies for personal data capture, aggregation, and integration progress, the following philosophical stance will also snowball in importance:  an individual’s data is his or her inalienable property.

[3] Temporality is a dimension common to each of the following data slices.  Each slice is like a layer of bedrock, and data archived in each aggregates many fossilized traces of one’s virtual life.  Time-stamps are common in each digital trace, making chronological sorting easy.  Who will standardize the aggregation and integration of these slices, as we once standardized the USB port?

[4] Lifenaut.com offers a digital (and biological!) time-capsule for would-be immortality-seekers.

[5] By condensation I mean something like summary, and by selection I roughly mean meme-discrimination.

[6] Arguably, neither is any good without the other, so my answer is “both.”

RECOMMENDED:

DATA AGGREGATION & VISUALIZATION INFO:

http://learning.blogs.nytimes.com/2012/02/07/what-story-does-your-personal-data-tell/#
What story does your data tell?

http://www.ted.com/talks/jer_thorp_make_data_more_human.html
New York Times data analyst on visualization

DATA COLLECTION & AGGREGATION TOOLS:

http://mobisocial.stanford.edu/muse/
Creates a slice of your personal history using your EMAIL, with capabilities for BROWSER HISTORY (best in Firefox).  The program runs securely on your local machine, so there’s no chance your data will make it to the cloud.  I’ve experimented with this program with interesting results.

https://openpaths.cc/
Capture your LOCATION DATA.

https://chrome.google.com/webstore/detail/gdfhmgiphpkoodfifkmiecmcmlmmhaip
Capture BROWSER HISTORY.  a friend of mine built this.

LIFE CASTING INFO:

http://en.wikipedia.org/wiki/Lifecasting_(video_stream)
http://www.justin.tv/

LIFE CASTING / DIGITAL AUTOBIOGRAPHY LOCKER TOOLS:

www.lifenaut.com
Interactive time capsule, digital self-storage space (digital locker)

https://www.lifenaut.com/learn-more-bio/
Also, they store your DNA…free (suggested donation $399)