Emergent weirdness in AI

Daniel Kahneman on ‘Emergent Weirdness’ in Artifical Intelligences
Nov 28 2011
http://www.theatlantic.com/technology/archive/2011/11/daniel-kahneman-on-emergent-weirdness-in-artifical-intelligences/249125

Emergent weirdness is a good bet.
Only deduction is certain.
Whenever an inductive short-cut is applied, you can search for cases in which it will fail.
It is always useful to ask “What relevant factors are not considered?” and “What irrelevant factors affect the conclusions?”
By their very nature, heuristic shortcuts will produce biases, and that is true for both humans and artificial intelligence, but the heuristics of AI are not necessarily the human ones.

heuristics

related:

December 7, 2011
https://franzcalvo.wordpress.com/2015/03/09/twitters-trending-algorithm

http://www.nytimes.com/2015/07/10/upshot/when-algorithms-discriminate.html

bvt123

Jürgen Schmidhuber at TEDxLausanne

When creative machines overtake man: Jürgen Schmidhuber at TEDxLausanne
Mar 10, 2012
https://www.youtube.com/watch?v=KQ35zNlyG-o

http://www.tedxlausanne.org
Machine intelligence is improving rapidly, to the point that the scientist of the future may not even be human! In fact, in more and more fields, learning machines are already outperforming humans.

Artificial intelligence exper t Jürgen Schmidhuber isn’t able to predict the future accurately, but he explains how machines are getting creative, why 40’000 years of Homo sapiens-dominated history are about to end soon, and how we can try to make the best of what lies ahead.

7:15 as you are interacting with the world, you are observing more and more data.
All the time, your brain is trying to find novel, unknown regularities in the data, trying to make sense of the world, trying to be a better predictor. … encode the data.
You can measure any novel regularity as follows: before you have discovered the regularities through some sort of learning algorithm you need so many computational resources to encode the data.
After having discovered the regularities, you need less computational resources, because any regularities mean that you can save computational resources such as synapses or computation time.
You can measure that, that’s a real number.

There are multiple grand truths

Structure Data 2014: Democratizing Artificial Intelligence with APIs
https://www.youtube.com/watch?v=iHVeoJBtoIM
Apr 11, 2014
If you think Siri is smart, you haven’t seen anything yet. Hear how APIs are putting cutting-edge capabilities in deep learning, cognitive computing and artificial intelligence into the hands of developers everywhere.
Elliot Turner — CEO, AlchemyAPI
Stephen Gold — VP, WW Marketing and Sales Operations, Watson Solutions, IBM Software Group

00:40 cognition as a service
02:05 programatic computing
04:00 in congnitive systems, there is no absolute outcome
05:00 the grand truth
05:40 context is key. You can have multiple grand truths, depending on the situation. Vampires are real in certain contexts, but they’re not in others.

“… the data, which in the end would form the ‘grand truth’ that Watson would regard as baseline facts.”
http://cornellsun.com/blog/2012/10/24/hello-watson-students-design-tech-support-program-from-jeopardy-super-computer-watson

There seems to be no term in the NN literature for the set of all cases that you want to be able to generalize to. Statisticians call this set the “population”. Tsypkin (1971) called it the “grand truth distribution,” but this term has never caught on.
Tsypkin, Y. (1971), Adaptation and Learning in Automatic Systems, NY: Academic Press.
http://www.developpez.net/forums/d1216717/general-developpement/algorithme-mathematiques/algorithmes/intelligence-artificielle/rdn-methodes-d-evaluation-performances

the notion of context as a critical feature to understanding anything. Bateson (1979) was adamant in making the point that “without context, words and actions have no meaning at all” (p. 16).
Words and actions need to be situated in one or more relevant and meaningful contexts, in order to develop any degree of complex understandings.
The use of metapatterns for research into complex systems of teaching, learning, and schooling. Part II: Applications.
Bloom, Jeffrey W., & Volk, Tyler. (2007).
Complicity: An International Journal of Complexity and Education. 4(1): 45—68.
http://internationalbatesoninstitute.wikidot.com/jrnlarticles:2

Turing Test: passed

Do Feelings Compute? If Not, The Turing Test Doesn’t Mean Much
by Geoff Nunberg
July 01, 2014
http://www.npr.org/blogs/alltechconsidered/2014/07/01/323984864/do-feelings-compute-if-not-the-turing-test-doesnt-mean-much

At an event held at the Royal Society in London, for the first time ever, a computer passed the Turing Test, which is widely taken as the benchmark for saying a machine is engaging in intelligent thought.
But like the other much-hyped triumphs of artificial intelligence, this one wasn’t quite what it appeared.
Computers can do things that seem quintessentially human, but they usually take a different path to get there.
IBM’s Deep Blue mastered chess not by refining its intuitions but by evaluating hundreds of millions of positions per second.
Watson won at Jeopardy not by wide reading but by swallowing all of Wikipedia

related:
The Turing Test Is Not What You Think It Is
by Alva Noë
June 13, 2014
http://www.npr.org/blogs/13.7/2014/06/13/321755270/the-turing-test-is-not-what-you-think-it-is

https://franzcalvo.wordpress.com/2015/06/10/robots-cognitive-abilities-of-a-two-year-old-child

Turing Tests in Creative Arts
http://www.npr.org/sections/alltechconsidered/2015/08/07/429084124/shall-i-compare-thee-to-an-algorithm-turing-test-gets-a-creative-twist
http://bregman.dartmouth.edu/turingtests

Anticipatory computing

Computers That Know What You Need, Before You Ask
March 17, 2014
http://www.npr.org/blogs/alltechconsidered/2014/03/17/290125070/computers-that-know-what-you-need-before-you-ask

artificial intelligence is getting even smarter.
The next wave of behavior-changing computing is a technology called anticipatory computing — systems that learn to predict what you need, even before you ask.

Science and Fiction – A Springer Series

Science and Fiction – A Springer Series
2014
http://www.springer.com/series/11657

Readers can look forward to a broad range of topics, as intriguing as they are important. Here just a few by way of illustration:
Time travel, superluminal travel, wormholes, teleportation
• Extraterrestrial intelligence and alien civilizations
Artificial intelligence, planetary brains, the universe as a computer, simulated worlds
• Non-anthropocentric viewpoints
Synthetic biology, genetic engineering, developing nanotechnologies
Eco/infrastructure/meteorite-impact disaster scenarios
• Future scenarios, transhumanism, posthumanism, intelligence explosion
• Virtual worlds, cyberspace dramas
Consciousness and mind manipulation