Image Image Image Image Image Image Image Image Image Image

Bartlett School of Architecture, UCL

Scroll to top

Top

No Comments

Learning as a defining element of Artificial Intelligence

Learning as a defining element of Artificial Intelligence

| On 28, Apr 2015

For more than 50 years, robotics and scientists have put their effort into developing artificial intelligence, and research into AI has been done to a very high level. However, intelligence has always been one of those concepts that is very difficult to define. Let’s just take a step back and reconsider that what’s required to be intelligent?

In 1950, Alan Turing published a famous article in the journal Mind, he originally put forward a simple behaviorist definition of intelligence. And Turing Test, as it is known, formulates the issue of whether  a machine is intelligent in terms of whether it could pass the following test:

Turing and Turing test

 

The game is played with a computer(A), a human(B), and an interrogator(C). The interrogator stays in a room apart from A & B, then asks questions of them. The objective of the interrogator is to determine which one is the human while the objective of both A & B is to convince the interrogator that he/she is the human and the other is not.

If a computer passes the test, it will necessarily be intelligent. That means intelligence is not the character of the object it self, but is attributed to the object by an observer based on his judgment.

However, that’s just a general idea of who we think is intelligent, it ignored one of the key elements of human intelligence, that is the ability to learn and improve the performance of himself. Let’s take wasp as an example:

wasp

 

A wasp’s routine is to fly outside for food, bring it back to the burrow, leave it at the entrance, go inside to check that  everything is well, come out, and then drag the food in. If we move the food a few inches away while it is inside, the wasp, on emerging from the burrow, will drag the food back to the threshold, but not inside, and will then repeat this procedure of entering the burrow to check if there is any intruders. The wasp never thinks of pulling the food straight in. It blindly follows a fixed response pattern and fails to take change into account.

This example shows the importance of memory and the application of past memory to a current situation. On the other hand, learning from others is also a prove of intelligence.

In 2004, Canadian artist Marc Böhlen and J.T. Rinker created UWM (Universal Whistling Machine).  The artists, interested in whistling as an ancient form of communication. As someone whistles a tune into the machine, it will learn by recording and response similar but not the same melody.

two uwms

 

 

Conceptually, it is a witty take on artificial intelligence, referencing Turing’s computing machinery and intelligence. Difference in this interaction is vital, because it suggests that the machine does not simply record the whistle, but managed to learn by understanding and response intelligently.

In 2014, Korean designer Minsu Kim created “The Illusion of Life”, a machine that can even imitates the gentlest breaths and lowest vocal tones of our private conversations. Kim hopes to stress the intimacy that seems to be one of the few irreplaceable facets to remain uniquely human.

the illusion of life

 

the illusion of life 2

Actually, intelligence does not necessarily mean to be human or life-like. But this could suggest a  new form of human-machine exchanges. I’d like to explore more of machine learning in my project that responses to dancer’s touch intensity in realtime performance.

 

References:

1. Ayse Pinar Saygin, Ilyas Cicekli & Varol Akman: ‘Turing Test: 50 Years Later’ in The Turing Test (James H. Moor ends.), 2003, Dordrecht, Kluwer Academic, pp. 23-33.

2. Jay Friedenberg: ‘Intelligence’ in ‘Artificial Psychology’, 2008, New York: Psychology, pp. 109-126.

3. Rodney A. Brooks: ‘Elephants Don’t Play Chess’ in ‘Cambrian Intelligence: The Early History of the New AI’, 1999, London: MIT, pp. 111-131.

4. Penny, Simon: ‘Twenty Years of Artificial Life Art’ Digital Creativity, 2010, Vol. 21 (3), pp. 197-204.

5. JT Rinker: ‘Two UWMs’ (http://www.realtechsupport.org/imgs/two_uwms.jpg, 27 February, 2015).

6. Jeremy Hance: ‘Insect intelligence: paper wasps display strong long-term memory’ (http://news.mongabay.com/2008/1201-hance_insects.html, 12 August, 2004).

7. Beckett Mufson: ‘The Illusion of Life’ (http://thecreatorsproject.vice.com/blog/the-illusion-of-life-machine-can-mimic-human-whispers, 15 April, 2014).

 

 

 

 

 

Submit a Comment