Artificial Intelligence and the Web

Creating artificial intelligence or an artificial mind has been a preoccupation of computer science (and science fiction) for decades, since the concept was introduced by scientists such as Alan Turing and John McCarthy. The original idea was built on the assumption that the human brain functioned like a machine – a body made up of small working components which individually were not conscious but together make a mind. Cartesian philosophy also provided tentative encouragement, as to Descartes being a person meant simply to be “a thing that thinks”, as well as which doubts, perceives, affirms, denies, wills, does not will, that imagines also, and which feels” – an array of qualifiers which largely translate to the kind of stimulus responses which machines are perfectly capable of.

However the belief that this could translate simply into machine code has proven rather naive, for various reasons. Humans are capable of not just storing massive amounts of information, but are able to retrieve and use it “with remarkable efficiency” (Brookshear, 2012: 485). Although storing data is not a huge challenge for machines, understanding it as information in a wider context (as ‘knowledge)’ is. This brings us to the Web, and a more nuanced understanding of intelligence. The Web, and to a greater extent the Semantic Web, gives computer scientists the opportunity to work with a vast store of connected, contextualised information. Results derived from using this network do not attempt to mimic the human brain, but instead create results that are similar to human abilities, and yet are achieved in a different way (a constructionist rather than mimetic understanding of intelligence). The real and achievable aim with this is not to be able emulate what a human do in a given situation, but rather to solve a problem as well as possible.

Angie M says:

December 21, 2013 at 12:49 pm

First visit to this blog.
It is not likely that ‘stimulus/response ‘ will be part of a definition of AI that we can all accept. We want something more than a constructionist approach – maybe in between constructionist and the Asimov imagined robot with self-awareness and a positronic brain.

Machines that can learn, currently, only do so by means which have been programmed into them – ie they do not learn of themselves, though they may extensively use pattern recognition, inferencing, and ‘appear to learn’. They do not have an independent decision-making capability, (eg let’s go for coffee).

BUT AI might be recognised as a combination of people and machines, initially unaware, with the capacity to learn and affect the environment, with the ability to act as a unified entity. Eventually we might have AI sub-divided into categories, eg this is ‘group AI’ – this is ‘individual AI’ etc

The ‘knowledge’ we are seeking from the Web is put there in the first place by people, either directly, or as data which is processed and refined within the context provided. So maybe this ‘knowledge processing’ is a precursor to the development of AI.

Is this vision way off the mark – because it is a result of doing this MOOC!