Aixiom member wins the IvI Grand Challenge

NewsPublished October 19, 2010 at 1:13 PM No Comments

The IvI (Institute for Informatics, University of Amsterdam) challenge consisted in defining the holy grail of computer science.

Here are our winning thoughts:

Computer sciences, or informatics, has globally become a synonymous of automation and communication. Machines, factories, traffic, clouds, cars, to spam filters, medical equipment etc. are automatically controlled/monitored by computers, making this relatively new science very successful and popular.
The question “What is the holy grail of Informatics?” could be rephrased in: where are we going with computer science? what is its ultimate goal? what will be the discovery that will actually put an end to computer science? The answer is scarily simple and immediate: Intelligence.

Although this is a clear answer, the problem is that there is no clear definition of intelligence, and specifically there is no clear direction which points towards the way to achieve this.
Nowadays computer software can do many things, even better than humans in certain tasks, but they still cannot be called intelligent: Is the software which can build a car intelligent? and the one that can play chess? and the one that can recognize your facial expressions?
This is because solving problems with computer science is generally achieved with hard coded or learned rules.
Currently, the paradigms used in computer science can be compared to instincts, rather than intelligence, thus following a basic principle: “if this condition occur, then do this operation”.
Using “if you are hungry, then eat” as an example, it is clear that the main problem of this approach is that each answer generates more questions which need to be modeled: “how do i eat?”. This could be answered with “go to the refrigerator and cook something” or “grab a spear and go hunting”. But how to do that? Each question generates more questions, which needs to be successfully answered to achieve the first task.
The the probability theory paradigm shift in the 90′s which moved from hardcoding this rules and using logic to do some reasoning to learning what is the most probable outcome, brought success stories for Artificial Intelligence. However, although very powerful, probability just hides the learned set of rules, hence being subject to the same problems of “instinctive” software.
It is clear that another paradigm shift is needed: A way to reuse information.

For instance, millions of pages about a single subject are written every day on the internet, but they cannot be summarized automatically into a single book. Digital life forms learn from making mistakes, and only the procedure/rule which is fit survives and evolves. However, once a pair of robotic legs have learned how to stand up from the ground, the process of learning how to walk needs to be learned again from scratches. Without human intervention no information about how to achieve the first task and start from there can be brought forward in the process.
In the animal kingdom, or “intelligent” kingdom, the re-use of information is key. We reuse concepts that we learn, It is at the basis of science society: we are not “reinventing the wheel” all the time.
Therefore, in my opinion, the holy grail of informatics (“informatics” standing for “Automatic Information Processing”) is finding an answer to the question “how can information be automatically reused?”.


Share|

Sorry, the comment form is closed at this time.