Creativity and Artificial Intelligence: a conceptual blending approach

About the book

"Creativity and AI: A Conceptual Blending Approach" is the result of several years of research I took in the area of Computational Creativity. Personally, it means the Masters (2000), the PhD (2005) and some more recent work I did in 2006, all under the supervision of Prof. Amilcar Cardoso and within the context of the Computational Creativity Lab of the CMS group in CISUC (if you're interested in understanding the structure of our research center, please follow the link).

Computational Creativity

During those years, the area of Computational Creativity grew considerably from a "tiny" workshop in AISB'99 to the standalone event that we have today. The size of the community has grown considerably since then (although it has still much more to grow, in the same way that the area itself has plenty of "loose ends" to explore) and I was happy to know many of the people that were and still are the leading players in area, not only for being outstanding researchers but also because they are very creative themselves!...

The book thus contains my personal path, frequently influenced by others, but also reflects many of the central topics of discussion we all had. It is also clearly a dissertation about Computational Creativity, with its many lively discussions. I am confident that my book is of value to those interested in knowing what Computational Creativity is about and where to start in looking for promising research topics of the area.

Artificial Intelligence

My background is AI, my Master's thesis was related to symbolic machine learning (in the form of Example Based Learning, Inductive Logic Programming, Ontologies...) and the PhD applied also semantic networks and genetic algorithms. In technical terms, the reader doesn't have to really know this area (neither the book is a good introduction to it, anyway), but those who are aware of it will certainly identify many issues that are common in any AI project, such as Knowledge Representation and Search Strategies. The book is naturally biased towards this AI perspective, and particularly to the dilemmas around Computers and Machines (can a computer really be creative? When can we say it is and it is not? Can we say we have AI without Creativity?).

I tried not to make it too technical (for those who really want to get into it, the CD has the code of the system, and in the "extras" section, I list some of the latest work) and to make it a good read.

Conceptual Blending

Not having a Cognitive Linguistics background myself in any way, I did take strong effort to understand the area, to find the links that could connect my intuitions on Creativity (and Language, in fact) with my formal background of AI and Computer Science. I did find what I was looking for in the Conceptual Blending Framework, of Gilles Fauconnier and Mark Turner. That became, at some point, the leading trail of the work (and eventually of the book).

Curiously enough, I didn't find many links between AI and Cognitive Linguistics, which seems quite odd if we consider the apparent affinities (Language, Reasoning, Communication, Evolution, Conceptualization, Memory,...). I hope that the book becomes a real contribution for filling this gap.

Its appearence in the "Applications of Cognitive Linguistics" series of Mouton de Gruyter is quite adequate in scope and context. It is indeed an example of an application of work produced within the Cognitive Linguistics community. In being so, I believe that it can serve as an example of how to "apply" research ideas from an area that sometimes is criticized for not being easily "applicable".

...well, if you're curious to know more, there is only one way... You have to get it!

"The book involves an in depth journey towards finding a computationally plausible model of creativity. Inspired by a multidisciplinary analysis of work on creativity, the author elaborates on what kinds of processes, principles and representations could lie underneath the act of creativity, eventually proposing a model based on the conceptual blending framework, from Fauconnier and Turner. Written by a researcher with a background in artificial intelligence, the book is the result of several years of explorations and discussions that cover several viewpoints. The author presents a top-down approach that intends to guide the reader from the general ideas and discussions towards the actual implementations and experiments. The key features include: unique in its choice of topic, easily readable, top-down approach, and from general ideas to actual experiments. The CD features: description and results of experiments, examples and datasets, and the software implemented for the experiments." (in Mouton de Gruyter and Amazon).


© Francisco Câmara Pereira