At the end of her comment on Dave's post, Frances introduced two very interesting topics that I've been edging toward through my last two posts:
- the connections between humans and non-human in understanding learning, and
- power relations in networks, especially "where the machine is seen as amoral and humans as 'democratised.'"
I want to explore in this post the first issue: the connections between humans and non-human in understanding learning. Connectivism says that knowledge can involve non-human entities. I find a telling footnote in a list of Principles of Connectivism in George Siemens' online book Knowing Knowledge. The salient principle reads "Knowledge may res ide i n non-human appliances, and learn ing i s enabled/facilitated by technology." The footnote reads:
The concept of knowledge resting in non-human appliances (mediated by artificial intelligence or directed by intelligent agents) is controversial. As with the discussion on context-games, how one defines knowledge largely determines whether one will accept this definition. As I mentioned in the preface, I have largely avoided the use of the word information in this text. It could be well argued that all knowledge is simply varying shades of information, and information itself is transformed into knowledge when we have a personal relationship with it (i.e., we internalize information). This discussion, from my perspective, is unnecessary for the purpose of this book. In order to have any practical discussion of information and knowledge, we need to discuss it as if it is something that a) describes some aspect of the world, and b) something on which we can act. This simple definition provides the basis for viewing knowledge as being able to reside in non-human appliances.Of course, this is too short a selection to be taken as a definitive statement of Siemens' position, but it can be a useful point of departure for my own thoughts. First, I notice a tension in how Siemens situates knowledge in an appliance, by which he seems to be suggesting modern, electronic appliances with artificial intelligence and intelligent agents. In his principle and then again at the end of the footnote, Siemens says that knowledge can reside in non-human appliances. He compliments the verb resides with resting when he says that "the concept of knowledge resting in non-human appliances … is controversial." To my mind, reside and resting suggest a passive holding place for knowledge, and this is very old hat indeed. We've been using non-human appliances such as cave walls, clay tablets, and papyrus scrolls to hold knowledge for millennia, both as extensions to our memories and vehicles for communication. Nothing new here. Of course, we can store more information in smaller spaces than before and we can share that information more readily with more people, and those are important differences, but they seem to me differences of degree rather than kind.
What is new is captured in Siemens' verbs mediated and directed. I don't know how active a role George is positing here for non-human appliances—and the meanings of these two particular verbs could be slanted either way toward more or less activity—but this is a difference in kind from the non-human knowledge appliances that came before, or so it seems to me. One of the virtues of a printed book is that it does NOT interact with the text, with the information contained within its covers; rather, the book preserves unaltered the text so that what I read today is the same as what I read yesterday (minus my own marginalia) and the same as what I will read again tomorrow. The text is static. Today, text is dynamic—along with other forms of information/knowledge such as image, number, audio, and so on (allow me to use information and knowledge interchangeably for a moment). Few these days will visit a third time a web page that hasn't changed, and while those changes are still mostly made by humans, they are increasingly being made, mediated, and directed by non-human appliances. This represents a fundamental shift in the relationship between human and the non-human knowledge/information appliance.
I see this difference most clearly in terms of Edgar Morin's distinction between closed systems and open systems. A stone is an example of a closed system: it is self-contained, needing next to nothing from the eco-system to maintain and persist as it is. Life forms, including humans, are examples of open systems: they are not self-contained, but absolutely require a dynamic exchange of energy, matter, organization, and information between themselves and their eco-systems to persist as living forms.
A book is more like a closed system (not completely closed, as I will argue another time, but for now, let's just say it's more closed); the recommendation engine on Amazon is more like an open system. The book does not take in more energy or information, process it, rearrange itself according to the inputs, and release new energy, information, and waste into its eco-system. This is not to say that the book is not a dynamic system. It is. Rather, the book is dynamic more in the way a stone in a stream is. Dynamic eddies and swirls develop, shift, grow, and wane as the water rushes about the rock, but the rock does not take in those as inputs. It just remains what it is until stronger outside forces push it into being something else. We want something similar from our books. We want them to stay put as they are. Just try re-writing Huckleberry Finn to see what a fuss you will raise.
The Amazon recommendation engine, on the other hand, does take in information and energy from its eco-system. It then processes that information, rearranges itself accordingly, and releases new energy, information, and presumably waste into its eco-system. This engine is more like a rather single-minded amoeba, or maybe a pigeon. I'm not sure how high up the intelligence ladder to go, assuming there is such a ladder, but this non-human appliance is able to monitor its environment, take in information, recognize options, make regular judgements based on that information, and release new information into the eco-system, which then becomes part of the reiterative feedback into the recommendation engine. Like all life-systems, the Amazon recommendation engine has the ability to self-organize, or simulates that ability very, very well.
Interacting with an open-system appliance is different from interacting with a closed-system appliance, or it sure feels different. But if so, then how? Let's think on that for a day and talk about it tomorrow.
I would see the Amazon recommendation engine (RE) as a human activity system (containing technology) in Checkland's Soft Systems Methodology terms OR as a subnet with human and non-human (inc tech) nodes in network terms. Taking the latter view it's the connections (and their nature) and how the network grows that matter in how it works. For example, you can only make a recommendation when you have made a purchase in Amazon's RE unlike Tripadvisor. That is why I would trust (but not blindly) the former rather than the latter.ReplyDelete
Gaming REs is not only possible but almost inevitable, and that why I treat them with caution.