Diana Bang's ETEC 531 e-portfolio

The connectionist model of design contends that the brain operates from the bottom up rather than the top down approach and has been utilized in recent artificial intelligence research. A simplified version of this model is that thought is considered to develop though associative and low-level series of connections. For example, brain neurons are stimulated by the impressions given to them by the body’s sense organs. These impressions and responses then develop patterns in the brain’s neural pathways. When some networks are strengthened we achieve learning and at another level patterns form into conscious thoughts. Also, though time these patterns will continue to change. In relation to AI research, they’ve developed neural nets (i.e. web of neurons) which try to copy the parallel processing capabilities of the human brain. Minsky and Papert (1988) refer to these copies as the fundamental parts of “percepetrons” (as cited in Murphie & Potts, 2003). Thus, if this is indeed the way the human brain functions then the implications for AI research is that consciousness will develop with the proper connection between low-level components and the appropriate environment to learn. Ultimately, then the focus of AI should not be about trying to create a replica of the human brain as creating consciousness in machines is a matter making the “right” connections. What are the implications of achieving machine consciousness? Is it vision of a doomsday Terminator-like scenario or akin to WALL-E?

I’d like to think that anything is within the realm of possibility considering all the technological advancements that we’ve gone through and will continue to surely unfold. According to a CNN news article, brain downloads will be possible by the year 2050: http://articles.cnn.com/2005-05-23/tech/brain.download_1_computer-human-brain-downloads?_s=PM:TECH. Interestingly though, one of the key concepts that they’re trying to contend with is how to develop consciousness in computers. One of the aspects that needs to considered though is the fact that human memory is not accurate (i.e. people tend to forget things) whereas computer memories are more precise (i.e. the only real issue you may run into is where you last saved a document but it will still exist in the computer’s hard drive or on a usb key). I wonder then, how personal memories could be uploaded to computers. Computers have several ports in which you can save and upload files- however if you consider the human body, we do not have accessible ports for connecting to a computer via a usb key for example. The human brain could be examined via stimulating electrodes surrounding the head, but how exactly would we be able to capture memories like data files? Generally, computer files are encoded, and so would the contents of our memory be also encoded and if so, would it be as easy as a file transfer?

Thus, at this point in time I’m a bit uncertain whether it is truly realistic to upload our memories to computers as of yet since some have argued that doing so would mean losing the body and its situation in the world which is important to consciousness. Also, we still do not fully understand how the human brain operates and how we can mimic this in computers. This is critical to the idea of developing the connections between the human mind and a computer. If we can understand how the human brain functions and translate that into a computer- then perhaps we would be another step closer to achieving this possibility. Another issue that needs to be addressed is using different “technologies” (i.e. the human brain and computers) to simply make a connection and pass information between them. For example, N. Katherine Hayles (1999) contends that using a “white box” (like computers in which we can view and understand all the inner components) is not the equivalent of a “black box” (i.e. the human brain) in which we still do not understand the mechanics of it. Hayles (1999) asserts that AI researchers claim that the way machines imitate human behaviour offers a model for the human brain. There’s faulty logic in this however as we cannot consider these different aspects (a computer vs. a human brain) to function the same way.

Additionally, thoughts and memories are not something that can be merely produced or reproduced in an artificial setting (i.e. laboratory) outside of the natural world. Thus, it is difficult to consider that uploading memories to a computer will be a reality as of yet if we still do not fully understand how the human mind works. For example, Steven Rose contends that computers simply retrieve digital information whereas the mind works with meanings in a creative and imprecise manner. Currently, we are able to capture and record our memories through the aid of electronic devices (i.e. digital photos, journaling/blogging, and through the use of videos). Yet, it hasn’t come to the point where we can simply connect to a computer and it automatically downloads the contents of our brain. Plus, if it did- where would this information go? Who would be in control of it and who would have access to it? There’s a lot of questions regarding privacy and confidentiality concerns that may be raised because of this. Also, if this were possible I wonder how the computer would interpret the information contained in our memories. Would it be accurate? And how would we know the level of accuracy it achieved since our memories can already be faulty at times. After all, to err is human.

One of the key concepts considers the possibility of creating consciousness in machines. Related to this is the question of whether the brain contains consciousness or thoughts. For example, Alan Turing (1950) suggested that if a computer could trick a person into thinking they were communicating with another person (via an interface) then it could be deemed as intelligent. Yet, John Searle (1980) argued that this was not a sufficient test and rejected the idea that computers would ever become completely intelligent. Also, another aspect to consider is information theory in that the carrier of the information does not matter nearly as much as the patterns of information that are carried. According to Murphie & Potts (2003) this is a critical idea since it suggests that two different material objects (i.e. a brain and computer) can carry the same patterns and processes in their materially different ways. However, in order to design machine intelligence we must understand how the brain functions. For example, there is much debate about whether the brain operates as in a hierarchical manner where information is processed in a system (i.e. this translates to a top-down approach in artificial intelligence). Several issues arise with this theory since it assumes that human thought is a “coherent, ordered, complex and more or less unchanging program” (Murphie & Potts, 2003, p.153). In order to effectively mimic the brain, one must develop a complex system before it can do any thinking and as such attempts to build “thought factories” have been unsuccessful as of yet (Murphie & Potts, 2003).