Could CODIL model how the brain processes information because I have aphantasia?

 On the 18th January 2024 the Aphantasia Foundation published the following note onto LinkedIn  which is very relevant to this question.

 

Aphantasia Network

🤖 What happens when you ask a large language model like ChatGPT to "Think of a Horse" and describe the image that comes to mind? The answer is as intriguing as it is enlightening, especially for those of us with aphantasia.

 

When ChatGPT was asked to think of a horse, its response was: 

 

"As an artificial intelligence, I don’t have the ability to visualize or imagine in the way humans do. I don’t have the ability to form mental images. So, when you ask me to think of a horse, I don’t actually conjure up an image of a horse. Instead, I access my training, which includes information about horses, their characteristics, behvaviors, and the context in which horses are typically mentioned."

 

This reflects our experience with aphantasia! We might not "see" in our mind's eye, but we have a profound capacity to understand and interpret information based on our learning and experiences.

 

🌐👀 The parallels between AI and aphantasia offer a unique window into human thought and AI potential.

 An AI version of a "blind Mind's Eye" generated by Bing

This is relevant to the CODIL project because I am neurodiverse and have aphantasia and my first job involved me working as a human chatbot. My task, as a graduate level clerk, was to monitor research and development correspondence in an international veterinary organisation, answering queries and producing monthly report highlighting issues relevant to future company products. No computers were involved but when I became aware of Bush's 1945 essay "As we may think" I decided to find out how computers could be use to provide better and more accurate management information. In order to learn more about computers in 1965 I moved to a major oil marketing company using LEO III computers to process sales and produce management reports. Two years later I was recruited by English Electric Leo to do market research on the requirements for the next generation of large commercial computers which would hopefully offer Integrated Management Information software accessible through user terminals. This gave me the chance to learn about the problems other major computers users were having with their "black box" commercial systems, and also meant I had discussions with the engineers who were responsible for designing computer hardware. I quickly decided that what was needed was a generic interactive user interface which could help company management handle complex real world tasks, As a result  proposed the CODIL language, which could be processed more efficiently if the computer CPU had better facilities for associative addressing.

 

In effect what I proposed was a computer clerk that could work symbiotically with humans handling commercial information in complex and ever changing marketing situations.  I decided the  computer clerk must use the  human vocabulary and process information in a human-friendly way.  If anyone had warned me, at the time, that I  was building a system which modelled human intelligence I would have considered the proposal far to difficult and started to look for other simpler solutions. But what I was doing merely involved transferring the way I had processed information (with my aphantasic brain) onto a computer - and to me this seemed a comparatively simple task.

 

So the statement in the above post that "The parallels between AI and aphantasia offer a unique window into human thought and AI potential" is very relevant. Much of what we know about how the brain works comes from case studies involving people where there is evidence of a misfunction. The fact that I have aphantasia means that I cannot rely on visual memories, and instead have to process "visual" information in symbolic terms. This "defect" could well mean that the CODIL model, actually reflects the way human brains processes symbolic information when not cluttered with visual information, and that I have accidentally reverse engineered how human short term and long term memory work to make intelligent decisions.

 

Of course the above does not prove that in the mid 1960s CODIL was a prototype reverse engineered working model of how the brain stores and processes real world information. However one of the purposes of this blog is to explore this further, and look at the relationship between the CODIL language, human and artificial intelligence, and the pathway by which intelligence evolved.