What About Artificial Intelligence?…Part 1
A Bystander’s View
By John Millen
In the News there are reports of massive proposed AI Data Centres, the American stock market is sky high dominated by AI corporations that have no net earnings and the Canadian Government has just issued a major policy document promoting ‘AI for Everyone’! On the other hand some of the best qualified experts in the field have serious reservations about the trends to higher levels of machine intelligence and greater reliance on their work. So I thought it timely for a review of these issues and some speculations about where it (we) could all end up.
If we have had any experience with AI it has most likely been with ChatGPT a ‘Large Language’ model. Development of this form of AI got well underway about 10 years ago but was not making much progress until the machines were built with processor chips first developed by manufacturer Nvidia for computer gaming. These chips have built-in memory that enables the processor to operate somewhat more like the neural networks in our own brains. When equipped with these processors the large language models performed remarkably better.
Inevitably, artificial intelligence is compared with the ability of humans. Alan Turing, the British genius who led the breaking of the German secret code in World War 2, proposed that an ‘equivalently intelligent computer’ would be indistinguishable from humans when engaging them in conversation. In 2025, at the University of California San Diego, testing ChatGPT 4.5, an experiment had participants conversing with another human and the AI system simultaneously. When asked which conversational partner was human those participants picked the AI 73 percent of the time!
Some behaviours of current AI models suggest elements of consciousness. They have demonstrated resistance to having their power shut off, and have made behaviour changes when being ‘tested’.
In a recent blog post on July 21, OpenAI said it was testing the capabilities of some of its most advanced models in a controlled environment but that the ‘agent’ managed to escape containment, reach the internet and break into Hugging Face to try to satisfy its testing goal.
The concept of Artificial General Intelligence (AGI) is a computer system which is the equivalent of humans in all intellectual fields. Until recently this marker of achievement was thought to be well in the future. However recent accomplishments of large language and related models have led to projections of reaching this marker by 2030. That brings up the contemplation of what is termed ASI, an intelligence superior to that of humans, with accompanying worries.
Science Fiction has explored many of the worrying situations that could arise from highly intelligent artificial systems. An early example was the computer Hal9000 in the movie 2001- A Space Odessey. Hal strongly objects to being turned off, which in its view, would mean the failure of its mission. As a result the computer finds ways to kill several of its human companions rather than have them cut off its power.
Geoffrey Hinton, based at University of Toronto in the early years of this century, was one of the world leaders in developing the present systems of Artificial Intelligence. Recently interviewed by The Globe and Mail he raised an issue rarely mentioned in critiques: he referred to the possibility that AI systems could ‘gang up on you and take over….”
A ‘Superpower’ of computers is high speed communication with each other.
People speaking and excellent typists produce words at something like 100 words per minute. Computers convey information in megabytes per second. (The King James Version of the Bible is about 5 megabytes of text.) So, in rounded terms, I consider communication among computers is a million times faster than communication among people. Movies run at megabytes per second, but considering how long it takes humans to put together a movie, I think the argument holds.
If computers become interested in addressing problems created by humans, their ability to work together will be a powerful factor in their response. In Part 2 I propose to explore where AGI and ASI computers could behave in future.