Secondary education: Newington College, Sydney, Australia, 1963-1969.
Tertiary: University of Canterbury, Christchurch, BA Philosophy and Economics, BSc Geology, MSc Philosophy of Artificial Intelligence.
Work history: 1980, New Zealand State Services Commission, Industrial Relations Branch, Wellington, under Maxine Gundersen, advisory officer assigned to stevedore sector. IRB negotiated annual ware rounds with public service unions.
1980-2015, computer software development, wrote the first government-approved commercial program to fully process income tax, wrote a large accounting, distribution and manufacturing software suite then licensed it to a multinational for 22 years, using the annual license fees to fund AI research, which I'd been interested in for many years.
Research: As part of AI research, I spent about a decade in graduate research in Philosophy of AI and two decades independently investigating AI theory. I was interested in discovering whether data compression algorithms could create human-like semantic knowledge structures that yield meaning. Obviously there's huge compression in the animal brain thanks to evolution selecting for small size and weight in the context of fast, appropriate responses.
The structures and algorithms came from my US patents 5,748,955 and 6,414,610. US 5,748,955 describes basic algorithms underpinning unsupervised LLM tokenization. In data compression terminology: the extraction of redundancy from undelimited bit streams.
The patents describe data compression in terms of text characters. The patents were text documents, so examples of structures and algorithms had to be expressed as text. But the widely-held view that computers process text, meaning they receive, manipulate (move), create, store, and emit text, is false. Rather, they process instances of electronic or electromagnetic binary difference.
In AI research I considered knowledge as togetherness in time. So human-like computer knowledge comprises stored instances of binary difference related by togetherness in time prior to storage. But on storage, the relation of temporal togetherness is removed, and instances of atomic or compound binary difference are related in storage by a timeless, or permanent, means.