I have demonstrated (see my previous articles and my book Precise
AI) that understanding language is a field phenomena and not a point phenomena. Therefore, in general, the meaning of an input sentence to a human-quality AI Assistant is dependent upon the context in which the input occurs.
There are two main types of sentence:
i) Maximum Effort (ME) sentences, where the sentence itself defines its meaning.
ii) Least Effort (LE) sentences, where meaning is determined by context.
An example of an LE sentence is: “so is his wife Mary”.
Whereas the meaning of “I visited france last year” is immediately clear, the meaning of “so is
his wife mary” depends on the context in which it occurs. However, distinct meanings exist when “so is his wife
mary” is found in different contexts, as demonstrated in the following two cases:
i) my friend peter is german
so is his wife mary
ii) alan is a dentist
so is his wife mary
Clearly the meanings associated with “so is his wife mary” depend upon context, and also as this input is a statement it must create appropriate and permanent memories. The memories are:
i) peter is married
his wife is called mary
his wife is german
ii) alan is married
his wife is called mary
his wife is a dentist
and so, if a year later the following question arises:
what does alan’s wife do?
It will be simple for an AI to respond: “She is a dentist.”
These examples should provide sufficient insight into the essential role of context in mediating an input sentence to meaning.
If we are to study intelligence and the mind, then we will either need a form of remote sensing or insights that can be verified experimentally – and it is this latter approach that I have adopted.
To understand the movement of the planets we needed the telescope.
To understand the composition of the stars we needed spectroscopy.
To look inside the human body we discovered x-rays.
For details within the human body G N Hounsfield invented the CT scanner – an inventive combination of X-rays and mathematics. Now we have NMR scanners – physics plus mathematics.
To study the human mind is a more difficult problem than those listed above and yet there is one very powerful technique that is surprisingly little used – and that is analysis and intuition.
Perhaps the best example of this occurred in 1874 when Jacobus van’t Hoff determined from a study of the optical properties of carbon compounds that the carbon atom had a tetrahedral structure. A wonderful example of human
intuition followed by detailed analysis. He was later awarded, for other work, the first Nobel Prize for chemistry in 1901.
In the 21st-century one similar way to investigate the mind and explore the properties of intelligence is to study conversations. With careful analysis and insight it is possible to determine many of the properties of intelligence and the associated memory structures.
A good example of the role of context in a conversation can be found in a 14 minute YouTube video on the Precise Intelligence channel entitled: Alexa & Siri and Conversational Intelligence.
In 2025 Apple promised the world a new Siri in 2026, however as of January 2026 this context dependent capability seems to exceed Apple’s current technical competence. Therefore customers must now settle for the limited conversational capabilities offered by Generative AI. Will such systems offer the context required for human-quality conversations ? It seems unlikely.
