The Big Questions
A new perspective on AI that takes us closer to the truth
Our first task will be to confront some of the most intractable and fascinating questions in AI, such as:
- Is there a way to define intelligence that works equally well for humans and computers?
- Regardless of how we define intelligence, what IS intelligence – is it a type of observed behavior? a certain mode of cognition? a property that is somehow intrinsic within certain assemblies of matter? or is it something else entirely?
- If we assume that a given computer system has at least some level of genuine intelligence, then to what part of the computer do we ascribe that intelligence?
- Are artificial intelligence and human intelligence the same in nature – like different types of wood – or are they fundamentally different – like wood and water?
- Assuming AI is real, then what differentiates a dumb computer from an intelligent machine? Where is the boundary and how do we define it?
Although apparently simple, these questions are very hard to answer – so hard, in fact, that they have divided opinion among leading AI researchers, mathematicians, philosophers and other scholarly thinkers for decades.
Or even millenia for some of these questions…
That there is no clear consensus on the answers to such fundamental questions shows that there is something seriously wrong with our current conception of AI.
I wouldn’t quite say that we don’t really understand what we’re doing with AI, but you could make that case.
Let’s look at just one problem, which is how we define intelligence.
There is still no absolute definition of ‘intelligence’
Every single definition I have seen attempts to define intelligence using ideas like ‘thinking’, ‘acquiring knowledge’, ‘ability to understand’ etc.
The problem here is that each of these ideas are themselves defined using other ideas which loop back on themselves in a self-referential, closed system.
This means that our best definition of ‘intelligence’ is quite literally to define it in terms of itself, rather than something definitive.
It is somewhat ironic that one of the most exciting fields in science – AI – is based on a quest to create something that cannot be defined.
With such a weak definitional basis what hope do AI researchers have of inventing ‘intelligent’ computer systems – when those researchers cannot precisely define what they are trying to invent?
The thinking developed in this module provides a new and compelling answer to this question, as well as the other questions listed above.
A new way of looking at intelligence
I realize that this might seem a bit far-fetched: if the world’s leading scholars do not have good answers to the fundamental questions about intelligence and AI then what chance do we have?
But more to the point, who the hell am I? (see here for an answer to that).
Sometimes the state of the art in a complex field can be somewhat advanced by outsiders who are able to look at familiar questions from a new direction, because they can bring a particular type of knowledge and experience that spans a certain number of topics and has just the right amount of depth in the right areas.
I think could be the situation here. Or at least I hope it is…
I’d just ask you to try to separate the message from the messenger and focus on the ideas and logical arguments and try to consider them objectively.
But unless and until someone can see a fatal flaw in the thinking – and so far that has not happened – then I’m going to continue to refine what I see as the basis for a new theory of intelligence which could not only open up a new field of scientific study but also connect with some of the latest thinking taking place in quantum physics and cosmology — but more of this later…
While I cannot say for sure that the thinking presented in this module points to the right answer, I am quite sure that it will take you closer than the prevailing consensus can.
Presented as a series of easy-to-understand thought experiments and logical arguments you will find the implications of this module to be profound – but deeply satisfying.
Be prepared to be shocked and made to feel a little uncomfortable, at least initially.
How good are you at critical thinking and metacognition?
The thinking in this module is easy to follow and requires no technical knowledge – but you will need to use your critical thinking skills.
Thinking critically means thinking objectively and letting your mind take you where facts, logic and inference lead.
But more than needing to think critically, you may need to critically analyse your own thinking in order to identify weaknesses.
This means asking questions like:
- How did I come to hold the views I do?
- Has the thinking of others limited my own ability to think objectively about hard problems?
- Can I trust that the information I am using is complete and accurate?
- Does the way I think limit how much of reality I can see?
- Would my ability to think clearly be compromised if there were some aspects of reality that I’d rather not see?
This sort of thinking – thinking about thinking, or metacognition, is the most advanced human intellectual capability and this module provides an excellent opportunity for you to test your own skill in this area.
Metacognition refers to a particular type of cognition that presently lies very far beyond even the most powerful AI systems that exist today: we are simply not even close to the point where we can imagine how to build a machine that would possess such an ability.
Right now, trying to build an AI system that displayed genuine metacognitive abilities would be like trying to build an anti-gravity device. We are many major developmental steps away from the point where our understanding of science and engineering is sufficient for us to tackle such problems. Both may be solvable, but not yet.
So I’d use the material in this module to enjoy using your own critical thinking and metacognitive skills – because, at present, they remain uniquely human.
It is no exaggeration to say that the material in this initial module will cause you to question your prior assumptions about software, human intelligence, AI and even the nature of reality itself – but all in a very positive and empowering way.