I have been meaning to write on this topic for a very long time. As a teacher, a biologist and a computer scientist the process we call reasoning is always to the fore, whether delivering a biology module on ethology, discussing fuzzy logic and AI or marking scripts and wondering what on earth is passing for ‘reasoning’ in the minds of my students. The recent ‘Brexit’ referendum brought it all together for me … just how does homo-semi-sapiens reason?
Artificial Intelligence or AI has had spectacular gains over the past decade or so. It started with IBM’s ‘Deep-Blue’ defeat of Gary Kasparov at chess in 1996 before retiring in 1997 after winning a controversial rematch. 20 years later Google’s computer beat the current world champion at GO! The latter board game was levels of magnitude less ‘logical’ and predictable than chess and regarded as more intuitive and complex. So much for board games. I struggle to play either so would stand no chance against a computer’s intelligence.
Next up in the real world the BBC recently reported a US AI defence system successfully outplayed incoming attack jet fighters by basically outfighting them with its own interceptors. So much for war then. I am not able to fly a jet, real or virtual, and I know little of strategic war routines having not been trained in the military so I must defer to the computers once again.
The question for me is not how smart is AI and how smart can be the very smartest humans but ‘just how dumb are humans, really?’
If we take our mammal cousins, the field mice, and look at their ‘thinking’ in the context of the life choices they have on rising in the morning. They have very little body mass and so have little food reserve. They cannot go without food for very long so each day they look for food. But they also need to reproduce for the species to survive and to do this they must look for a mate rather than forage for food. This is risky as they may, find a mate and then starve or fail to find a mate and starve.
How do they make this life and death choice? Well, very detailed study showed that they use a very simple algorithm which is ‘I ate well yesterday so I can look for a mate today’. That’s it!
This simple algorithmic approach to reasoning is widespread in the animal kingdom and are inferred from observed behaviours. This is why it belongs to the field known as ethology which is the study of animal behaviour. These algorithms are known to the higher primate homo-sapiens and we call them ‘rules of thumb’. Rules of thumb abound today, after all if you want to improve something then measure it. So here then is building block No1 in human reasoning
Reasoning method Number 2 was explained to me on BBC Radio 4 many years ago when a politician was describing ‘thematic reasoning’ which at the time was new to me and a revelation. Essentially things, issues, people, whatever, are ascribed a virtue, that is ‘good’ or ‘bad’. To whit an often repeated 1990’s experiment is to ask the general public which of the following processes were most responsible for global warming: electricity generated by coal fired stations, oil fired, gas fired or nuclear powered? Invariably the order was, worst=nuclear, best=gas.
It’s simple to understand their mistake. Global warming = a v bad thing; nuclear power = a bad thing ( this is 1990s post Chernobyl), bad begets bad so the conclusions are linked by theme. Apparently according to my radiophonic mentor all politicians understand this and are taught it at politician school which explains a lot...see referendum debates, invasion of Iraq etc etc.
Finally we have reasoning Number 3. It’s called magic. James George Frazer’s famous book ‘The Golden Bough; a study in magic and religion’ in the late nineteenth century does it for me. To cut a long story short, belief systems developed that introduced new ways of reasoning. They were/are characterised by rituals and symbols and embodied principles. The latter include an association inferring cause and effect, eg the Rooster crows just before sunrise ergo the sun’s rise is caused by the Rooster's crow. This is broadly called sympathetic magic the other main type is contact based and is called contagion. So for example the baptising of a baby with holy water is an example of contagious magic replete with symbolism and iconography.
So far so bad for homo-semi-sapiens. What about higher levels of reason? The kind of thing you get from education.
Take for example conditional statements, ‘if I do this, that will happen’. Every teacher knows by experience that students can have their minds trained so as to make correct conclusions using the conditional so long as they are given the initial knowledge conditions and the rules of the game. This is certainly the start of logical thought and in itself an educational achievement but ask also the same teachers (and setters of exams) what success they will have with double conditions. Eg, ‘ if this occurs then that will occur and then the other will occur as a result’...the nested ‘if’ of computing. Most students cannot do this.
Putting it all together, Homo-semi-sapiens in a developed society will have access to ‘rules of thumb’, thematic constructs, magical thinking and one step conditions.
The famous thinker Scott Adam’s seminal 20th century work ‘Dilbert: Thriving on Stupidity in the 21st Century’ describes this beautifully and is still elaborating his ideas today in his latest comic strips where engineer Wally is nominated for a Nobel prize for his AI humanoid; aka a block of wood.
Homo-semi-sapiens is on the march today aided by the connectivity of social media which allows for self-referencing and reinforcement. Authority is challenged as never before as ‘priest classes’ lose their grip unable to explain to their audience what is in effect unexplainable and resorting to using reason that is in common use (as described above).
It is ironic that we debate the capabilities of AI systems when the ‘I’ systems in general use are so primitive. Maybe if we wish to produce realistic AI systems we should start with our own protocols.
Am I being patronising to semi-sapiens? Maybe once I could have been intimidated into being less scornful. However the Brexit referendum has removed all such inhibitions.
