AI and The Art of Intelligence

Over summer, I met Olivia, the latest addition to my extended family. The introduction went as follows.

Smiling baby (Olivia) visits our home with her mother Carlota, who passes the child carefully to me. Olivia’s smile inverts in seconds. Distraught yells follow. I quickly return Olivia to Carlota. Moments later, calm returns – Olivia is smiling again.

I am not a child psychologist. One academic friend suggested Olivia was showing early signs of perfect rationality (reacting with horror at seeing a monster). A more generous colleague tells me that babies ‘see’ the world as one where they remain physically attached to their mothers and any challenge to this worldview risks a panic response.

I will leave you to make up your own mind.

For whatever reason, the baby reacted instinctively when she was out of her comfort zone. Her cries amounted to ‘get me out of here’, an example of what Daniel Kahneman might characterise as early ‘Type 1’ thinking (a fast and automatic decision when time does not allow for reflection).

Through life, inherited infant impulses grow into adult intuitions. These are worth much more than the early gift of nature passed from mother to child. They draw on years of lived experience, sometimes codified in our language as rules of thumb. At other times they lie hidden in maturing memory casks that our minds access instantly in a crisis.

We call on our intuition when confronted by unquantifiable doubts, what former Bank of England governor, Lord Mervyn King, calls ‘radical uncertainty’. Intuition also guides us when there are too many options to consider in time-constrained situations. Such decisions are error-prone, but our lives depend on them.

Some commentators frown on intuition. They see it as too quick and dirty. However, it is simply a category error to assume adult intuition involves little prior thought. Quite the contrary, it may take years for the outcomes of experience to distil in our minds in a way that allows these fast intuitive judgments.

Pure rationality, by contrast, runs differently. In complex decision situations, it arms itself in models that aim to give the right answer, quantifying any residual risk numerically. These models can be powerful but are subject to their own weaknesses.

With assumption layered on assumption, a model may give a false pretence of certainty. A derived conclusion is valid within its territory but may simply miss critical aspects of the terrain it is trying to explain.

Intuition and rationality are both fallible. The main point is not to fall into an either/or trap, but instead hold on to both/and thinking.

The French mathematician, Blaise Pascal, often credited with laying the foundations for modern decision theory, summed it up perfectly:

“Deux excès: exclure la raison, n’admettre que la raison.”

(Two mistakes: to exclude reason, to only admit reason).

In investment, there are dangers that come with intuition, taking hasty action when some cool-headed calculation could have led to a better outcome. Artificial Intelligence (AI) can help us. A modern computer can carry out operations in under a second that would take a human thousands of years, with a lower risk of mistakes. So, when high-quality information is available, AI can be an extremely valuable tool, subject to the laws of statistics (but that is another story).

However, investments in many asset classes, especially those embedded in networks, including banking, airlines, energy and nature, raise trade-offs that are partially intractable to computation. Many are exceptionally long term, and uncertainty of how the world will unfold is inevitably high. Asymmetrical risks, non-linear feedback effects, irreversible sunk costs and, of course, the unknowns inherent in an unfolding future are among the challenges.

And this is where lurks one big danger from AI – hubris. AI has no context outside the data and model and no sense of its own weakness. It follows instructions like a loyal foot soldier. Conversely, human intuition comes with the wisdom of the general who understands some risks lie hidden.

A useful heuristic in some circumstances is ‘safety first’. This is a mantra that has stood the test of time within the best financial institutions, the airline industry, banks, electrical utilities, and nature itself.

Sometimes, this means investing in so-called ‘redundancy’. But extra cost should always be measured compared with additional value. As one of the Founding Fathers of the United States, Benjamin Franklin, put it:

 “An ounce of prevention is worth a pound of cure.”

At other times, it may simply mean being brave enough to slow down, or even say ‘no’.

These are lessons of relevance in parts of our IT industry today.

So, back to Olivia. She was employing an inherited gift of evolution lying in her early consciousness. ‘Safety first’.

Children learn from us. But it is not a one-way relationship. Sometimes, we can learn from them too.

Additional Reading:

Iain McGilchrist: The Matter with Things. Perspectiva.

Daniel Kahneman: Thinking fast and slow. Penguin.

Mervyn King: The End of Alchemy. Abacus.

Pascal: Pensées. Penguin Classics.