"Robo" sapiens ?

I’ve heard psychologist Daniel Kahneman discuss human and investor behavior on previous occasions. This time, at the recent CFA Institute conference in Hong Kong, he moved his focus from “thinking fast and slow” to “noise and framing”, especially where human (“expert”) judgement of a situation is required. He opened my eyes yet again to these very real and deep-seated behaviors we all harbor. It’s astonishing how twisted our reality, individually and in groups becomes because of this. The societal and economic costs of these human behaviors or “inefficient” decision making is alarming. We can look for many examples in the world around us, but one notable study he quoted was in risk underwriting where the expected error rate due to human judgement is around 10%. His studies found the error rate to be well over 50%. And, according to him, this was shown to be the case repeatedly. So...rather toss a coin than employ qualified humans to make judgements on intricate cases where expertise is required?

Expertise can be acquired through rigorous trial-and-error but the feedback on decision-making must be immediate to embed the learning algorithm and then also not be “contaminated” by internal noise or false signals. Think of our own biases or the filtering and framing of situations that happens almost instantaneously in our minds. We are very much a function of our mental wiring which in turn is a function of our past learning experiences. Add to that our inherent loss-aversion coupled to our inherent over-confidence - a conflict within which is second to none! This is where machine learning or artificial intelligence (AI) has an advantage. Machine’s can now learn by creating opposing identities or by inventing scenarios. All of these with twists and turns and this can be done millions of times over and over in a single day, even in minutes! Interestingly though, AI often suffers from inappropriate framing or, to put it another way: misinterpreting the context. This appears to be a main focus of current research and development. Before we dismiss the technology's ability to overcome these obstacles, one must remember that the rapid progress in AI has exceeded even the most optimistic scenarios by integer factors! So, future progress can be expected to be rapid as massive amounts of data add to the learning context for machines. The biggest concern, in Kahneman’s opinion, is where AI’s rapid progress will leave human society. Given his credentials, I take his observations of the progress and value of AI seriously - as an opportunity and as a potential threat. Us humans are inherently “noisy” in our thinking and AI is a very viable and many would say even a preferable alternative to human judgement in many cases.

For further reading, go see what Alphabet’s DeepMind is about at deepmind.com.

Alphabet has been a key holding in our PCS Global Equity model portfolio since the portfolio’s inception more than three years ago.