Navigating AI landscapes: Why adaptability beats prediction

THERE is a question that always comes up when people learn someone sails alone across oceans: how do you sleep?

- ANDILE MASUKU

The answer, offered recently by solo sailor Noa Hopper navigating between Lisbon and Tenerife, is instructive well beyond the sea. Turns out you don't get eight hours. You don't get the recommended rest. Instead, what you get is a new rhythm, calibrated not to your preferences but to your actual circumstances.

In a busy shipping lane, Hopper sleeps on deck in twenty-minute bursts. In open ocean, where he has not seen another vessel for three days, he allows himself forty minutes below. The interval isn’t arbitrary. It’s calculated from a simple fact: his visible horizon is 8 miles (12.87 kilometres).

At ten knots, he reaches that horizon in roughly an hour. But if another vessel is doing the same speed toward him on a direct collision course, both ships become visible to each other only twenty to thirty minutes before impact. So twenty minutes it is. Not a prediction. A calibration.

While tracking the volume of AI disruption commentary flooding my feeds over recent months, I’ve been looking for ways to make sense of all the data, insights and speculation swirling about me in what often leaves me feeling like I have my mouth attached to the end of a firehose on full blast.

For the most part, the issue isn’t so much that the commentary is wrong, but it is almost uniformly doing the wrong thing. It's trying to predict. Almost never calibrating.

A fortnight that tells the whole story