Opinion Express View: Not even over a beer

It’s not a coincidence that a nearly 4,000-year-old ode to Ninsaki — a goddess of brewing — in ancient Sumerian contains the oldest surviving recipe for beer. Like language and agriculture, the yeast-assisted intoxicant is a mark of civilisation, an acknowledgement that life is not merely about tasks — it is about taste as well. And community and revelry and leisure. In the millennia that followed, cities rose and fell, emperors came and went. The brewer’s art, though, has survived and thrived. There are flavours and textures, different grains — rice beer, wheat beer, barley beer — and with subtle changes in the recipe, new drinks still emerge. Now, the hops masters’ art faces a new challenge.

Researchers in Belgium — their findings were published in Nature Communications — have devised ways for “predicting and improving… complex beer flavours through machine learning (ML)”. Essentially, a programme has been trained on a variety of highly-rated beers and can provide recipes that can make a better beverage. The subtle art of the brewer, or even the joy of discovering a recipe through tinkering, will be reduced to numbers, processed by a sophisticated abacus. The tech optimists may argue that, at least when it comes to the big companies, there is no art — just massive labs and consumer surveys. But even the largest breweries couldn’t predict taste — which can be as mysterious as love or happiness or god.

Advertisement

In 1954, Roald Dahl wrote ‘The Great Automatic Grammatizator’, a short story in which a mechanically-minded man, unable to write, invents a machine that we would today call a Large Language Model. It can write in the style of any writer and is faster than any human. In the end, the machine and the man who invented it killed prose and poetry. That story is coming true in many spheres. Sadly, soon, perhaps people won’t even be able to commiserate about the age of art, or lament the technocratic, engineered, data-driven world they live in — over a beer.