AI performing tasks once done by humans — causes and evaluation

Artificial intelligence is now performing tasks that were until recently the exclusive preserve of human workers. This shift has both technological and commercial causes, and in my view it is a positive development overall, though only if the disruption it creates is properly managed.

The primary cause is technical. Modern systems are trained on vast quantities of data using computing power that simply did not exist twenty years ago, which allows them to recognise patterns in language, images and medical scans that once demanded human judgement and years of training. The second cause is economic. Once a system works, it operates around the clock without wages, holidays or fatigue, so any company that declines to adopt it risks being undercut by competitors who do. These two forces reinforce one another, since commercial demand funds the research that makes the next generation of systems more capable still.

On balance I regard the trend as beneficial. Automation removes work that is repetitive, dangerous or simply dull, and it has already made services such as medical diagnosis faster and more accurate, which frees people for tasks requiring empathy and creativity. The benefits, however, are not evenly distributed: data-entry clerks and call-centre staff lose their livelihoods while shareholders gain, so governments must fund large-scale retraining and ensure that productivity gains are shared rather than concentrated at the top.

In conclusion, artificial intelligence is spreading because it has become both highly capable and remarkably cheap. Whether the outcome is positive depends far less on the technology itself than on the policies that accompany it.