Artificial intelligence is often hailed as a game-changer across many industries, promising to revolutionise productivity, reduce costs, and open new frontiers of innovation. Even the UK government talks about it in such terms...
However the AI reality so far falls far short of the hype. AI’s supposed benefits seem impressive at first glance, but disappointing if you look more closely...
The Productivity Paradox
The research lab METR put AI’s productivity claims to the test by running controlled trials with experienced computer programmers. These developers were given AI tools to help them write code, with the expectation that they’d work faster and more efficiently.
The result? The exact opposite. Although developers believed they were developing 20% more efficiently due to AI, in reality, they were 19% behind.
The main reason for this was the time spent correcting AI errors, and then developing new processes to work with the AI tools rather than just tackling the problems directly with their human brain-things.
And real world experience is increasingly showing that rather than unlocking hidden productivity, businesses are facing increased errors, dependency on faulty systems, and more time spent on debugging.
But AI can reduce costs...
That is not to say that AI is of no use. The grand narrative of businesses these days is "cost reduction," not innovation.
And AI can certainly be used to automate some mundane tasks or reduce the need for some jobs, I'm thinking chat bots straight away, or any kind of advice line, and this has to be tempting for businesses when costs are increasing thanks to tariffs and taxes.
Final Thoughts
It's to be expected that AI is going to make mistakes, especially on more complex tasks. The trick is to learn how to work with it so one can weed out these errors.
Probably over time AI will evolve to be more accurate, but it would seem for now it still requires significant human steering!