Mind the productivity gap

Excel was in the news recently, as the Office for National Statistics (ONS) admitted to a rather embarrassing error. The type of error we’ve all made, where we choose the wrong cell to divide another cell by; the difference here is that the ONS’s mistake not only made it into the public sphere but was massively wrong.

The topic in question: productivity. This is defined as output divided by hours worked. Now imagine what happens if you click on the wrong column; say, if you divide 2022’s output by the hours worked in lockdown-stricken 2021. The answer is so far out it wears a flowery shirt.

Rather than correctly showing that Britain’s productivity retracted by 1.8% between 2022 and 2021, the

ONS’s first set of figures suggested we had become 22% more productive.

What struck me wasn’t the Excel mistake, or the lack of quality control, but that nobody thought to themselves, how? A country’s population doesn’t become 22% more productive in a year unless something big has happened.

And by big, I mean epoch-shifting big. Think the arrival of labour-saving textile machinery and skilled workers to operate that machinery in the 18th century. The gradual implementation of electricity into our workplaces in the late 19th century, along with those companies savvy enough to take advantage. The arrival of computers in offices in the 1970s and 1980s, when word-processing skills – and indeed spreadsheet skills – became key.

There’s always resistance to such change. The Luddites even became a word in our vocabulary. And as any Gartner fans will know, we can rapidly move from excited hype about a new technology into the “trough of disillusionment”. Sometimes it takes years, decades, for a product to climb up from the trough and into the “plateau of productivity”. But we’ve seen it happen time and time again, whether it’s Bluetooth or blockchain.

Common sense suggests generative AI, popularised by OpenAI’s ChatGPT, will follow a similar pattern, but every new technology weaves a different path – and today’s pace of change makes it feel like we can jump from over-hyped to pits of despair to productivity, and back again, in the space of a week.

The thing is, GPT-3 – the language model behind ChatGPT – is already useful. I know because Barry Collins goes on about it so much that if I hadn’t given the green light to this month’s feature, he and his army of sentient robots would have revolted. Awful as it is to admit, however, Barry is right. Generative AI is already being put to practical use, and if we don’t get on board we’ll become the Luddites: the people who needed their kids to help them use the VCR, except that this time it’s not a physical machine that needs operating but a chatbot.

And, as I alluded to at the start, it’s not just about us as individuals. As a country, our productivity has fallen behind other G7 nations in the past decade. Yet history shows that we aren’t a nation of shopkeepers but also innovators and risk-takers. We shouldn’t be worried about generative AI taking our jobs. Instead, we should be investigating exactly how we could be doing our jobs better. Fortunately, Barry reveals exactly that from p26.