American sports don’t often grab my attention but this gridiron story published last week was an exception: ”Bears Pull Off a 27-20 Upset Over Packers.” It wasn’t the summary of the game that interested me but the byline.
The story was written by a sports reporter named ”Automated Insights” – a machine rather than a human. Automated Insights is the name of an American company with computer software that sifts mountains of data, identifies key patterns and trends then ”describes those insights in plain English with the tone, personality and variability of a human writer”. It also provides reports on share prices, property markets and historical events.
There’s nothing new about robots. They have transformed manufacturing and transactional jobs, including cash handling and grocery deposits, for decades.
But the arrival of robot journalists illustrates how the automation of tasks once performed by people is moving into new realms. Rapid improvements in computer processing power and artificial intelligence software means machines are able to perform increasingly complex tasks, including those that require abstract thinking and judgments.
So far, robots have been effective at replacing jobs that require a single process – like the welding robots in factories that made many welders redundant or automatic teller machines that substituted for human cash handlers at banks. Most knowledge jobs, such as journalism, consist of ranges of tasks, so automating one activity may not make a worker entirely superfluous. The diversity of most knowledge has made them harder to replace.
But for how much longer?
The combination of fast computational speeds and ”machine learning” promises to give knowledge workers, such as professionals, managers, engineers, scientists, teachers, analysts, administrative support and even journalists, new tools that will augment core aspects of their work involving decision making and judgment.
Smart machines with this level of intelligence are likely to be commercially produced over the next decade with big implications for knowledge workers.
A recent report by the McKinsey Global Institute predicted automation of knowledge work would be one of the dozen most disruptive forces in the world economy between now and 2025.
First the good news. These new machines will make knowledge workers far more productive.
”Such tools could both extend the powers of human workers and allow them to offload tedious detail work,” the institute’s report says.
The economic impact of these ”knowledge automation tools” could be between $5.2 trillion and $6.7 trillion per year by 2025 due to greater output per knowledge worker.
Great social benefit will surely follow as the powers of smart, high-skill knowledge workers are extended by machines with the capacity to think for themselves. Maybe the delivery of complex healthcare services could be made faster and better or life-saving scientific discoveries accelerated.
But, like all disruptive technologies, the benefits will not be distributed evenly among workers. New smart machines will be able to perform a growing number of functions that once only humans could do. For example, an executive who might now assign a small team to pull out information needed for a sales presentation could simply ask a computer to do the entire job.
Demand for some categories of knowledge worker might dwindle in the same way typists were made obsolete by word processing programs on desktop computers.
McKinsey estimates that knowledge automation tools could take on tasks equal to the output of 110 million to 140 million full-time workers by 2025. That impact would be disproportionately felt in advanced economies like Australia because wages are higher.
Like previous waves of manufacturing and transaction work automation, the automation of knowledge work will be controversial.
”As computers transform knowledge work … debates about the role of thinking machines in society will undoubtedly intensify,” the report says.
Displaced workers will have to be retrained and those entering the labour market for the first time will need to be equipped to work effectively with smart machines.
Will our education system be up to the challenge?