AI may have had its false starts but now the capabilities it offers looks set to change the nature of work for many in a new industrial revolution, writes Sopra Steria Sweden’s Felicia Ljungvall.
Around 200 years ago, three out of four Swedes were employed in agriculture. Since then, we have witnessed three industrial revolutions: mechanisation, electrification, and computerisation.
These shifts have completely transformed the nature of work in Sweden. Now, as little as 2% of all Swedes work in agriculture, while 75% are employed in the so-called service sector, doing jobs that didn't even exist 200 years ago.
Today, new AI services are launched almost daily, and some are saying we are now entering a new industrial revolution, with AI at its heart. In the first episode of our new YouTube series “The Digital Miracle,” we ask the question: will AI create a new revolution, and if so, how will it affect what we do for work?
The development of AI
Like many other technologies, AI has gone through various stages over time. There have been high expectations and hopes for AI that were not met and did not deliver the desired results.
This led to a reluctance to invest in the technology, resulting in little development. However, about five to six years ago, there was a sudden rapid development in AI with the availability of vast amounts of data and information and new capabilities created that significantly improved AI’s functionality.
Since then, AI has acquired the ability to take over more cognitive functions traditionally performed by humans -- capabilities we hadn't previously managed to digitalise. Algorithms have become more intelligent, sometimes even more so than humans. This is primarily what has driven the new AI revolution.
Which job roles are affected by AI?
The extent to which AI affects different professions varies. Örebro University has developed a model that examines this. It is based on 52 different abilities which can be categorised into four ability groups: cognitive, physical, psychomotor, and sensory. Professions are then assessed on how important these 52 abilities are for each role.
Using this model, researchers have found that professions that are more cognitive, non-physical, and non-social are the most exposed to AI. Examples include economists and developers.
Firefighters and nurses, on the other hand, are less exposed to AI because their jobs are more physically, psychologically, and socially demanding.
How are jobs affected?
To get a clearer picture of how AI will affect our work, we can look at substitutive and enabling technologies. Substitutive technologies make certain jobs and skills redundant, an example being the loom, which almost entirely replaced the need for weavers.
Enabling technologies, on the other hand, make people more productive in their existing jobs and can create entirely new jobs. For instance, the invention of the telescope created new sciences and tasks without replacing previous jobs.
So, will AI be substitutive or enabling? Probably both.
A clear example is the ability to recognise tissue changes in radiological images. Machine learning can analyse many images and determine if there is a cancerous tumour or a tissue change.
In such a case, AI has an error margin of approximately 7.5%. A human and trained radiologist has an error margin of about 3.5%, which means the human performs better in this instance.
However, if you combine AI's decisions with human decisions, the error margin can be reduced to 0.5%. This demonstrates that humans and AI have different abilities to detect, for example, a cancer tumour, and that the combination of both can lead to enormous improvements in quality.
The insights in this blog post are based on interviews with Daniel Akenine, National Technology Officer at Microsoft, and Erik Engberg, PhD candidate in Economics at Örebro University.