When automation works against output
Robots are no magic bullet for our country’s productivity and skills shortage woes
By now most people would have heard prophecies of a jobless future where machines do all our work and entire industries are displaced due to robotic automation.
In December, a McKinsey Global Institute report found that as many as 375-million workers worldwide may need to switch professions by 2030 due to automation.
However, Elon Musk’s capitulation that excessive automation at his Tesla Motors factories directly caused production delays should ring the warning bells for local firms.
Robots are no magic bullet for our country’s productivity and skills shortage woes. In fact, technology is most effective when it empowers human performance capability, not when it replaces it.
While reports suggest a future where robots take over nearly all aspects of human labour, Tesla’s automotive peers have shown that in many respects there is no replacement for humans as an integral aspect of the manufacturing process.
For example, Toyota has relied on human skills at its Georgetown, Kentucky, assembly plant, de-automating in many respects to rely on humans to install an array of options for luxury car customers for its top-of-the-range Camry models.
Closer to home, local vehicle-making factories rely on tens of thousands of people to produce a range of makes and models at the country’s multiple car manufacturing plants for BMW, Volkswagen, Ford and Toyota. In 2017, the country produced 588,000 units, down from 600,007 in 2016 but showing strong signs of recovery: domestic production is expected to reach 635,000 units, according to recent industry figures.
Instead of excessive robotic automation, local manufacturers are looking into how artificial intelligence (AI) and data can help improve their operations by optimising processes and eliminating waste.
In contrast to robots assuming an entire task in the production process, AI is applied as a prescriptive tool that helps humans instead of replacing them.
AI solutions also don’t assume control over processes. Instead, they take measurements from various sources — machinery and the sensors embedded within such machinery, for example — and show manufacturing staff where the best operating region is for each parameter.
And while it’s easy to become overwhelmed by the sheer number of parameters and prescribed values in the manufacturing process, AI tools can prioritise parameters to highlight the most important ones to adjust and maintain at their prescribed values.
AI has the further potential of facilitating skills transfer between experienced workers and the younger, inexperienced workers entering the manufacturing workforce.
AI algorithms can track and analyse repetitive tasks within the manufacturing production process, for example quality control processes, and match the data generated to a predefined set of variables.
The algorithm then looks for inconsistencies that could point to a fault in production, which operators can quickly rectify.
Any insights the engineer uncovers through this process are added as input to the AI’s processes, which retains the learnings. It’s a 100% true, accurate repository of all process data and insights, including the input from the most experienced humans using the system.
What’s exciting is that these learnings are accessible to any new staff members, meaning there’s an accurate set of best-practices and learnings embedded in the system that can greatly speed up the time it takes new staff to operate at an optimal capacity.
The key lesson for those investigating robotic process automation is that having humans in the loop doesn’t condemn the manufacturing process and may actually be a requisite for success. There is an inherent versatility to human actuation that is difficult to realise with robots, one that is perfectly suited to augmentation by insights gained from AI solutions.
• Cronje is DataProphet MD.