Sundeep Sanghavi from Data RPM looks at how machine learning and predictive maintenance could erase recalls for good
Oh dear, Samsung. It’s not been a year to remember, so much as a year to learn from, hasn’t it? While it scored one small victory in the Supreme Court today against Apple, in general, 2016 has been full of relentlessly bad news for the tech titan.
After its flagship phone the Galaxy Note Seven caused fires, Samsung was forced to issue global recalls with US authorities taking action to prevent the dangerous cell phones from being taken on-board aircraft.
Later, the Korean electronics giant faced another recall debacle: washing machines. Following reports that top-loading devices were exploding, Samsung recalled 34 models — and some 2.8 million units — to prevent further damage and injury. What makes this worse is that the incident is effectively a repeat of an (albeit smaller) recall that Samsung issued for washing machines in Australia three years ago.
Still, Samsung is hardly the first, and will unlikely be the last, business to issue a recall due to faulty products. Pharmacist Johnson & Johnson was forced to recall 31 million bottles of Tylenol in 1982 – worth $100 million – when seven people died in the Chicago-area after products were tampered with cyanide. More recently the Takata airbag debacle saw models from a range of car manufacturers recalled after faulty products designed to keep people safe resulted in deaths. Clearly, problems with recalls are endemic.
Using data to end recalls
The thing is, the reality does not have to be like this. There is no need for businesses to issue recalls routinely, or for consumers to suffer the potentially dangerous consequences of dodgy equipment. Predictive maintenance — that is using artificial intelligence (AI) and machine learning (ML) to predict, ahead of time, when things will go wrong — can drastically reduce the chance of things going wrong, and therefore the likelihood of recalls occurring.
Big players already forecast that predictive maintenance can put an end to recall culture. According to reports from Big Four consultancy group McKinsey, predictive maintenance could save global businesses an incredible $630 billion a year by 2025. Despite being a complex subject, the reason for this outcome is relatively simple.
This is because, through a process of so-called meta-learning, machines will be able to learn from previous results, leading to developments that can prevent recalls and save lives. Machines on factory floors will, with the power of AI and ML, be able to foretell problems in machinery and prevent them from becoming serious problems.
As things currently stand, even the most forward-thinking businesses are not using data from their machinery to its utmost potential. Currently, those businesses employing technology — such as sensors to report output — are still employing human teams, complete with human errors, to filter through the signals.
Yet with a combination of ML and AI, these sensors can be monitored by advanced machines that can provide businesses, such as Samsung, with the prescriptive information needed to stop disaster from happening. Only by properly employing machine learning can businesses reverse their fortunes when it comes to costly, and reputation damaging, recalls.
The views expressed are of the author.
Geektime invites global tech and startup professionals to share their opinions and expertise with our readers. If you would like to share your point of view, please contact us at [email protected]