Overcoming internal AI challenges

What are the internal barriers slowing adoption of artificial intelligence and how can these be overcome?

So you’ve heard about how artificial intelligence (AI) can bring efficiency and other benefits to different aspects of your manufacturing processes. You’ve identified areas of your business that need automating. What’s the next step you need to take?

It’s not enough to simply invest in intelligent vision systems and sensors, install smart software to help monitor the health and performance of machinery, and then sit back and let the magic happen. It’s much more complicated and failing to take the right action could pose barriers to being able to implement AI successfully. Here are some of the main challenges.

By 2020, the discrete manufacturing industry, which is related to products that can be counted, seen or touched, will be spending £35 billion on the internet of things (IoT). This will account for 16 per cent of the total £220-billion spend globally, according to 2017 research by Boston Consulting Group.

As manufacturers invest more in IoT, there will be an explosion of connected devices and sensors on assembly lines and in production plants, all constantly communicating with each other and generating large volumes of data. However, the more connected factories become and the more devices and sensors in communication, the more robust manufacturers’ IT infrastructure will need to be to cope with the wealth of data that needs to be processed.

“Simply put, more data means more opportunity for insight and therefore more value. But more data also means more modelling and that will require access to smarter, more powerful computing,” says Tate Cantrell, chief technology officer at Verne Global, which provides datacentre solutions on an industrial scale, supporting companies to deliver intensive, machine-learning applications.

By upskilling employees and developing a long-term strategy for digital transformation, manufacturers can create more efficient, productive factories that will reshape the industry

Mr Cantrell argues it’s imperative that manufacturers consider high-performance computing (HPC), particularly in the cloud, as it enables AI application to return results more quickly.

To explain the way it works, it isn’t too dissimilar to connecting the dots on a piece of paper. The more dots there are, the longer it will take to draw a picture. Likewise, the more data points, the longer it will take to connect them up, process the data and then return the results, although we’re talking a matter of milliseconds.

“By choosing a cloud-enabled approach to HPC, users can scale up their overall computing capabilities. And by incorporating AI into HPC applications, companies can ensure even smarter and more efficient computing,” says Mr Cantrell. “Applications that are built with the cloud in mind enable companies to stay ahead of the innovation curve.”

Those manufacturing businesses that fail to invest adequately in their IT infrastructure, Mr Cantrell concludes, end up limiting their ability to push boundaries and will fail to realise their ambitions.

Once you have the data, you need to be able to make sense of it. While software can do the crunching, present unstructured data in digestible formats, offer insights and even make recommendations, it can’t tell you what specific actions you need to take.

If manufacturers are to act on the AI-driven insight that their IoT ecosystem is generating, then they need to rethink their workforce and the type of people they are hiring.

A 2017 report by Engineering UK found there will be demand for an estimated 186,000 engineers every year until 2024. Many of these jobs will be in roles such as automated system engineers and manufacturing software engineers. As such, a new breed of manufacturing employee will be needed, who is tech savvy with an interest in data-driven decision-making.

“Job roles will inevitably change and some will be made obsolete, but humans will remain fundamental. Retraining and upskilling staff to work alongside machines will be key to ensuring [manufacturers] have the necessary skills to work efficiently,” says John Kirven, senior proposition consultant at Canon. His role is to support manufacturing businesses by understanding their processes, challenges and goals, subsequently linking them with the technology and services that meet their needs.

The type of skills likely to be in demand are an ability to operate automated manufacturing systems, ability to work with computerised systems, and ability to read and write code. A broad understanding of machine-learning, predictive analytics and algorithms is also likely to be required.

To close the skills gap, Mr Kirven believes manufacturers will need to engage with universities and colleges. This will help in demystifying any misconceptions young people may have about manufacturing that jobs are low paid, or there’s little room for professional growth or progression. The truth is it’s a high-value industry.

Another benefit of engaging with educational institutions is reaching young people, who may not have previously considered a career in manufacturing, with transferrable skills.

“AI has the ability to streamline processes and boost productivity, but this can only be achieved if manufacturers utilise machines to their full potential and, in doing so, invest in their staff as well,” says Mr Kirven.

“Ultimately, by upskilling employees, bringing in the necessary expertise [and talent], and developing a long-term strategy for digital transformation, manufacturers can create more efficient, productive factories that will reshape the industry.”