%page_title%

This article was translated and originally appeared on polskiprzemysl.com.pl

The topic of artificial intelligence has become firmly established in the public sphere, even becoming a trend, although the level of knowledge about these technologies remains relatively low. While Generative Artificial Intelligence (Gen AI) is still a new phenomenon, other types of AI have been around for decades. A good example is systems based on human-like intellectual abilities, such as abstract reasoning or common sense, which have even surpassed human abilities. On the other hand, some types of AI are still unclear and remain topics for the future.

Just ten years ago, certain types of AI were a technological curiosity used by a few companies, and scientists tested the usefulness of algorithms to play and win logical games against humans. Since then, artificial intelligence has come a long way and with new solutions such as ChatGPT, GitHub Copilot for developers, Sora and Midjourney, AI is now on everyone’s lips and companies are racing to find new ways to use it.

However, there is a significant difference between how consumers use artificial intelligence and how businesses use it. Challenges such as budgeting, ensuring security, analysing use cases, focusing on return on investment and the need for specialised skills are taking up entrepreneurs’ attention, leaving little room to observe the buzz around this phenomenon. Measurable productivity, process efficiency, labour optimisation, cost reduction and revenue growth are paramount.

According to BCG’s report, From Potential to Profit with GenAI, executives rank AI and Gen AI among their top three technology priorities for 2024. Fifty-four percent of respondents expect artificial intelligence to deliver cost savings this year. Of these, around half expect savings of more than 10%, mainly due to increased productivity in operations, customer service and IT. It should be noted, however, that only 6% of executives have begun significant upskilling in AI and Gen AI, while a substantial 66% are ambivalent or dissatisfied with their progress.

Looking at the automotive industry, the Zebra Technologies report ‘Over Half of UK Automotive Industry Companies Now Using AI for Machine Vision’ reveals that 43% of leaders surveyed in Germany and 56% in the UK are currently using some form of artificial intelligence, such as deep learning applied to machine vision projects. On the other hand, 34% of leaders in Germany and 24% in the UK say they are not using any form of artificial intelligence in similar projects and do not see a need for it. There are also executives who expect AI to do more for them and deliver better results.

Advances in machine vision are an example where, as in other industries, manufacturers are at different levels of maturity in terms of achieving results through artificial intelligence. Modern machine vision solutions take analysis to a whole new level of accuracy, compliance and quality in production processes, and provide engineers with new tools to work more efficiently, demonstrating the true value of the technology.

At one of its production sites, the Bosch Group develops solutions for diesel engine injection systems for the automotive industry. Injection nozzles are a critical component in delivering diesel fuel to the engine’s combustion chamber. Bosch needed a vision system to further automate reading and verification processes, improve nozzle identification and reduce the number of parts that needed to be manually inspected.

With this solution, the plant achieves a production level of 7,000 parts per day. The percentage of faulty rejects has dropped below 5%, a significant improvement. The system uses vision software to control the entire process, allowing the team to reduce costs while achieving faster and easier configuration.

Artificial intelligence (and users) need training

Deep learning neural networks are powerful, advanced AI tools that mimic the human brain – specifically, in the case of machine vision, convolutional neural networks that emulate the brain’s visual cortex, which processes images.

Engineers sometimes expect neural networks to work flawlessly, but it is important to remember that they have limitations. It is important to communicate this to all stakeholders. Neural networks can achieve remarkable results, but they need to be used carefully. Realistic expectations should be based on areas where they excel beyond human capabilities or conventional rule-based machine vision. Examples of such areas include surface defect detection, object recognition or counting, reading difficult characters, or detecting unexpected deviations (anomalies) from a pattern.

There are many data-related issues that need to be addressed for an organisation to benefit from artificial intelligence. Mixing training and test data sets, insufficient and uneven sample sizes, ambiguous and inconsistent data labelling, and environmental factors must be taken into account to ensure that deep learning-based solutions work properly.

A plan is needed to harness the potential of AI

With the adoption of the EU Artificial Intelligence Act, businesses need to cut through the noise to understand the value that AI can bring to their operations. The EU AI Act establishes a common framework for the use and deployment of AI systems in the EU, including the classification of AI systems with different requirements and obligations tailored to a risk-based approach.

The Act acts as a new catalyst for manufacturers to invest in the partnerships and technologies needed to create digital factories and develop intelligent manufacturing operations. More automated and autonomous workflows, better supported workers, and predictive and prescriptive analytics can be facilitated by AI and the vast amount of valuable production data.

What manufacturing process needs automation and could benefit from AI? What type of AI would be most appropriate? How will compliance be ensured and documented? What staff and partners will companies need to make this happen? These are questions that are not easily answered by the current buzz around AI, but they need to be addressed.

Progress without unnecessary hype

At the moment, it’s not a question of whether artificial intelligence will create or eliminate jobs, despite what many media headlines suggest. Like cars, phones and the internet, the development of AI will create many new jobs and industries. What we are seeing now is manufacturers equipping their engineers, developers and data scientists with new and improved AI-based tools. This enables them to do their jobs faster and more efficiently, and to automate certain tasks using AI solutions.

Manufacturers, like other industries, may face challenges in hiring, training and retaining employees. In these cases, business leaders are turning to automation to fill staffing gaps, speed up employee training, and support existing teams. Employees who leverage AI capabilities will stand out because they have the knowledge and experience that manufacturers need in their facilities.

Companies will also prioritise the democratisation of AI and machine learning as a strategic priority. Whether it’s engineers, data analysts or developers, employees will be upskilled and given educational resources and out-of-the-box AI tools that take over some tasks and support them in others. Some solutions can be built to low/no code standards, meaning they are ready to use out of the box and do not require specialised training. An example of such a system is a deep learning OCR tool. Other tools are more advanced and function more like out-of-the-box environments for developers and data analysts. They are used to create solutions using the platform, tools and libraries provided.

Ultimately, upskilling employees and optimising front-line operations with new ways of working will become the norm, not the exception, in the war for talent. Those who can adopt and utilise new AI tools without being distracted by the hype will secure a future advantage for themselves and their customers.