The article was first published in Top Engineer magazine.

Industry’s interest in artificial intelligence (AI) has surged dramatically with the emergence of generative AI and the popularity of large language model based tools, such as ChatGPT. However, we are far from seeing AI’s full potential. World-class AI experts from Silo AI advise companies to direct AI investments towards their core business to generate real and enduring value – as a technology, AI benefits significantly from economies of scale.

Silo AI is the largest private AI laboratory in Europe, employing about 300 people worldwide. The company’s strength lies in deep AI expertise, exemplified by the fact that more than half of its staff members hold PhDs, and by an extensive track record of applying AI across multiple industries.

Silo AI was born when its founders recognized the ongoing AI talent drain in Europe, as the US and China were spearheading AI development. If Europe were to retain AI talent in the region and maintain its technological competitiveness globally, the need for a European counterpart to the AI companies in the west and east was imminent.

AI is now at the same stage as aviation was over a hundred years ago.

The idea was partly rooted in the distinctions in value systems between countries, highlighting the necessity to incorporate the European value system into AI development, but also to serve practical product development needs across industry by connecting top academic AI research with engineering talent. Since then, Silo AI has built AI solutions with hundreds of customers’ for bespoke use cases across multiple industries, guided by their vision to build AI for people and focusing on where AI can create real value for their customers.

We had the opportunity to interview Niko Vuokko and Jaakko Vainio, Chief Technology Officer and Chief Operating Officer at Silo AI, both leading experts within AI with decades of experience between them. Vuokko leads the Smart Things business unit, developing solutions for tangible objects, such as mining machines, while Vainio is responsible for the Smart Applications unit focusing on digital business and cloud-based solutions.

How do you define artificial intelligence?

Vuokko: Artificial intelligence is a vast field and far from what one person could comprehend. One concrete implementation of it is large language models – a trend that has been in the headlines but is still only a small part of AI.

Vainio: There has been a significant increase in interest in generative AI. In the background, scientific development is progressing rapidly, and new models are constantly emerging. However, we mainly think of AI in terms of the value it produces for the customer. For example, we create neural networks, but we also use generative AI and large language models, or even more traditional methods if a customer has very little data.

Vuokko: One could say that AI is largely what will provide a competitive advantage in the future: the core of how technology should be understood. If you look at history, the machine industry electrified in the past because electricity made it easier to add functionalities. Then came electronics, and next, devices could be connected to the internet, allowing software to be integrated into them. AI is the next major revolution, and soon it will be everywhere, as it simplifies and accelerates the achievement of better results.

At what stage is the industrial utilization of AI currently?

Vainio: It depends a lot on the industry, companies, and their sizes, but we are not very far along yet. There are exceptions, of course, and there is already quite advanced usage. Before AI can be utilized, a company needs to do its homework on digitalization: there must be data and IT infrastructure in place to build on top of. One significant factor is that there is still a shortage in AI talent. Companies should also learn to recognize AI’s possibilities better.

Vuokko: For many valid reasons, the industry is slow to adopt new technology. AI is now at the same stage as aviation was over a hundred years ago. The first commercial airplanes could carry two to three passengers, and air travel was still a bit exciting. But when people managed to reach their destination by air, it was a magical experience, which has nothing to do with how aviation is perceived today. It tells us that we are facing many decades during which the scale of AI will increase a thousandfold.

How does Finland fare in the competition?

Vainio: Finland is not at the forefront of AI development, but we are not lagging behind either. This is a small market, and there are few large companies here. As we know, significant product development projects require large investments. On the other hand, Finland is one of those countries that perform well relative to their size.

Like other countries, we face the challenge that companies are conducting proof-of-concept experiments that end up waiting for the right value creation mechanism. It is not enough to have the right AI model; the solution must also work in the right place at the right time. For example, the service should not crash overnight if that is a requirement.

It is not worth spreading the grains everywhere; choices must be made, and investments concentrated accordingly to reach a large scale.

How can industry promote the green transition through AI?

Vainio: The green transition requires a lot of action across society. In some cases, AI and machine learning models work well, such as in improving resource efficiency or logistics. For example, production can be optimized to reduce waste or goods can be moved efficiently from one place to another, which reduces fuel costs. AI also brings tools to the energy markets that make pricing and transmission of renewable energy more efficient.

Vuokko: The green transition means doing things smarter and controlling things at a more detailed level than we are used to. AI allows industrial processes to be manipulated with more complex goals. After all, it is very challenging to balance all the numerous goals: quality, deliveries, process safety, sustainability, to name a few. AI enables decisions to be made at a microsecond level and to continuously fine-tune industrial processes.

Vainio: AI also works as a planning aid. One new area is energy-efficient machine learning models, as computing often consumes a lot of energy.

What are the biggest challenges to widespread AI utilization?

Vuokko: The core challenge is that big disruptions happen very rarely. When a revolution comes, the way of doing things changes, and that is when the actual benefit arises. The more agile a company is, the easier it is to embrace new technology. When the revolution is over, the goal is to optimize the current state and cement the processes, including procurement principles and everything else that keeps the machine running efficiently in that state of the world – the idea is to reject changes that make the company inefficient. But with a new revolution, operations need to change.

Vainio: Companies often haven’t identified what the broader path to leveraging AI at the strategic level is. Typically, they want to reduce costs a bit and keep up with others. There are, however, various options for AI utilization, and what is best for your business is not self-evident unless you have the expertise or seek it. It is extremely easy to fall into a trap if you only look at the easiest and most obvious use cases that do not provide much help.

Vuokko: AI is one of those technologies that benefit significantly from economies of scale. Individual pilot projects can be disappointing because their scale is so small that they don’t achieve economies of scale. On the other hand, companies must also understand that the benefits emerge when significant investments are made. It is not worth spreading the grains everywhere; choices must be made, and investments concentrated accordingly to reach a large scale.

Do companies have fears about AI?

Vainio: Fears are usually misplaced. For example, machine vision solutions often involve installing cameras in industrial facilities, which can raise concerns since they are wrongly associated with employee surveillance. Cybersecurity is also one thing that is discussed a lot, although AI does not pose a greater security risk than other technologies.

However, there are few concerns that everything provided as input to ChatGPT is shared outside the organization. Nothing related to the core business should be put in an external service. The ethical questions regarding AI are also naturally important when discussing its applicability in society as a whole.

What tips do you give to companies that want to make progress in utilizing AI?

Vainio: As a rule of thumb, AI should be utilized primarily in your own core business. Support functions may yield small benefits, but for example, an energy company’s biggest efforts should go into energy production or distribution. You also need to understand how to get started and learn to identify use cases. If you don’t have the capabilities for this, you can start with the help of a consultant.

Vuokko: Two things need to happen simultaneously. One is how to build essential core capabilities needed to implement change in the company. In addition, there is a need for ideas on how to bring the technology to customers, going into another cycle that genuinely creates value. Many companies push forward for months without thinking about how to get the concept into the hands of the customer and, thus, fall into a pit. So, you must simultaneously consider the prerequisites and implement the concept in practice. This also leads to learning.

Vainio: For smaller companies, the good news is that it may even be easier to adopt AI when starting from a clean slate. For companies without data centers, solutions can be built using innovative technologies directly in the cloud.