Due to UI, the most at risk is the middle management

The use of artificial intelligence (UI) in companies is often misunderstood as replacing people, while it is actually process optimization. UI allows companies to improve efficiency and productivity, leading to better business results. Instead of replacing jobs, it can help employees focus on more strategic tasks while UI takes on routine and recurrent tasks, emphasizes Darko Butinawho has participated in the development and growth of many international companies in his career – Netceter (Switzerland), Cognism (United Kingdom), One PM (Switzerland) and Securities Grid (United Kingdom), Mimovrste (Slovenia), Trimo UAE (ZAE), Elektro Energija (Slovenia) and others.

The fourth industrial revolution is associated with automation, large data and digital twins that have accelerated and cheaped innovation. With the arrival of large language models (LLM), artificial intelligence also penetrates the field of management. How can UI specifically contribute to better, faster and more thoughtful decisions in the company?

I do not want to reduce the importance of middle management, but the fact is that, using artificial intelligence in all processes within the company, many roles change significantly. On the one hand, many people will have to take on different tasks, as some parts of traditional management, especially middle management, become less needed. On the other hand, customer relationships will be increasingly important. These changes will be so large that they will also affect broader social relationships. The difference between those companies – as well as individuals – who will be able to use the UI effectively and those who will not be huge. Using UI tools has been proven drastically increasing productivity. This means that a company with the same resources can do more or reduce costs, which makes it more profitable. If the company is more profitable, it also has more money for investment.

What impact does the UI have on strategic decision -making?

If the strategy is updated – and is helpful to help -, many decisions can be transferred to people closer to customers. Because they know what framework they should decide, they no longer need constant confirmation of superiors. This is how you relieve the middle management, which has so far been responsible for transferring the strategy. But, hand to heart, in larger companies often get stuck – information is not flowing because middle management creates its own gardens. However, if you have a strategy that UI is constantly refreshing and communicating forward, you may even no longer need that work of management.

Are we already at the point where the leaders will trust UI in making important decisions?

Trust is the key. Of course, this is not a blind trust – UI is based on the principles of statistics, which means that there is always a certain degree of uncertainty. But if you put the model properly, it can be more reliable than a person. Let’s say: research has shown that on average, doctors correctly make a diagnosis in 70 percent of cases. UI reached 97 percent with the same data. So – why not use it as a « junior associate », which you first control it, and then gradually trust him?

How, specifically, is used in companies today?

I divides this into four areas. The first is personal productivity – let’s say that Copilot helps you write a letter. The second is the renovation of internal processes – the process you needed five people before, you can now do better and otherwise with fewer sources. The third area is aimed at improving products or services with UI elements – for example, improved customer support or faster processing of demand (such as triage of legal matters). The fourth is the development of our own UI solutions in the products or services themselves – with the UI becomes a key part of the solution that the company provides. In all these cases, the company can work faster, more efficiently and with fewer errors.

Data security is one of the key concerns of companies, especially when sharing internal data with external UI providers.

Of course there are reservations. But we remember the time when people have doubted the security of clouds. Now all companies use Office 365, Google Workspace and other Saas platforms. The same will happen with UI – companies will learn to choose trusted providers and understand what is happening with their information. Most UI leading tools are on big players such as Openai, Google, Microsoft. If the company does not want to share data outside, it will have to have technical knowledge to drive the tool within its systems – but there are very few such companies.

Businesses often care that they do not have sufficient quality data that would be suitable for UI processing.

The data is never complete. However, you can do analyzes with them – this applies to humans and artificial intelligence. If a person can do a meaningful analysis based on this information, then this can do so. Of course, you must first prepare the data properly – clean, normalize, check that it is correct. And even with that, I can help today. Not only in analyzing, but also in the quality of data quality. In some cases, UI is even faster and better in detecting errors than human.

Which tool to choose in the flood of domestic and foreign providers, many times young businesses?

Deciding which tool to choose must always come from the company’s strategy – first you have to know what you want to achieve and then choose the right tool.

The implementation of the UI must not be carried out in terms of « introduce chatgpt, but then we will see. » No. First, it is necessary to look at your strategy, to define the processes you want to optimize or change, and then choose the tools that fit it.

As for the tools themselves – I would recommend one today, but maybe another in three months. Development is rapid. How to check? You are looking for references, business model, financing rate … In short, as always – you trust, but you also check.

What is the biggest challenge in tool integration?

If the model you rely on suddenly starts acting differently, it can be a big problem. Especially if they are integrated into it. Model coaching is not like writing a code – if something is wrong, you can’t just fix one line. If the model is incorrectly trained, you often have to train it again – with new parameters, a different data set, or even a completely different approach.

And more – for most users there is still a black box. You can’t know exactly why the model decided the way it was. You cannot always reconstruct the path to the final answer. This is one of the great restrictions on the current development on the one hand, and on the other, this also hides a great resemblance to us (people).