Towards the end of 2024, it looked like the hype around AI was cooling down. The authors of this Outlook even contemplated what to focus on beyond AI. Then 2025 started off with a bang! The Chinese startup DeepSeek launched a large language model (LLM) which fundamentally challenges Silicon Valley on both price, computation requirements, and functionality. The moats for LLMs are hereby declared gone with the wind. In addition, Donald Trump took office as POTUS, and the Silicon Valley tech bros all lined up to shine in the fringes of Trump’s spotlight, ready to fuel the global AI arms race. AI is now global politics, a question of power and money. The hype and noise around AI are as loud as ever, and there is a strong need to help leaders navigate wisely and sensibly.
First, we need to separate out the political noise and commercial hype from the pragmatic reality of mathematical algorithms applied to large amounts of data (aka AI). AI is the natural next step of digital transformation, where we apply mathematical algorithms wrapped in software to data and digital processes for the purpose of automation and optimization.
Large language models opened the lid of a treasure trove of opportunities for completely novel solutions and approaches. However, it is still about mathematics, not magic. The Upwork Research Institute [1] found that 96% of executives believe AI will enhance productivity. At the same time, 77% of employees find that AI reduces productivity. Is this a story about useless technology? The authors of this Outlook offer a different perspective; through their interactions with thousands of leaders across industries across Europe, they find that typically less than a quarter of the executives present in the room are personally using AI tools for professional purposes on a weekly basis. This indicates a fundamental leadership problem; boards and executives are making strategic decisions about the use of AI without having the required knowledge about what AI is (and not), what it can do (and not), and how to successfully apply this toolbox to enable the business strategy. AI adoption is not about ChatGPT or Copilot licenses. It is about understanding the dynamics of large amounts of data in automating and optimizing an organization's processes and the opportunities and risks involved in using prediction models in decision-making.
From the perspective of leadership and strategy, value-creating AI literacy will have to start at the top of the organization. The EU AI Act Article 4 [2] on requirements for providers and deployers of AI systems to ensure sufficient AI literacy came into force on Feb 2nd this year. However, AI literacy requires a certain level of knowledge about mathematics, statistics, and information theory. Not a lot, but more than a crash-course on prompting. This is challenging. At no point in history have boards and executives stood in front of such an urgent and absolute need to learn hard skills.
https://artificialintelligenceact.eu/article/4/
Problems can be calculated. AI may be a useful toolbox for this purpose, but sometimes even overkill. Conventional analytical methods may be more than sufficient.
Secrets can be predicted, and this is where the AI toolbox may be especially useful. Advocates of generative AI emphasize its ability to automate simple or repetitive tasks. By handling time-consuming routine work, AI is expected to free up users to focus on more complex and strategic challenges. However, executives must evaluate whether the output of generative AI, which often requires significant refinement, genuinely results in labour savings.
Mysteries, however, are by nature almost impossible to predict, as they depend on the dynamics of human interactions and intuition. This is where human leadership and human capabilities are essential. Many jobs in the categories ‘problems’ and ‘secrets’ will be displaced and replaced by AI, just like with any other technological advancement through history. The mysteries, however, are almost impossible to replace by digital technologies. This is where leaders should prioritise their leadership efforts.
We invited Arnulf to a podcast to dig deeper into the challenges and opportunities of leadership in the light of AI
Listen to our podcast with Jan Ketil Arnulf
[1] https://investors.upwork.com/news-releases/news-release-details/upwork-study-finds-employee-workloads-rising-despite-increased-c
[2] https://artificialintelligenceact.eu/article/4/
[3] https://www.ketilarnulf.no/en/home-2/