Topleadership of AI disruptions

This is the first part of three on top leadership of AI disruptions

AI is more than merely a new technology. It is a universal catalyst that enhances human functions, cognitive, communicative and creative, for all individuals willing to use it. Compared with previous technological disruptions, AI has a far more intrusive potential, because it fundamentally alters the conditions for leadership, judgement, capabilities and competition. For this reason, AI adoption is also most effective when it happens bottom-up.

The task of top management is therefore primarily to encourage and motivate the use of AI, within carefully considered security frameworks.

AI has particularly disruptive characteristics …

Today, AI is a universal tool for enhancing human capabilities in analysis, formulation and experimentation. It democratises knowledge and enables individuals and small units to conduct research and development that previously required entire research organisations. Going forward, we can therefore expect major breakthroughs and economic effects from, for example, Inverse Design, In Silico research or Robotics sensoring. 

However, AI will affect society more broadly and far more unpredictably. It differs from previous technologies in that:

• AI becomes increasingly self‑learningthe more it is used and the more data it learns from.
  • The closer it comes to Artificial General Intelligence (AGI), including the development of its own senses and its own consciousness, the faster the pace of development accelerates.
  • Today, AI is still “merely” statistical models applied to knowledge bases, but once the AGI threshold is crossed, AGI will itself become a knowledge base. This will dramatically accelerate the speed of learning.
• AI operates through pattern recognition rather than sequential processing, similar to the human brain. This means, among other things, that AI will introduce a new form of intelligence (“alien intelligence”) once it passes the AGI threshold. Otherwise, it is not intelligent.
  • AGI solves problems by autonomously assessing which problems are worth solving. This creates the possibility of breaking through the boundaries of our current understanding, which today is constrained, in particular, by the limits of our senses and by our cognitive biases.
• AI is based on vector based data coordination. This means that answers are generated contextually, are user‑specific, and are created from scratch each time. As a result, deep personalisation can be achieved at very low marginal cost.

… and a large but complex potential, …

These characteristics make it difficult to predict the effects of AI. Projections are further complicated by the fact that AI adoption will be asymmetric. Some individuals will be quick and highly effective in exploiting the opportunities, while others will be more cautious. Certain industries will also be constrained in their ability to apply AI, for example due to rigid regulation.

Nevertheless, the potential is both large and evident, and the technology has therefore found broad and rapid application, especially in knowledge‑intensive domains.

• In mid September, for example, Bundesbank President Joachim Nagel spoke at the BIS Innovation Summit : “At the Bundesbank, generative AI is now broadly available for our staff to use. This technology is already augmenting existing processes in almost every business area and complementing previous work. From document drafting to advanced data analysis, AI applications are not only supporting us in areas like banking supervision or economics, but also in a wide range of internal and administrative services.”
• Similarly, the Vice Chair of the Federal Reserve, Philip Jefferson,argues that AI marks the beginning of a new era in which future tasks will look fundamentally different. The technology will become an integrated part of human development, with a “general‑purpose” impact on the economy.
• The views of central banks are also shared by finance ministers. Ahead of the G20 meeting in October, the BIS concluded that AI is not only a technological issue but also a political and leadership challenge. AI can change the structural and functional modes of the economy through, for example, self‑service solutions, automation and data analysis. This opens up new applications and new types of tasks. The BIS concludes that success above all requires robust governance and a willingness to invest in both equipment and people.

… for which expectations are extremely high

In several areas, expectations of AI’s usefulness are so high that it is even seen as a potential structural response to demographic challenges (such as ageing populations) and to shortages of key skills.

This also applies among AI developers themselves. At Microsoft, more than a quarter of the company’s code is already written by AI today. Satya Nadella expects that 95% of all code will be AI‑generated by 2030.

The enthusiasm for AI is also evident in countries without a tradition of technological leadership.

• Saudi Arabia, for example, has worked intensively to persuade OpenAI to establish data centres in the country and is seeking to attract TSMC and Intel to manufacture chips locally, even though very few residents currently have sufficient STEM education merely to maintain such facilities.
• This year, Albania also became the first country to appoint an AI‑generated ministerial figure, Diella, with responsibility for public procurement. This illustrates how deeply we, as humans, accept and trust AI, and how quickly and unpredictably the technology is moving into the structures of society.

The development towards AGI is progressing rapidly …

Most of the boundaries that were previously considered almost impossible to break have now been crossed:

• The ability to reason and solve problems reached a breakthrough in 1997, when IBM’s Deep Blue defeated Garry Kasparov in chess.
• Natural language understanding did the same in 2011 with IBM’s Watson. Today’s generative models can even understand nuance, idioms and humour.
• AI now creates art, music and text that challenge the idea that creativity is unique to humans.
• Real‑time perception, enabling self‑driving cars and robots to navigate complex environments.
• Emotional simulation, where chatbots can imitate emotional understanding (EQ), albeit still from a superficial and analytical perspective. According to a recent study by the RAND Corporation, one in eight young people in the US uses AI chatbots for mental health advice. , and 93% of them believe they receive useful guidance.

… even though highly complex challenges remain

The remaining challenges include in particular:

• Hallucinations, where models invent facts. Techniques such as Retrieval‑Augmented Generation (RAG) reduce the problem, but do not (yet?) prevent it effectively.
• The development of senses is necessary for AGI to emerge. Interoception and proprioception are two of the major organising principles of intelligent behaviour that are still missing. Today, they represent a form of “glass ceiling”. They are required for robust and self‑regulating AGI.
  • Most AI models assume “cognition without physiology”, whereas biology is better described as “cognition because of physiology”.
  • Without self‑awareness, AI lacks the ability to consistently assess what is “right” or “good”, as well as what constitutes “common sense” and contextual understanding. This also precludes genuine emotions such as joy or grief.
  • From a technical perspective, however, this is probably only a matter of time.
• Security. AI models are enormous and therefore run on large, centralised data centres. This creates risks related to lack of redundancy (for example in the event of outages) and broader data security. Smaller and more specialised models may offer a solution. However, it will take more than five years before AI solutions can become sufficiently decentralised and portable to operate offline. The demand for data volumes is enormous.

To be continued in the next blog post