The reason AI is disruptive lies in the fact that the technology makes it possible to multiply the pace of evolution and innovation. AI is universally enabling and self-learning, especially once the AGI threshold is crossed. Among other things, this enables deep customization, magnified by the fact that AI represents not just a higher intelligence, but a new form of intelligence. By contrast, the precise effects are difficult to predict, as we tend both to overestimate them and to focus on worst-case scenarios.
This is the second post in the series
This blog post is the second in a series about the impact of AI on innovation and development:
- Where does AI impact today?
- Why Is AI Disruptive?
- AI’s effect on robotics and experimental automation
- AI’s effect on climate and meteorological research
- AIs impact on social sciences
- AI’s effect on teaching, learning, and psychotherapy
- AI’s impact on biotechnology and pharmaceuticals
- AI’s effect on materials research and quantum chemistry
- AI’s effect on theoretical physics and mathematics
- What should investment strategies in AI take into account?
LLMs have been historically well-received by users
AI generative services have been met with unprecedentedly positive reception, and this demand pull is only growing stronger. For example, the language model ChatGPT has more than 400 million individual weekly users, andLLMs are increasingly being used in sensitive and decision-critical contexts..
The truly significant perspectives of AI, however, will begin to emerge over the next 2–3 years, as research and development accelerate dramatically. The effects of this will extend far beyond the efficiency boom brought by LLMs and AI agents.'
But the disruptive effect will come from increased research and development
Predictions are generally based on extrapolations of what already exists—the known and the unknown—since evolution is normally a gradual process. It takes time to fully recognize and then harness the potential of the small mutations that form the core of evolution. The unknown usually becomes known over long adoption cycles.
What is unique about AI, however, is that its disruptive potential is both broad and fast-moving. It is a technology that anyone can use. At the same time, it is self-learning and therefore develops exponentially. Finally, it allows for deep customization and will ultimately give rise to a new form of intelligence.
AI is “universally enabling” …
AI is universally enabling because it has the potential to broadly impact all users. At first, it enhances our individual cognitive abilities, mental doping.
Yet AI will “only” deliver more of what we already know at the industry and societal levels, as long as it is used merely to boost the efficiency of individual employees (primarily through LLMs).
AI’s potential is disruptive because it makes it possible to democratize the distribution of knowledge and intelligence. This gives individuals and small businesses new opportunities to conduct research and development at relatively advanced stages with limited capital. From there, however, the real potential for value creation will lie with companies that already have consumer trust (existing relationships) and credibility (brand strength) . These firms are best positioned to bring products successfully to market.
… and it is self-learning, especially once the AGI threshold is crossed
AI is becoming increasingly self-learning as it approaches the AGI stage, and thus the development of consciousness and its own senses. This exponentially accelerates the pace of progress.
What is new about AI is that, like the human brain, it is built on pattern recognition rather than sequential analysis. Traditional computers are based on relational structures, while AI focuses on vector relations. For this reason, the speed of AI’s self-learning depends primarily on software capable of integrating sensory input, as well as on hardware that can execute processes simultaneously.
However, the pace of AI’s self-learning will be asymmetric. Some users will be faster and more effective than others in exploiting the opportunities from LLMs and AI agents to increase efficiency and productivity. Other sectors will experience slower self-learning potential, for example as a result of institutional boundaries (rigid regulation). This will particularly be the case in sectors where safety requirements are high, such as healthcare.
Moreover, AI will change the approach to research and development. Traditionally, a researcher formulates a hypothesis and then tests it. In the future, AI itself will increasingly generate hypotheses and pre-test them. Humans will then decide which hypotheses merit broader clinical trials and development.
This enables deep customization
AI enables deep customization. This is evident in areas such as education and healthcare products, where treatments and products will increasingly converge (see later posts). In general, deep customization will shift the value chain segments where the highest earnings lie (perceived value).
AI breakthroughs will therefore particularly occur in customized products rather than in universal blockbusters like the discovery of penicillin. This creates opportunities for entrepreneurial individuals who can structure and target research and development toward methods for deep customization.
It is a new form of intelligence, not just higher intelligence
AI introduces a new form of intelligence. Yuval Harari , for instance, consistently refers to it as Alien Intelligence rather than Artificial Intelligence. If it were not new, it would not be intelligence—it would merely be machine optimization within familiar boundaries.
At first, AI will uncover new correlations in VERY large datasets, such as climate and meteorological data or psychometric data. These findings will challenge our understanding of natural logic when some correlations prove to also reflect causal relationships. Our perception of the world generally depends on our grasp of logic and of cause and effect (causality).
Although correlation searches can lead to groundbreaking developments, they are above all a form of advanced machine optimization—made possible by a generative, self-learning technology built on massive computational power. But a new form of intelligence means that AI can break beyond the boundaries of our understanding, boundaries that today are defined by the reach of our senses and by our cognitive biases.
The development of AGI especially entails the creation of consciousness, sentience. This presupposes, among other things, the development and integration of senses. In itself, this should lead to entirely new fundamental understandings of the world around us.
Still, the exact impact is difficult to predict, because we often overestimate the effect…
Conversely, it is important to remember that what is new often seems frightening:
New technologies often trigger an intellectual panic. In antiquity, for example, Socrates feared that writing would destroy thought. In the Victorian era, people feared that the telegraph would lead to social isolation, and in the 1980s, many believed that video games would make young people violent.
… and in that process, we often focus on the worst case
With new technologies, there is also a tendency to focus on worst-case scenarios rather than on the overall risk. We mostly pay attention to “tail risks,” even though we know that the future usually falls within the confidence interval. Risk is the product of potential impact relative to probability.
All in all, there are thus opportunities for historically significant leaps in innovation. But when and in what sequence this happens depends, as always, on the capital, courage, and persistence of entrepreneurial investors.