Page 20 - The Annual AI Governance Report 2025 Steering the Future of AI

The Annual AI Governance Report 2025: Steering the Future of AI

predictions on what percentage of tasks can be reliably and profitably replaced by generative
AI and at what timeline. Similarly, experts are uncertain what types of productivity gains can be
achieved by new discoveries, such as “new materials, new drugs, or new services.” McKinsey
32
research sizes the long-term global AI opportunity at $4.4 trillion annually, or a contributed 3.7% Socioeconomic Theme 2: AI and
of global GDP growth, in added productivity from corporate use cases by enhancing productivity
across industries, especially in customer operations, software engineering, marketing, and
R&D. In contrast, studies from MIT indicate a much more conservative growth estimate of
33
0.7% annually, suggesting a “nontrivial, but modest effect.” 34

New Value Streams in Education
There are emerging value streams in education by enabling more personalized, adaptive,
and data-informed learning experiences. Tools such as Intelligent Tutoring Systems (ITSs),
Natural Language Processing (NLP), and Automated Performance Enhancement (APE) systems
offer the capacity to individualize instruction and assessment, allowing educators to respond
more precisely to diverse student needs. These technologies may contribute to improved
35
educational outcomes by supporting real-time feedback and streamlining administrative tasks
such as grading, thereby reallocating educators’ time toward more complex pedagogical and
mentoring activities. Furthermore, AI-enabled analytics and planning tools are beginning to
inform curriculum development and institutional decision-making by aligning educational
content with student performance patterns, local priorities, and anticipated labor market
demands.

Additionally, AI is contributing to the advancement of immersive learning through technologies
such as augmented and virtual reality (AR/VR), which are being explored for their potential to
simulate hands-on experiences in disciplines ranging from science to the humanities. These
tools may facilitate more engaging and context-rich educational environments, particularly in
remote or under-resourced settings. AI also supports the development of lifelong and flexible
learning pathways, potentially expanding access to upskilling and reskilling opportunities in
rapidly evolving sectors such as data science, healthcare, and cybersecurity.

New Value Streams in Science

In 2024, for the first time and likely not the last, the Nobel Prize was awarded to a discovery that
was enabled by AI: AlphaFold. AlphaFold was made to predict the structure of virtually all the
200 million proteins that researchers have identified, and it enabled the development of new
protein structures; new iterations of AlphaFold predict the structure and interactions of all of
life’s molecules, which can unlock new materials, crops, drugs, and research. 36

Generative AI can speedup the discovery of new materials and molecules using a process
called generative design. For example, by training AI models on simulations from quantum
physics, we can make more accurate predictions about how materials behave; these models,
when paired with more traditional approaches, help scientists get a better understanding of

32 Walsh, D. (2025). A new look at the economics of AI. MIT Sloan School of Management.
33 Mayer, H., Yee, L., Chui, M., & Roberts, R. (2025, January 28). Superagency in the workplace: Empowering
people to unlock AI’s full potential. McKinsey & Company.
34 Acemoglu, D., (2024), The Simple Macroeconomics of AI. MIT
35 Escotet, M. Á. (2023). The optimistic future of Artificial Intelligence in higher education. Prospects.
36 Deepmind (2024), AlphaFold 3 predicts the structure and interactions of all of life’s molecules.

11