“We were so scared of our IPO”: Leaders Unplugged with former On-CEO Marc Maurer
Marc Maurer shares how ON grew from startup to IPO by defying hype, focusing on purpose, and leading with humility in this candid Leaders Unplugged episode....
Strategy
Why the future of AI marks the end of scenario planning
Forget about the “three horizons”. How artificial intelligence is going to affect everything is such a massive unknown that you need a whole new approach to strategizing for the future.
AI-based technologies will change our lives, our businesses, and our societies – but the fact is, nobody knows to what extent, or even what the specifics will look like. While some aspects of the current hype around ChatGPT might disappear, the recent developments are substantial, multidimensional, and exponential, and create positive feedback loops that accelerate themselves.
Our world changed forever in the moment that Ashish Vaswani at Google and his co-authors published their paper “Attention is all you need” in 2017, suggesting a new machine learning architecture. That was the paper that enabled the development of transformers, on which all GPTs (generative pre-trained transformers) are based.
Generative AI has already changed jobs and processes, led to strikes, enabled new customer interactions, and forced governments to work on new regulations. The results of a representative research study I conducted in October 2023 among Swiss non-executive board members show that artificial intelligence ranks as the most crucial topic that boards must pay attention to. There is a clear consensus that both the benefits and risks of generative AI are vast.
Because the development of AI is based on technologies that indicate exponential progress, forecasts are impossible. This is especially true as developing technologies are interacting with each other. My thought experiment went like this: I researched a list of technologies that interact with the development of AI, which resulted in 24 entries (see box). In other words, we need to follow the trends of 24 technologies. If we now calculate (using ChatGPT-4) how many interactions 24 technologies have, we learn that there are 16,777,239 combinations of cross-fertilization that we should follow. Clearly, this is not possible, which poses a problem for strategy development.
Why scenario planning is not working
Every strategy has three components. We must analyze and diagnose the external environment and our internal resources and capabilities. Once we can answer the question “What the heck is going on?” with a coherent narrative, we must identify choices involving significant strategic trade-offs. Finally, once we decide, we need to align the resources and actions within the organization.
Scenario planning is the established methodology to deal with uncertainty in the external environment. Scenario planning was developed within Shell about 50 years ago. The main unknown was the oil price development, which was affected by a set of other developments (see Figure 1).
Once the main scenarios are understood, strategic initiatives can be planned. These initiatives are proactive steps the organization can take to navigate or capitalize on the potential futures envisioned in the scenarios. Organizations often use the Three Horizons Framework by McKinsey to manage and prioritize these initiatives. This framework categorizes initiatives into three “horizons” based on their time frame and nature:
- Horizon 1 focuses on core business activities that ensure current profitability and stability.
- Horizon 2 involves emerging opportunities that may require substantial investment but promise future growth.
- Horizon 3 envisions long-term transformative ideas or innovations that hold the potential to reshape the industry or organization’s future. In other words, Horizon 3 initiatives consider investments in resources and capabilities that will likely be the foundation for a competitive advantage in the future, even if the future business model is unknown.
By mapping initiatives across these horizons, organizations have been able to balance their portfolios of strategies to maintain current operations while investing in future growth and innovation.
But here is the catch: Scenario planning is particularly effective when dealing with a single critical uncertainty, like oil prices, where varying prices can create different future states. This allows organizations to develop targeted strategies that respond to specific changes in that variable. The future of AI presents a vastly more complex challenge for scenario planning due to the many interrelated uncertainties involved.
This complexity makes it exceedingly difficult to construct scenarios that are coherent (internally consistent and believable), plausible (credible given what we know about the world), and likely (having a reasonable chance of occurring). The interactions between the numerous variables can lead to emergent behaviors and outcomes that are not readily foreseeable by extrapolating current trends.
AI technology inevitably leads to the biggest education initiative in the history of organizations. Because we just don’t know yet.
What to do instead: The six E approach
In my recent conversations with leaders in various industries (publishing, insurance, distribution, consulting, manufacturing, food processing), I learned that there are better ways to strategize for an AI future. Given that there are too many trends to predict a future with a useful likelihood, organizations must create an environment for learning.
This comes at the risk of “picking up the AI hammer to find AI nails,” as my friend Jim Pulcrano describes it. Nevertheless, diving in is inevitable, but should be done with a systematic approach, which I summarize with six guiding principles.
- Educate and Embrace: Given the fast development of AI, particularly generative AI, everybody must learn how to use these tools to experience the benefits and to understand the risks. I am convinced that AI technology leads to the most significant education initiative in the history of organizations. Leaders and employees need to embrace AI technology and discuss the learnings in formal and informal settings. Studies show that familiarity with AI tools reduces anxiety. Working with AI tools can also bridge the generation gap within companies or, at least, prevent this gap from widening.
- Execute and Evolve: Companies must identify AI-based use cases that help them increase revenues and reduce costs while posing few risks. Research shows that AI-based use cases in marketing and sales will likely increase revenues with little upfront investment. AI-based use cases in process optimization promise cost savings, and AI-based use cases, such as chatbots applied in customer interfaces, create benefits on both dimensions. Such use cases often start as isolated initiatives and evolve into company-wide initiatives with proper risk assessment and risk management measures.
- Explore and Evaluate: Because scenario planning is no longer feasible for strategizing, as explained above, organizations must explore and evaluate. The term “exploration” includes searching, investigating, and discovering new places, knowledge, or experiences. Exploration is driven by curiosity and the desire to understand and interact with the world around us. Applied to AI, the leaders I talked to and worked with set up formal exploration teams that searched for relevant signals. After the exploring phase, the learnings summarized are evaluated in workshops. This process must be inclusive. Involving and sharing the learnings relates to the mandate of "educate and embrace" that we discussed above. AI-related explorations are not projects with a start and end date but are part of an evolving learning process. Figure 2 illustrates such a learning process.
As we face the vast possibilities of AI, it’s clear that traditional ways of planning for the future, like scenario planning, don’t work anymore. The future of artificial intelligence is made up of many complex parts, with numerous technologies growing fast and affecting each other in ways we can’t predict.
Trying to guess what will happen using old methods is no longer helpful. Companies must switch to strategies that can change and grow as quickly as AI. This means following the ‘six E’ approach: educate, embrace, execute, evolve, explore, and evaluate.
Relevant technologies for AI development (list is not comprehensive)
- Quantum computing
- Edge computing
- 5G and advanced network technologies
- Internet of Things (IoT)
- Blockchain technology
- Augmented reality (AR) and virtual reality (VR)
- Cloud computing advancements
- Big Data analytics
- Advanced machine learning algorithms
- Neuromorphic computing
- Cybersecurity innovations
- Natural language processing (NLP)
- Robotics and automation
- Advanced semiconductor technologies
- Biometric technology
- Energy-efficient computing
- Digital twins
- Human-AI collaboration tools
- AI ethics and governance
- Deep learning and neural networks
- Advanced predictive analytics
- Wearable technology and personal AI
- Smart cities technologies
- Autonomous vehicles
Authors
Related
How to harness the power of nonmarket strategy
Forward-thinking leaders proactively shape their external environment, turn uncertainty into certainty, and create substantial value in the process....
Audio available
Luxury E-Commerce is dead: Here is what might come next
The traditional e-commerce model is on its last legs. In a disrupted luxury landscape, brand leaders are shifting focus to unified commerce, hyper-personalization, and deeper digital storytelling to completely reinvent the customer...
Shaping the game: Understanding Trump’s unconventional negotiation strategy
Donald Trump’s tariff tactics offer a bold case study in game theory in negotiations. By setting the rules early, limiting options, and projecting power through credible threats, his approach shows how negotiations...