Decision-making in AI Transformation

Summary:

• Age of AI comes with multitude of complexities, uncertainties and risks. At the same time, stakes are higher than ever: There’s a vast difference between succesful AI deployment and failure to do so.
• Because of complexities and uncertainties, decision-making is more difficult than ever before. Yet, given the speed of change in the business environment, acting as a deer in the headlights is the worst thing to do.
• Sources of complexity and uncertainty are many: Rapidly evolving AI technology, competitive landscape that has become AI-defined, digital ecosystem dynamics, and AI use case integration challenges to name some of the biggest ones.
• Decision-making roadblocks include analysis paralysis and delay tactics as ways to cope with high degree of uncertainty. However, in the Age of AI, the greatest risk is not making a mistake, but failing to make a decision at all.
• The primary remedy is to add clarity. To make decision-making easier for decision-makers in all roles. Adding clarity builds on many things.
• Path to clarity comprises of many methods. Those include AI Transformation Core Methods discussed in ealier articles of the series, Digital Strategy, and Decision-making Framework.
• Core Methods discussed in the article series so far include using AI use cases as a bridge between technology and business, and to assess use case integration requirements.
• Digital Strategy consists of things like business environment analysis, digital capabilities assessment, identification of strategic alternatives and options, strategic choice, and execution planning.
• Decision-making Framework covers things like governance structure, steering group facilitation, decision-making culture, risk calculation, operating model, and AI literacy.

Sources of complexity and uncertainty

Technology Evolution – Rapid AI technology evolution makes it challenging to keep up. To be able to tell signal from the noise. To identify breakthroughs that are poised to disrupt whole industries. For example, multimodal generative AI models that will create totally new way of human-machine interaction. When applied to all products and digital services, this alone will change everything across all industries. Something that all companies will need to respond one way or another.

Competitive Landscape – To be competitive is to create genuine customer value, combined with customer experience that differentiates, and to do this with minimum effort. AI use cases address all three domains at once. Players that succesfully integrate AI use cases with versatility and scale will become the industry leaders. Conversely, players missing Basic Competitive Requirements will get marginalized fast. At extreme, when sales stop over night, it’s too late to do anything due to financial limitations. Age of AI is destined to create many new Kodaks.

Age of AI will create many new Kodaks thru missed Basic Competitive Requirements. To prevent that is to build foundational digital capabilities early enough - before the threat appears on radar screen.

Pace of Change – Changes in technology and competition create uncertainty by themselves. Speed of ongoing changes makes the situation even more challenging, especially when combined with organizational inertias and need for digital capability build-up. Competitive pressure and related strategic risks make timing a key factor.

Nature of AI – Digitalization of business processes and systems comes with significant complexity. However, digital systems implemented with traditional software code remain deterministic. After comprehensive testing, we know exactly how the system will behave. AI changes this by being inherently undeterministic. Neural network behavior cannot be explained exhaustively. From decision-making perspective this creates new layer of uncertaintly. Fear of making mistakes increases.

Digitalized business processes are still deterministic. AI is inherently undeterministic. This limits AI application domains - and adds to uncertainty in decision-making.

Digital Capabilities: To do AI transformation at scale calls for signficant amount of highly versatile digital capabilities. Choosing when to focus where becomes complex task. Natively digital challengers have significant advantage on their side in terms of smaller organizational and technological inertia – leading to reduced complexity.

Natively digital challengers are dangerous due to smaller organizational and technological inertia. Incumbents need to watch out.

Digital Ecosystem: Selective outsourcing of digital capabilities is an essential element in the management toolbox. However, rapid technology evolution creates equally fast changes in digital solutions. Choosing between Make and Buy becomes more complex with increased uncertainty in terms of betting future winners within the ecosystem.

AI Use Case Integration – Integrating AI use cases into business operations is inherently systemic and complex undertaking – covering technical, organizational and process aspects alike. Some of the required capabilities can be outsourced thus reducing the overall complexity but still many need to be created in-house.

Maintaining balance between outsourcing and in-house capabilities is crucial to sustain innovation and differentiation - both essential in the Age of AI.

Decision-making roadblocks

Under high degree of uncertainty, the two most common decision-making roadblocks are analysis paralysis and risk-aversion. They are closely related but different.

Analysis Paralysis happens when decision-makers request excessive analyses in a futile attempt to eliminate all uncertaintly. In some cases, this may even turn into simple delay tactics to buy time in the hope that time alone will take care of the uncertainty. In some cases, that comes with high cost. Taken to an extreme, at bankruptcy, there are no longer uncertainties. Age of AI has the potential to create extremes with elevated probability.

Attempt to eliminate all uncertainty is futile and often harmful. Ability to decide under uncertainty is a mark of strong leadership. But always shooting from the hip is not optimum either.

Risk-aversion is about fear of mistakes. Uncertainty connects to the moment of decision-making while risk is about implementation after the decision has been taken. From business and capital’s perspective, risk-aversion is a difficult problem to address. It all boils down to incentives. If de-facto incentives – both monetary and cultural – favor caution, managers avoid getting caught of “bad decisions” at any cost. For example, head of business function will be accountable for investment decision not delivering on the business case. However, he will not be held accountable for company bankruptcy. Under high level of uncertainty, for him, it makes sense to postpone the investment decision for as long as possible – potentially risking the future of the whole company.

Incentives have profound effect on decision-making. Getting them right becomes crucial corporate governance issue. Age of AI makes it more so.

Avoiding serious suboptimization in decision-making becomes crucial governance issue. Complexity combined with pace of change makes this a challenging governance problem. The first thing is to recognize that minimizing tactical risks may lead to signficant strategic risks, even existential ones. It is for corporate governance to make sure that such a scenario is avoided.

Dealing with complexity, uncertainty and risks

With all those complexities and uncertainties in play, decision-making becomes exceptionally difficult. Telling truly important information from marginally important or mere noise is not easy. Placing the bets in an optimum manner becomes non-trivial task. And most of all, avoiding pitfall of undecisiveness becomes crucial.

Key question then becomes: How to deal with all this complexity and uncertainty?

The primary remedy is to add clarity. To make decision-making easier for managers on all levels and for decision-makers in all roles. Fourtunately, there are many complementary ways to achieve that.

Q: How to deal with all this complexity and uncertainty? A: The primary remedy is to add clarity.

Core Methods

Article series on AI Transformation has introduced a handful core methods to alleviate complexity and uncertainty through better situational awareness and by applying new tools to address specific problems:

• AI-defined Competitive Landscape – With AI embedded in all aspects of value creation, industry players create higher customer value and better customer experience with enhanced efficiency – leading to higher productivity. Productivity Frontier is a useful way to visualize how AI is shaping the competitive landscape within an industry with an important lesson: Falling too far behind the Productivity Frontier creates an existential risk.
• Digital Capabilities for AI Transformation – AI Transformation builds on digital capabilities of different nature: Strategic Management, Digitalization, Cloud Migration, Data Management, AI Engineering, Software Engineering, Operating Model, and Data Culture. Shortcomings in any of these become constraints hampering AI use case integration.
• Use Cases at the Heart of AI Transformation – AI use cases with crisp value proposition connected to business model elements bring clarity. Business Model Canvas is a well-known tool to do that visually. Strong AI use case portfolio emerges from coherent and mutually reinforcing use cases.
• AI Technology Evolution – Use cases can be used to assess technology evolution’s impact on business. Generic use cases act as proxies of technology evolution with increasingly powerful enablers. Focus on a certain application domain reveals gains of underlying AI technology evolution. In this way, it is possible to pinpoint what each AI technology generation can and cannot do.
• Use Case Integration Requirements and Constraints – Assessment of AI use case integration builds on two complementary perspectives: Integration Requirements and Integration Constraints. IR Factor signals AI use case integration complexity, effort and cost in a compact easy-to-use format. Correspondingly, IC Factor indicates company-specific constraints on integrating the AI use case in question. Together, IR and IC Factors cater to systematic integration assessment and informed decision-making.
• Outsourcing Digital Capabilities for Use Case Integration – To alleviate AI use case integration challenges, both strategic and tactical outsourcing are essential. Strategic outsourcing addresses platform-like digital capabilities to minimize engineering related cognitive load on business domains. Tactical outsourcing aims at reducing use case specific integration complexity and the effort required from business domains. When business domains consider what to outsource, the questions to ask include What is our role in the company? What is expected from us in terms of value creation?

Put together, core methods introduced in the article series add significant amounts of clarity and thus become powerful decision-making aids. AI use cases in the middle serve that very purpose. They provide an intuitive way to connect business needs and opportunities to underlying enablers from AI technology to digital capabilities – while supporting the option of outsourcing some capabilities to digital ecosystem.

Together, core methods add large amount of clarity for decision-making.

Digital Strategy

Strategic homework provides the necessary foundation for understanding and navigating digital landscape. It is essential part of decision-making preparation. Strategic homework done well reduces complexities and uncertainties decision-makers will face later. Good digital strategy is something to lean on when encountering operational decisions. Combined with core methods discussed above, digital strategy empowers leaders at all levels to make informed decisions that drive the company forward in the Age of AI.

When done right, digital strategy empowers leaders in the Age of AI.

In this context, digital strategy refers to current state assessment and forward-looking plan on how to flourish in the Age of AI. It does not replace business strategy but it does intertwine intimately with business – if it didn’t, it would be useless. Here’re the core elements of such strategy:

Business Environment is mandatory part of every strategy. The environment is about markets, competitive landscape and regulatory framework. In case of digital strategy, these fundaments do not change but understanding digital elements of the environment becomes essential. For example, how customers’ value creation processes are affected by AI. Or how competition is utilizing AI to improve their competitive position. Technology evolution makes yet another crucial business environment element to assess – impacting markets, competition and digital ecosystem alike.

Digital Capabilities is the internal side of the story, complementing external description created with business environment analysis. Digital capabilities are to be understood broadly, not limiting only on technical issues. By definition, no company can influence its environment in any significant way. Therefore, the way it acquires, organizes and manages its digital capabilities will determine success or failure. In the Age of AI, the difference between success and failure can be decisive.

Digital Strategy needs to outline how digital capabilities are to be acquired and organized.

Analysis alone is not enough. Good strategy presents alternatives and options in terms of how to go after market opportunities, how to position in relation to competition, and how to utilize technology. In other words, strategy is about offering and value proposition, and how value is to be created with the resources available.

Once alternatives and options have been presented, it is time for strategic choice: What we do and what we don’t do. In ideal situation, argumentation on choices made is part of strategy communication. This is done for improved buy-in and alignment. Those are crucial when it comes time to implement the strategy.

Strategic choice is about what we do and what we don't do (and why)

Strategy ends with Execution Plan that may take many forms but should at least provide strategic roadmap: An outline of key phases planned and milestones targeted in order to implement the strategy. Strategic roadmap is an essential tool to connect digital capability acquisition and build-up to business justification – ideally directly traceable to business environment fundaments and strategic choices made. Good digital strategy comes with easy-to-follow storyline from environment to execution, with everything in between.

Easy-to-follow storyline from environment to execution plan marks good Digital Strategy

Interplay: Core Methods and Digital Strategy

AI Transformation core methods complement Digital Strategy in many ways. Together, they are a perfect match to alleviate pain of complexity and uncertainty. Here’s an overview of the interplay between the two:

• Competitive Landscape – The concept of Productivity Frontier, and the way AI use cases move it, is good way to close the gap between high-flying strategic slogans and concrete ways to improve processes, products and services with AI.
• Digital Capabilities – Digital capabilities can be categorized in many ways but the eight-piece model introduced in the article is a strong candidate to be used also in digital strategy. Its main merit is comprehensive approach combing technology, processes and organization – all essential enablers for AI Transformation.
• AI Use Cases – Taking the discussion to use cases makes it more concrete and tangible than typical strategy is. However, synergies are significant as business model defined as part of strategy is also the baseline for AI use case mapping. Treating use cases as portfolio rather than isolated point-solutions is a strategic choice by itself.
• AI Technology Evolution – By showing how technological advances make even existing use cases much stronger, the article demonstrates the importance of staying in the race for higher customer value, better customer experience and enhanced operational efficiency. It is for digital strategy to recognize this and to make statement on how to do that in practise.
• Integration Requirements and Constraints – Detailed assessment of AI use case integration requirements and constraints comes after digital strategy. Ideally, digital strategy provides guidelines on how to prioritize use cases and digital capabilities build-up. Details can be added later.
• Outsourcing – Strategic and tactical outsourcing are about digital capability acquisition from the surrounding ecosystem. Outsourcing is an essential part of strategy implementation as in-house capabilities can never serve all emerging needs. As minimum, digital strategy needs to provide guidelines for both strategic and tactical outsourcing. For example, Core vs. Non-core analysis should be part of the strategy project itself. In addition, ideal digital strategy would already outline Capability Acquisition Plan.

Decision-making Framework

As the final step of dealing with complexity, uncertainty and risks, let’s look into decision-making framework. Core methods and strategic homework discussed above are basically all about preparatory steps to enable informed high-quality decisions. Conversely, decision-making framework is about decision making itself and what else can be done in order to ease up the act of decision. Decision-making framework is a collection of many different things:

Governance Structure – With governance structure in place, operational decision-making takes place in steering groups with clear mandates rather than in line organization. In effect, this makes decisions collective rather than individual – by head of business function, for instance. This eliminates much of the personal risk related to poor decisions.

Steering Group Facilitation – Decision-making quality lies largely on good facilitation. In effect this means that decisions are always well prepared. Steering group facilitator makes sure that poorly prepared presentations seeking decisions do not get on stage at all. However, here lies the pitfall: With complexities and uncertainties in play, skilled facilitator is able to tell when to let things through despite gaps in the presentation. In doing that, he leans on the decision-making culture and senior leadership support.

Good steering group facilitation can make wonders in terms of decision-making quality and effectiveness.

Test Questions – Steering groups and their facilitators should use simple test questions: Would further analysis and planning decrease level of uncertainty any more? Should we instead proceed and verify our assumptions in the markets and in real use?

Very often it is better to "get out of the building" and test assumptions with markets and actual customers.

Decision-making Culture – Organizational culture with learning emphasis sees uncertainty as normal part of the business landscape. It views initiatives of high uncertainty as an opportunity for learning and growth. Incentive structures reward not only outcomes but also smart risk-taking. Risk acceptance is in-built with cultural understanding of the necessity of risk-taking in order to innovate and renew. Within the organization, there’s an understanding of how implementation risks differ from strategic and existential risks – leading to better acceptance of implementation risks.

Risk Calculation – Practise of taking calculated risks is well established. Decisions are made with the best available information and where potential negative outcomes are recognized and mitigated. Steering group facilitation plays central role in making this happen.

Open Communication – Rationale behind decisions is communicated effectively with feedback actively sought and valued. Again, steering group facilitation has key role to play.

Leadership Support – Senior leaders play central role in establishing decision-making culture. This is part of vision and strategy communication where business environment complexity and inherent uncertainty is clearly pointed out – with the recognition of the implied decision-making practises needed to succeed in such environment.

Operating Model – Modern operating models have the assumption of uncertainty and unpredictability in-built. DevOps with Continuous Delivery is the prime example. With such operating model in place, decision-making becomes much easier as nobody expects absolute certainty at the point of decision-making. Continuous Learning is the default. However, pitfall is equally evident: No real learning takes place and course correction does not happen. Good governance addresses this pitfall.

Good Operating Model is based on the assumption of uncertainty and unpredictability. DevOps is an incarnation of this profound wisdom.

AI Literacy – AI literacy is about organizational awareness of what it means to operate in the Age of AI. It comes with awareness of how AI changes the business environment and what are the implied requirements on the organization and its way of working. AI literate organization recognizes how AI impacts the overall value creation. Without this awareness, there is no motivation for cultural or behavioral change.

Without AI Literacy, motivation for change is lacking.

Conclusions

Age of AI challenges decision-making more than anything before. With complexities and uncertainties present, the apparent pitfalls include undecisiveness, analysis paralysis and fear of mistakes. The biggest dilemma is that procrastination does not reduce aggregate risk. On the contrary, in the AI-defined competitive landscape, too much delay is prone to make risks existential, at worst.

The remedy is to apply multiple methods to systematically ease up decision-making. Methods link to both preparatory work and the act of decision itself. The ideal situation is evident: Informed and timely decisions, dealing with complexities, accepting uncertainties, and systematically mitigating risks with iterative approach and continuous learning.

Introducing new article series on AI Transformation:

Introduction: Digital Capabilities for AI at Scale