Understanding Options, Constraints and Impossibilities

Scope

The main thing in dealing with AI is understanding it in a business context rather than in a technological one.

This is done by organizing all subject matters (core competencies, domain expertise, processes) concerned to provide as many as possible variations of the (to be) filled in cells of the AI Business Model for Interacting Services.

The aim is to induce representations on understanding AI in a business context, so as to generate the set of divergent views (set of problem statements) on:

• what AI business are we in¹
• where do we want to go with AI
• how do we get there

and subsequently investigate these divergent views.

This obviously touches upon core business and core competencies directly, and thus strategy.

Essentially, it boils down to formulating an AI strategy to act as compass for the direction to take, by carrying out a strategic analysis with the aim to embed AI in the overall business strategy.

Data is again used to validate divergent hypotheses and divergent views sufficiently validated are systematized (order, measure or metric).

Scanning closely though all the views ensures once more the whole landscape of possibilities to be covered.

 

AI Strategy Development

 

 

No surprises here, start with mobilizing domain expertise and gathering information.

Begin by pulling together diverse teams of stakeholders with whom to brainstorm the key issues and strategic decisions regarding AI.

Address chief concerns with respect to the future, such as

shifting needs (abstract requirements) and preferences (expectations habits)

technology affecting capabilities

impact of changing regulation and environment

Capabilities are the tangible and intangible back office assets or resources tapped to create value.

the challenge is to formulate strategies that align (match map) positioning in the environment (market) with back office capabilities available or planned

assess capabilities well aligning with the environment through determining

the current value chains, by clustering all the activities that create value for a product, service or experience, working backward (reverse engineering¹) from the end point of the value proposition delivered

whether the present situation is what is needed to execute the current strategy and achieve goals now, or whether organizational capabilities need to be enhanced or developed for going forward

certain assets or resources that are not so core and which could therefore be outsourced

Identify the core strategic AI capabilities (AI Business Model for Interacting Services) needed to produce value in each cluster

isolate the core set of AI capabilities central to distinctive success

determine degree of alignment

strategy is about making choices and selections

Set a time horizon for how these concerns might pan out and set it far enough in the future that uncertainty is there, but not so far to become prone to unrealistic or wishful thinking.

Factor in economic and other cycles and bring in key and predictable trends.

Look for supportable evidence and expert consensus around things that can be researched and ascertained.

And brainstorm the key uncertainties, documenting them extensively.

 

Structuring, Writing and Using Scenarios².

The aim is to identify driving forces and critical uncertainties to develop plausible AI scenarios further to the AI Business Model for Interacting Services.

Scenario development covers identifying the driving forces and critical uncertainties for the time to come and to develop scenarios that are most likely to happen. The best way to perform all of these steps is to organize workshops, brainstorm sessions and whatever else helps to find creative solutions.

Identify driving forces. Discuss what are going to be the big shifts (driving forces) due to AI in the creative domain (societal, economical and technological) in the future and see how this will affect the organization. Make a model of the position of the organization or line of business (concrete satisfiers), possibly relative to fellow organizations or competitors. This model classifies the organization over a limited number of dimensions that capture either critical elements driving requirements and preferences or important attributes characterizing continuity of operations.

Identify critical uncertainties. Once driving forces identified pick up those that have the most impact on the organization. Determine the factors (costs and revenues; know-how and know-what; information and ICT architectures) related to realizing AI within commensurate application domains (AI Business Model for Interacting Services). Match the organization or one of its concrete satisfiers lines to each factor. In the best case, this can be done quantitatively (costs or prices are dimensions relatively straightforward to map). Other dimensions might require surveys or expert opinions to quantify and map, as when perceptions on quality or ratings are needed. Analysts may elect to assess organizations and businesses based on intuition or knowledge of the environment (market, industry, society). The final step is to classify the organization based on these assessments.

Develop a range of plausible (arguable) scenarios, ordered in an array. Choose orientations and model the representations into arrays in such a way that cells be as close to mutually exclusive as possible. Once topics, time horizons and key trends and uncertainties are identified, analyzed and properly represented (formally described), select the most important and uncertain ones and narrow down this list to the most critical factors. Each cell now specifies a scenario.

Depending on what direction each of the uncertainties will take, possible (achievable) scenarios for the future can now drawn.

Write out each scenario and fill in the details.

 

 

Discuss the various implications and impacts of each scenario and start to reconsider the strategy: set mission and goals while taking into account every scenario.

All critical issues are now defined, uncertainties uncovered and multiple scenarios constructed. The key value-adding step is now to ask what the implications of each scenario might be, what might have to be changed, reconfigured or added.

 

Synthesis

Premises

there is sufficient concordance on possible ways AI might help interacting services to systematically retract, embed and section information as to better deliver individualized solutions based on understanding and context (AI Business Model for Interacting Services)

computational systems allow to smoothly FAIR data with minimal intervention and provide the necessary support to deal with increasing volumes, complexities and creation speed of creative digital assets

things are linked to core business and core competencies, and thus have become Chefsache

The next step involves choosing for an AI-development plan of action.

The options vary between full-scope (organization driven) AI-development and narrow-scope (domain driven) AI development.

Deciding requires some insight into the broad-context.

If this broad-context is already sufficiently understood and available to guide the selection of relevant external business-drivers, then a domain driven AI-development plan of action will just demand but inside-inwards and inside-outwards perspectives, usually.

For full-scope AI all four perspectives will be needed, typically inducing the following sequence:

inside-inwards → outside-outwards (optional) → outside-inwards → inside-outwards (usually together with a detailed inside-inwards)

Ideating on the possible solutions can be done by filling in the models with the above in mind.

Upon sufficient alignment with the problem statement, convergence emerges by evaluating how different solutions fit into action plans.

 

Remarks

Note that deploying AI profitably in outside-in schemes may involve having to model utterly complex domains, that are highly dynamical and often lack sufficient reliable (hard) data. Moreover, deploying AI in inside-out schemes neither is obvious. Domains can be complex and dynamical, and there also may not be sufficient reliable data.

In doubt, stick to inside-inwards. This might, given the right constraints, very well include applying AI to help back office systems help front office systems deal with the whole spectrum of inside out and outside in schemes.

Strategic Positioning

Engaging in choosing for an AI-development plan of action necessarily entails positioning AI either as a feature for an inside-out or an outside-in strategy (scheme). Essentially there are two (dual) sets of (exploratory) key questions to be addressed by organizations in the creative domain wishing to assess the likelihood of being able to successfully apply inside-out or outside-in AI.

First Set

Is there sufficiently valid and relevant specific and specialized knowledge (data, information) available on:

targeted segments’ needs and behaviors

how to best solve relevant problems in the environment

kind of value provided per targeted segment

Is there a strong fit between

targeted segments’ needs and behaviors and

internal value proposition, overall organizational (business) model and an outward oriented organizational culture

Upon a strong fit between targeted segments’ needs, value provided and an outward oriented organizational culture (indicative of an outside-in scheme), successfully applying AI largely will depend on the factual modelability¹ (identification, analysis and representation) of available (and digitally processable) knowledge (data, information) of the environment (market).

If that is the case, which is not trivial, then there might be scope for applying AI on gathering, systematizing and interpreting the quantitative and qualitative data on which an organization can base its direction and activities. Note that with in the main intuitive (gut feeling) acting, working without strong underlying data on market segments or value created and with minimal and aggregated internal record keeping only, there is little scope for looking to apply AI in outside-in alike schemes.

The outside-in approach to AI works when there is sufficiently useful context-aware understanding of what constitutes value to different segments.

Second Set

Are there:

internal domains with interconnected items of knowledge and representations of problem solving processes, supported by effective protocols and efficient processes

coherent sets of (unique) value propositions² due to specific excellence in specialization or segmentation

indications on effective use of company resources and core competencies as main drivers of value creation

Is there a strong fit between:

internal specialization or segmentation and

internal value proposition, overall organizational (business) model and an inward oriented organizational culture

Upon a strong fit between internal specialization and value proposition (indicative of an inside-out scheme), successfully applying AI will depend on the factual modelability³ of the pertinent internal domains. If that is the case, then there might be scope for applying AI on services to access and process information better.

 

 

Method

In general, there are two types of strategic advantage an organization may pursue with AI: lower (internal or external) transaction¹ costs or more uniqueness (distinctiveness).

The two dimensions, broad versus narrow and lower transaction costs versus more uniqueness, define four generic AI strategies.

AI Strategies	organizations typically aim at deploying AI as a means to
	lower transaction costs	more uniqueness
broad-scope	general, aggressive and overall cost-cutting to realize low-cost inputs and labor gaining economies of scale through improving efficiency and effectiveness minimizing overhead the idea is to invest in low-cost, state-of-the-art operations and continuous improvement initiatives	general, purposeful and overall investing to realize improvement in inputs and labor globally building awareness developing innovative capabilities to stay on the cutting edge the idea is to invest in human resources and other ancillary activities
narrow-scope	limited, aggressive and targeted cost-cutting to realize specific low-cost inputs and labor gaining economies of scale through confined improving specialization and segmentation minimizing overhead the idea is to invest in state-of-the-art innovations in specialization and segmentation	improving inputs and labor to support segmentation of small-batch manufacture and distinctive craftsmanship locally building awareness developing specific, innovative capabilities to stay on the cutting edge with specific knowledge and abilities the idea is to invest in human resources and other ancillary activities

A full-scope AI-development would typically use these perspectives in the following sequence.

inside-inwards	start with developing a broad understanding of what would be required within each domain in scope
outside-outwards (optional)	then gain wide understanding of the overall business-ecosystem, in its own terms, independent of the organization
outside-inwards	establish a broad to detailed understanding of how others would interact and transact with the organization, from their perspective
inside-outwards (usually together with a detailed inside-in)	finally specify a detailed AI architecture for each domain, each from its own perspective, drawing on the previous perspectives for guidance

In practice, there will always be iteration between these perspectives.

Inside-inwards and outside-outwards can be switched-over if preferred. In effect, the sequence above defines the precursor-order: an outside-inwards perspective will depend on clarity about the overall context, as described by outside-outwards; and an inside-outwards perspective will only be viable if all of the other views are already in place.

A narrow-scope domain AI development will often need to use only an inside-inwards and inside-outwards perspective. This will be the more valid if the broad-context is already understood and is available to guide the selection of relevant external business-drivers. If the broad-context work has not been done sufficiently, a narrow-scope development may cause fragmentation problems (business-AI divides).

For domain AI, inside-inwards and inside-outwards perspectives will usually suffice, for full-scope AI all four perspectives will be needed.

 

Footnotes

1Scenarios – critical uncertainties: factors (costs and revenues; know-how and know-what; information and ICT architectures) related to realizing AI within AI Business Model for Interacting Services commensurate application domains.