Solving the AI Leader's Dilemma: Saying Yes to GenAI "In Thing" or GenAI ROI First decision

Nupour Mukherjee Lalit S Kathpalia PhD

Assume you are the AI Board Director of a Global company and have asked to approve/disapprove all the use cases coming to you from departments and business lines asking for GenAI to be embedded in their lines of business, operations, and product management. Some have a genuine need and others are recommended by consultant or even they think they want it because of a promise of being the "in" thing". As a CXO in AI, you must determine how to make a framework checklist for aspiring generative AI use case business owners /tech owners coming to you, as an eligibility criteria to get selected for adopting generative AI in their unit. How would you determine genuine cases and their eligibility for genAI adoption with the highest ROI and positive perception? Nothing sort of a Nightmare right : there is aspirations , people , egos and a sheer passion to do above and beyond to be dealt with . The answer came to me , from detailed sessions at ISB's leadership for AI , that goes above and beyond a technical use case only focus of GenAI but overall how to think as an AI leader. It is the framework and my framework that seems to be working well is to follow three different frameworks "Munger , Kotter and Business Canvas Framework : Here's how

Framework Checklist for Generative AI Adoption:

1. Use Case Description and Alignment:

Clear Business Problem: Is the proposed use case addressing a specific business problem or opportunity?

Alignment with Strategic Objectives: Does the use case align with the company's strategic goals and priorities?

2. Munger's 0.1% Transformation Guidelines:

Magnitude of Impact: Will the implementation of Generative AI lead to significant transformational impact, even if in a small area of operation?

Value Creation: Can Generative AI create substantial value disproportionate to its implementation cost?

Competitive Advantage: Does Generative AI offer a competitive advantage in the industry?

3. Kotter's Change Management Principles:

Urgency: Is there a sense of urgency and necessity for adopting Generative AI in the identified use case?

Guiding Coalition: Is there strong leadership support and a coalition of stakeholders driving the adoption process?

Vision and Communication: Has a compelling vision for the use of Generative AI been communicated effectively to stakeholders?

Empowerment and Incentives: Are employees empowered and incentivized to embrace the changes brought about by Generative AI adoption?

4. Business Model Canvas Method:

Customer Segments: Who are the primary beneficiaries of the Generative AI application (e.g., internal teams, external partners, customers)?

Value Proposition: What specific value does Generative AI bring to the identified customer segments?

Channels: Through what channels will the Generative AI solution be deployed and accessed by users?

Revenue Streams: Are there potential revenue streams associated with the Generative AI solution (e.g., cost savings, increased efficiency)?

Key Activities and Resources: What key activities and resources are required to implement and sustain Generative AI adoption?

Cost Structure: What are the upfront and ongoing costs associated with implementing Generative AI, including training, infrastructure, and maintenance?

Partnerships: Are there potential partnerships or collaborations that could enhance the success of Generative AI adoption?

5. Eligibility and ROI Analysis:

Data Availability: Is there sufficient high-quality data available for training Generative AI models?

Complexity of Output: Does the use case involve generating complex outputs where Generative AI can provide value?

Resource Availability: Are the necessary resources (e.g., computational power, expertise) available for successful implementation?

Cost-Benefit Analysis: What are the estimated costs and benefits associated with adopting Generative AI in the identified use case?

ROI Projection: What is the expected return on investment from implementing Generative AI, considering both tangible and intangible benefits?

Risk Assessment: What are the potential risks and challenges associated with Generative AI adoption, and how can they be mitigated?

6. Perception and Stakeholder Management:

Internal Perception: How is Generative AI perceived by internal stakeholders, including employees, management, and board members?

External Perception: How might Generative AI adoption impact the company's reputation and perception among external stakeholders, such as customers, investors, and regulatory bodies?

Stakeholder Engagement: What strategies will be employed to engage and involve stakeholders throughout the adoption process, addressing concerns and garnering support?

7. Continuous Improvement and Evaluation:

Feedback Mechanisms: How will feedback from users and stakeholders be collected and incorporated to continually improve Generative AI applications?

Performance Metrics: What key performance indicators (KPIs) will be used to measure the success and impact of Generative AI adoption over time?

Iterative Approach: Is there a plan for iterative refinement and adaptation of Generative AI solutions based on real-world usage and feedback?

As an AI Board this is an online checklist that I recently recommended a bank and pharma manufacturing unit , to do the initial screening for internal GENAI applications .

Hope this helps

checklist for users to realize whether their use case should opt for genAI?

Generic Framework Checklist: Use Case Description:

Briefly describe the proposed use case.

• Data Availability: Is there sufficient data available for training a Generative AI model?

Yes

No

• Complexity of Output: Does the use case involve generating complex outputs such as images, text, or designs?

Yes

No

• Creative Content Generation: Does the use case involve generating creative content?

Yes

No

• Resource Availability: Can we access the required computational resources for training and deploying Generative AI models?

Yes

No

• Cost-Benefit Analysis: Estimated Cost of Implementing Generative AI:
• Estimated Benefits of Implementing Generative AI:
• ROI Analysis: Estimated Return on Investment (ROI) from implementing Generative AI:
• Customer/Business Benefits: List potential benefits for customers or business resulting from implementing Generative AI in this use case:

Reasons for Ineligibility:

• If the use case is deemed ineligible for Generative AI, provide reasons why:

Alternatives Considered:

Have alternatives like automation or process orchestration been considered?

Yes

No

• Reasons for considering Generative AI over other alternatives:

Global Banking Companies Framework Checklist:

Note: Incorporates considerations specific to banking operations

Use Case Description:

Data Sensitivity:

Is the data sensitive and subject to regulatory compliance?

Yes

No

Fraud Detection and Prevention:

Does the use case involve detecting and preventing fraud in banking transactions?

Yes

No

Customer Experience Enhancement:

Will implementing Generative AI improve the customer experience in banking services?

Yes

No

Risk Mitigation:

Can Generative AI help in mitigating operational and financial risks in banking operations?

Yes

No

Compliance Requirements:

Will Generative AI assist in meeting regulatory compliance requirements in banking operations?

Yes

No

Global Pharma Manufacturing Companies Framework Checklist:

Note: Incorporates considerations specific to pharmaceutical manufacturing

Use Case Description:

Drug Discovery and Development:

Does the use case involve accelerating drug discovery and development processes?

Yes

No

Process Optimization:

Can Generative AI optimize manufacturing processes and reduce time-to-market for new drugs?

Yes

No

Quality Control and Assurance:

Will implementing Generative AI improve quality control and assurance in pharmaceutical manufacturing?

Yes

No

Regulatory Compliance:

Can Generative AI assist in ensuring compliance with regulatory standards and guidelines in pharmaceutical manufacturing?

Yes

No

Supply Chain Management:

Will Generative AI enhance supply chain management and distribution of pharmaceutical products?

Yes

No

+++++++++++++++++++++++++++++++++++++++++++++

My Lessons learnt

These frameworks aim to provide structured guidance for evaluating the suitability of Generative AI in different enterprise contexts, ensuring thorough analysis of cost-benefit, ROI, customer/business benefits, and risk mitigation.

It would be a bit to realize not to follow the "School of Linkedin Hype or the University of Whatsapp" or look at historical data and reapply the techniques and call it AI or be genuinely honest with myself and my clients and avoid calling or repackaging Automation , RPA as AI implementations. A Generative AI demands genuine rethinking of org design, and rethinking of architecture - this is lego framework, federated mode, integration , Security redesign , network compute redesign , infra spend redesign , org role redesgin and most importantly an honest rejig of Cost to implement , cost to serve , cost to upgrade , Cost to serve cost reduction , employee rebatching , customer and employee perception and any backup plans for loss of IP's or dealing with ghosts in the system. I am sure with this done and diligently done , a fat bonus and a clear conscience and admiration of many , and sustainable growth of the company , positive sjareholder and market perception will exponentially be a positive.

If you are looking for miracles and magic , put in the hard work and honest diligence . GENAI is like Ali Baba's cave , you have to say the password correctly "open Sesame" , get your carriage mules ready to get the treasures away securely in time , you have a trusted partner like Morjina (AI expert like me :-)) to support and counsel you on the way , before the 40 thieves arrive , and ensure that any data leakage or discussion (aka Ali baba's brother Quasim , does not get you or him killed (data leakage risk) and while taking the treausre we careful of the wily Jjinn in the cave ( AI hallucination) to prevent you from achieving your task.

IF are now ready to make decisions , use the above framework or remember the fable of Alibaba's cave.

If you want to chat about your AI use cases and deterministic models or Maxlmum likelihood estimate, DM me at Nupour Mukherjee