Intro:
To demystify the hype around generative AI and identify suitable use cases, I recently had the privilege of conducting three workshops with various user groups. Our goal was to evaluate common scenarios where data could be leveraged for decision-making and to assess the machine learning methodologies applicable to these use cases. The evaluation process was structured to first explain each use case, followed by a discussion on different themes. We then used Wooclap to vote and reach a consensus on the best approach for each use case.
Use case for discussion:
| Use Case | Description |
|---|---|
| Forecasting / Prediction | Using historical data to predict future outcomes, trends, or behaviors. |
| Planning | Assisting in the creation of strategic plans by analyzing data and providing actionable insights. |
| Segmentation / Classification | Grouping data into categories for better analysis and decision-making. |
| Recommendation Systems | Suggesting products, services, or content to users based on their preferences and behaviors. |
| Content Generation | Creating text, images, or other media using AI algorithms. |
| Conversational AI | Enabling machines to interact with humans through natural language processing. |
| Knowledge Management | Organizing and retrieving information to improve decision-making and efficiency. |
| Decision Intelligence | Enhancing decision-making processes by integrating data, analytics, and AI techniques. |
Methodology considered :
The working group in each workshop debated on how each of the below methodology would be effective to solve a specific use case
| Concept | Description |
|---|---|
| Gen AI | Generative AI involves creating new content, such as text, images, or music, using AI algorithms. |
| Classical ML | Classical Machine Learning focuses on using algorithms to learn from data and make predictions or decisions. |
| Optimisation | Optimization involves finding the best solution from a set of possible solutions, often under constraints. |
| Simulation | Simulation is the process of creating a model to study the behavior and performance of a system. |
Criteria for evaluation:
| Theme | Dimension | Explanation |
|---|---|---|
| Value | Revenue | The potential of the use case to deliver additional funding to the organization through sale of new products and services, cost savings allocations, or grant funding that will support long-term growth. |
| Value | Efficiency | The potential of the use case to meet or exceed performance goals with less costs and improved service operations, resulting in reduced investment in current margin. |
| Value | Risk | The potential of the use case to cost or exceed current margins resulting in reduced investments in service operations. |
| Value | Nonfinancial Risk | The potential of the use case to assist the organization in meeting its nonfinancial mission-related goals: These goals can include environmental (E), social (S), governance (G), diversity equity and inclusion (DEI) sustainability; innovation; quality improvement. |
| Feasibility | Technical | The organization's ability to meet the technical requirements of a potential use case. Considerations include the core capabilities of the technology itself, the availability of vendor support, the current state of the organization's technology infrastructure, and the technical talent required by the use case. |
| Feasibility | Internal | The organization's ability and openness to use and incorporate potential use cases. This includes the willingness of internal stakeholders to understand, trust, and effectively execute these use cases. |
| Feasibility | External | The extent to which the environment outside of the organization is conducive to successful execution of potential use cases. This includes consideration of the legal and regulatory environment, public opinion regarding use cases, and digital access, literacy, and customer engagement by users. |
Outcome:
Below is the diverse perspectives and insights from the workshops, which highlighted that we need not use Gen AI for everything
Not every problem requires an over-engineered solution. As the saying goes, “don’t bring a gun to a sword fight.”
Top comments (0)