The aim of this phase is to define what the decision aims to achieve and how success will be measured by identifying relevant objectives and structuring them into clear, usable evaluation criteria.
Image:
Definitions
Objectives
Objectives reflect the values and preferences of stakeholders and may encompass multiple dimensions, including environmental, economic, social, and technical considerations.
Criteria
Criteria translate broad objectives into measurable indicators that can be used to evaluate how well the alternatives achieve those objectives. Characteristics of a good criterion include, for example, relevance, non-redundancy, measurability, understandability, independence, sensitivity to differences and feasibility of assessment.
The objectives/criteria can be structured hierarchically as a value tree, with main objectives at the top, sub-objectives beneath them, and on the lowest level criteria against which the alternatives are measured (e.g. indicators). The set of objectives/criteria should comprehensively cover all the issues that are relevant to the decision, but on the other hand also be concise enough to be operational.
There is no general answer for the level of detail, but it depends on the aims of the analysis. Value-focused thinking can be used to enhance identification of the objectives. With questions like “What do we want to achieve, and why?” one can get deeper into the fundamental (main) objectives behind the problem. With questions like “What are the means of achieving these objectives” we can identify sub-objectives and criteria.
Outcome of the phase: A set of objectives relevant to stakeholders, with criteria under each objective for measuring how well the alternatives achieve it
Follow up Questions for the AI
Ask the following questions to get more information from the AI:
- Can you give examples of hierarchical objective and criteria structures in environmental decision problems?
- Can you elaborate the value-focused thinking process?
- What are the characteristics of good criteria [or a good set of criteria] in decision analysis?
Image:
Image:
Benefits and Risks of Using AI
Image:
Benefits (+)
- Support for comprehensively identifying objectives that may remain unnoticed by human stakeholders.
- Support for structuring objectives into hierarchical form.
- Support for going through the related literature about the problem and identifying objectives from there
Benefit Scoring: + Low Benefit
Image:
Risks (!!)
- Over-reliance on AI
- Even if AI can identify all the relevant objectives, an overly AI-driven process may make stakeholders feel ignored, which may reduce their learning and commitment to the process.
-AI’s difficulty in handling normative and value-laden aspects, and false sense of completeness and objectivity
- Objectives suggested by AI may appear “objective” but lack ethical legitimacy or stakeholder acceptance.
- Lack of contextual understanding about the local situation leading to a biased/incomplete set of objectives.
- In its suggestions, AI cannot know the local context and the background of the case, where underrepresented, marginalised or local values may be ignored (e.g. indigenous knowledge, cultural values).
Risk Scoring: !! Medium
Questions to Better Understand the Benefits and Risks of AI
Questions to make use of the benefits
- See the follow-up questions above.
- I have this problem: [description of the problem]. What objectives are typically found in these kinds of problems and what are the fundamental objectives behind these?
- I have these objectives: [description of the objectives]. Can you suggest how to structure these objectives hierarchically?
- I have these objectives: [description of the objectives]. Can you suggest measures for assessing these objectives?
Questions to avoid the risks
- What are the possible risks of using AI to support identification and structuring of objectives/criteria in environmental decision-making?
- [After getting suggestions of objectives from stakeholders] Are there any additional objectives that might be relevant in this problem?
Questions NOT to ask
- I have this problem: [description of the problem]. What objectives/criteria should be included in the model and what should be excluded?
AI can identify objectives, but it cannot decide what should matter most