Many people use AI to ask, "Which is the best?" However, such questions often yield overly general answers. AI does not know whether you prioritize price, speed, ease of learning, security, or flexibility if all those considerations are not explained.
For decisions with multiple options, AI is more useful as a comparison partner rather than a final answer machine. It can help organize criteria, highlight the weaknesses of each option, and uncover questions you may not have thought of. The final decision remains in your hands.
Why the question "which is the best?" is often insufficient
Every option almost always has trade-offs, which are exchanges between one advantage and another disadvantage. A lightweight laptop may have a smaller battery capacity. A cheap tool may limit the number of users. A marketing strategy that is quick to implement may not necessarily be easy to maintain.
If such context is not provided, AI tends to fill the gaps with assumptions. The results may sound convincing, but they may not fit your situation.
The problem is not solely that AI is "wrong." It could be that the criteria you are using are not yet clear. The process of comparing options can help make those criteria explicit.
Use AI to build a decision framework
Before asking for recommendations, ask AI to help you outline how to evaluate them. This is useful when you are still unsure about what to consider.
Example prompt:
I am comparing three options for the following needs: [explain needs]. Help me outline 5–7 relevant evaluation criteria. For each criterion, explain why it is important and what questions I need to answer before comparing options.
With this prompt, AI does not jump straight to conclusions. You get a list of aspects to check, such as total cost, implementation time, scalability, dependency risks, skill requirements, and ease of recovery in case of issues.
Do not accept all criteria automatically. Choose only those that are truly important for your situation. Five clear criteria are usually more useful than twelve criteria that are never used.
Weight the most important factors
Not all criteria hold the same value. When choosing an application for a small team, ease of use may be more important than advanced features. When selecting a service for storing sensitive data, security and data export capabilities may take precedence over price.
You can ask AI to create a weighted matrix. Weights are values that indicate the importance level of each criterion. For example, cost could be weighted at 20 percent, ease of use at 30 percent, security at 30 percent, and flexibility at 20 percent.
Use the following prompt:
Create a comparison matrix for the following options: [option A], [option B], and [option C]. Use the criteria: [list of criteria]. I want to weight: [criterion 1] at [x]%, [criterion 2] at [y]%, and so on. Use a score of 1–5, explain the reason for each score, and then calculate the weighted value. If data is insufficient, mark it as "needs verification," do not fabricate.
The part "do not fabricate" is important. AI may fill in details that sound reasonable even if it does not have enough information. For prices, feature limits, service policies, or specifications that may change, double-check the official sources.
Ask AI to present opposing arguments
One risk of using AI is that we may too quickly accept answers that align with our initial assumptions. If you lean towards one option from the start, AI can help test that decision by finding reasons not to choose it.
Example prompt:
Assume I tend to choose [chosen option]. Your task is not to agree with me, but to test this decision. Provide 5 strong reasons why this option may not be suitable, what conditions make it risky, and signs that I should consider another option.
This prompt does not make AI a judge that is always right. Its function is to open perspectives that are often overlooked when someone feels confident.
You can also ask for comparisons from different viewpoints:
- The perspective of everyday users.
- The perspective of someone who has to manage costs.
- The perspective of the technical team.
- The perspective of security and privacy.
- The perspective six months after the decision is made.
Different perspectives often change how we interpret the same options.
Use "what if" scenarios
Decisions that seem right today can become burdensome when conditions change. Therefore, do not just compare the current state. Also test the scenarios.
You can ask AI to analyze questions such as:
- What happens if the number of users doubles?
- What if the budget is cut by 30 percent?
- What if a key team member is no longer available?
- How difficult is it to transfer data to another option?
- Which part is most likely to become a barrier after three months?
Example prompt:
Compare [option A] and [option B] in three scenarios: light usage, rapid growth, and limited budget. For each scenario, explain which option makes more sense, the assumptions used, and the biggest risks I need to check manually.
Scenario analysis helps avoid decisions that are only good under ideal conditions.
Don't forget to separate facts, assumptions, and opinions
When asking for analysis, ask AI to categorize the results into three parts:
- Facts: information derived from data or verifiable sources.
- Assumptions: things assumed to be true because the data is incomplete.
- Opinions or judgments: interpretations based on certain criteria.
This format makes the results easier to audit. For example, the statement "option A is easier to use" is not an absolute fact. It may be a judgment based on the number of steps, user experience, or opinions from a certain group.
Prompt that can be used:
Organize the analysis into three columns: facts, assumptions, and judgments. Mark sections that still need verification. Do not disguise opinions as facts.
What does this mean for us?
AI is most helpful when decisions are broken down into checkable parts. You do not have to hand over decisions to AI, and you do not need to create complicated prompts. What matters is explaining the goals, options, criteria, weights, and information limitations.
A simple flow could look like this:
- Explain the problem and the available options.
- Ask AI to outline evaluation criteria.
- Determine weights based on real needs.
- Request a comparison matrix and reasons for each score.
- Test your favorite option with opposing arguments.
- Compare several scenarios if conditions change.
- Verify important facts before making a decision.
Use AI to organize your thoughts, not to avoid the responsibility of choosing. The best answers are not always the most assertive recommendations, but a clearer understanding of what you gain, what you sacrifice, and what needs to be checked before proceeding.
– Rio Yotto @rioyotto