How to use AI for analysis

"How do I use AI for analysis?" 

Large Language Models are skilled at processing tasks — interpreting, manipulating, or altering existing data. This used to be limited to text-based data, but the frontier models (i.e. ChatGPT, Claude, Gemini) have expanded to other forms including images, math calculations and reasoning, and structured data like tables and spreadsheets.

This means that AI is a great tool for analysis. 

For example:

• Identify trends in data: Use AI to quickly identify and explain trends within large datasets, regardless of the data type.
• Summarize research papers: Quickly understand the gist of academic or technical papers relevant to your work.
• Compile feedback: Aggregate and categorize reviews or feedback to identify common themes or issues.
• Simplify complex regulations: Break down complex legal or regulatory documents into easy-to-understand summaries.
• Summarize key findings: Automatically summarize the key points from a collection of documents, articles, or data sets.
• Compare document contents: Automatically compare and contrast the content of multiple documents to identify differences or similarities.
• Break down complex concepts: Simplify technical or complex concepts into easy-to-understand summaries.

… In other words, when you need to interpret, summarize, classify, simplify, or identify patterns.

‍

✏️ Analysis-Related Starter Prompts

• Analyze the key trends in [topic/industry] over the past [time period]. What are the main driving factors behind these trends?
  ‍
• Compare and contrast [concept A] and [concept B] in terms of their [aspect 1], [aspect 2], and [aspect 3].
  ‍
• Analyze the strengths, weaknesses, opportunities, and threats (SWOT analysis) for [company/organization/industry].
  ‍
• Identify the common themes or patterns in [dataset/collection of works]. How do these themes relate to [broader context]?
  ‍
• Explain the relationship between [variable A] and [variable B] in [specific context]. Are there any mediating or moderating factors to consider? 
  ‍
• Review the attached dataset. Summarize the main themes, along with any notable outlier opinions.
  ‍
• You are a data analyst. Your overall task is to review the below data and to summarize [MONTH] performance, though it may also be helpful to compare and contrast [MONTH] data with the previous month's data, [Q4 data], and the [2024 monthly averages]. First, review the data, take a minute to understand it, then share a few takeaways. For example, trends, outliers, dropoff points in the conversion funnel, or overall insights based on the performance here. These insights should be data-driven and written in a bullet-point format.
  • Review the following data and share the 3-4 most important takeaways. We will define important as the most significant net change and/or the most surprising or unexpected change.
  • Based only on the data I have shared and the takeaways you've listed, what are the top 3 challenges this business faces? What are the top 3 opportunities?
  • Provide some analysis based on everything we've done. Give me approximately 10 bullet points of top takeaways. Include major points. Also highlight anything that might be less obvious or intuitive, but may be significant or worth looking into.

‍

💡More Tips 

When using AI for analysis, remember: 

• Better inputs = better outputs. The size of the dataset, along with its quality, will impact the accuracy of your results. For the best outcome, use a large, clean, structured dataset.
• Choose the right tool(s) for the job. For reliable analysis, choose a frontier LLM (i.e. GPT-4o, Claude 3.5, etc.) or a specialized solution for data analysis. Leverage purpose-built analytical capabilities within LLMs when available — like ChatGPT’s “Data Analyst” GPT. And take advantage of multi-modal features like the ability to upload images and spreadsheets. 
• Trust, but verify. If the complexity of the analysis exceeds your ability to evaluate it, remember the three ways to reduce: Make it smaller, smarter, or safer.
• Don’t forget about “imprecise analysis.” There are also plenty of ways that imprecise analysis can be insightful, helpful, and time-saving. For example: Identifying trends and themes in large text files.

‍

📌 Want to Use Generative AI to Save Time?

Just remember RADIC:

• Research: To find or collect data on a topic.
• Analyze: To interpret or find patterns in data.
• Decide: To make a choice or recommendation based on data.
• Innovate: To combine or reimagine ideas in a new way.
• Create: To make something new — whether text, audio, video, image, or code.

RADIC is a simple framework that can help you to:

• Divorce your work from contexts like department, role, team, or industry.
• Break complex goals and responsibilities into smaller tasks.
• Consider the role of data in task completion. (What information is required to perform a given task?)
• Think about the degrees of sophistication and complexity in our work. (What do we do that’s routine and logical? What is more abstract and driven by intuition, instinct, or experience?)

‍

‍