How to Identify Opportunities for AI in Your Company’s Data

In today’s business landscape, artificial intelligence (AI) is transforming how companies handle their data, making inferences that would take humans far longer to process. But before diving headfirst into AI, it’s important to consider the type of data you have and the specific problems AI can solve. In this post, we’ll explore real-world examples from our own clients and provide some pointers on how to assess whether AI could be useful in your company.

[We examined this topic and more in our recent webinar, Making Your Data Useful with AI. Acess the full replay here].

Example 1: The National Judicial College

At ThinkNimble, one of our clients, the National Judicial College (NJC), was looking to better understand how judicial decisions are influenced by various factors such as age, race, or location. They had data on how different juvenile justice cases were adjudicated and were manually drawing connections between fact patterns. For example, they’d compare cases like a 16-year-old robbing a store in Arizona to a 14-year-old robbing a store in Massachusetts. While they could intuit some patterns, the task of making accurate and broad inferences became cumbersome.

We developed an AI solution for NJC that automated the recognition of similarities between cases, accounting for different variables and highlighting significant variations in outcomes. This type of AI could help NJC identify trends or biases in the judicial system that might otherwise be missed through manual comparison.

Key Takeaway:

If your company is manually sifting through large datasets and making inferences by pattern recognition, AI can help by automating these tasks, allowing you to draw more accurate conclusions in less time.

Example 2: AI vs. Simpler Solutions – Meals on Wheels

It’s also important to recognize when AI isn’t necessary. Meals on Wheels came to us wanting to use AI for medically tailored meals based on dietary needs. Initially, they envisioned an AI-driven system where users would input complex dietary needs, and the AI would generate results.

However, when we examined their data, we found it was highly structured, making traditional filters (i.e., checkboxes for diabetes-friendly meals) more effective than an AI model. Ultimately, a simpler solution worked just as well, saving time and resources.

Key Takeaway:

Start with the problem you’re trying to solve. If your employees can already find what they need quickly and efficiently, AI might not be necessary. Focus on whether AI will save time, enhance accuracy, or reduce tedious work before investing in it.

Example 3: Brumby – Training AI for Complex Judgments

Brumby, another client, needed help automating how they recommended diets for horses based on their size and weight. Previously, this process was handled manually by an expert who looked at a horse and made a judgment call. We used a machine learning model to train an AI on horse images, predicting their weight with a certain margin of error.

While we were able to develop a system that worked, the margin of error was too high. Upon further conversation, we realized the more valuable question wasn’t the exact weight of the horse but whether the horse was overweight or underweight. Shifting to a classification problem (overweight, underweight, or just right) made the AI more effective for Brumby’s needs.

Key Takeaway:

AI can help automate expert judgment, but you must identify the right problem for it to solve. Sometimes, a more straightforward question will lead to a more successful AI implementation.

Is AI the Right Solution for You?

So, how do you know if your company is ready to use AI to enhance your data processes? Here are a few questions to consider:

1. Are you manually making inferences from large datasets? If yes, AI can probably help automate that process.

2. Do your employees spend too much time searching through documents? AI can optimize search and retrieval processes, especially when keyword-based search methods fall short.

3. Do you have enough data to train an AI model? Data is crucial for AI success. If you don’t have sufficient high-quality data, AI might not be effective yet.

Lastly, make sure to approach AI with clear outcomes in mind. Don’t implement it just because it’s trendy. As with the Meals on Wheels example, sometimes traditional solutions are more effective than AI.

By understanding your data and the problems you need to solve, you can determine whether AI is the right tool for your business.

Conclusion

AI is a powerful tool, but its effectiveness depends on having the right type of data and a clear business problem to solve. By focusing on outcome-driven AI solutions, companies can save time, automate processes, and make smarter decisions based on their data.