How AI can boost the power of traditional RCA tools

Technical problems can significantly vary in complexity, from problems that can easily be solved by simply using knowledge and experience, to complex and impactful problems that cost a lot of money, time and customer satisfaction. Easy problems make up the high volume while with increasing complexity the number of problems tend to go down (as depicted in the problem solving pyramid below).

                                    Low volume but high complexity

                                            High volume but low complexity

With the rise of technological complexity, interconnectivity, and an increased need for alignment, we see the shape of the pyramid changing as more and more problems move up from the bottom out of the low complexity zone.  

An increasing number of problems move up from the low complexity level

As a result it has become increasingly difficult to solve problems effectively using only knowledge and experience. When we lack experience, we often tend to fall into guesswork which can lead to a high number of unnecessary actions that incur costs, take time, and sometimes even worsen a situation.

How is AI going to help us?

The role of AI in problem solving will depend on the location of the problem in the pyramid.

Low complexity problems at the base: We predict that AI is going to automatically solve low complexity problems that are at the base of the pyramid.

Mid to high complexity problems: With increasing complexity, the role of AI is going to change, from taking full control and lead (automatically solving problems) to rather becoming a sparring partner supporting humans in their problem-solving work.

Which role will traditional problem-solving tools play?

Problem solving tools (such as Fault Tree Analysis, Ishikawa, Six Sigma etc.) help us to structure our thinking and to navigate complexity. In particular in high complexity environments, they can reduce the risk that we jump to false conclusions (Trial & Error) and be of great help to find the root cause faster.

The needed “size and power” of a tool should match the corresponding complexity level.       

However, traditional tools are often outdated, scattered and partially require significant training and practice before they can be applied in a meaningful manner.

As problem-solving becomes more challenging and traditional tools tend to fall short, how do we close the gap?

The answer lies in advancing traditional tools by integrating them with modern technologies such as Artificial Intelligence (AI) and Machine Learning (ML). AI can be a great sparring partner providing information at the right point in time of a problem-solving process.

The Flow Behind Effective Problem-Solving

To understand how traditional tools can be advanced, it’s helpful to revisit a best practice flow for problem-solving. Good problem-solving typically follows five phases:

Step 1 - Understand the Initial Situation: Often, teams jump into action without understanding the problem, leading to delays in resolution.

Step 2 - Increase Understanding: This involves gathering key facts about the problem. Skipping this step often results in numerous irrelevant causes.

Step 3 - Develop Possible Causes: Based on the data gathered, teams develop causes. If steps 1 and 2 are done properly, the causes identified are likely to be relevant.

Step 4 - Determine the Most Likely Cause: Before diving into action mode, it is advisable to relate the identified causes back to the data gathered. This helps to remove guesswork and to align on the most likely causes.

Step 5 – Verify the root cause: As part of this stage teams become physically active and seek to verify their assumptions. Before doing so it is advisable to think how to verify a cause in the most efficient way.

AI in Step 1: Understanding the Initial Situation

A clear understanding of the problem often hinges on the initial description of the deviation. Ambiguous statements such as “not functioning” or “broken” leave too much room for interpretation causing delays in the resolution. A better approach is to identify the observable, measurable deviation.

How AI Helps:

• AI can help to make problem descriptions more precise. It can, for instance, prompt users to eliminate ambiguity from problem statements.

AI in Step 2: Increasing Problem-Related Knowledge

Key to understanding any problem is data. But what data is relevant? Problems can typically be described across four dimensions: What (Identity), Where (Location), When (Timing) and Size (Extent). The most effective approach is to gather data not only about what is not working but also about what is still working fine.

Example: If you switch on your TV and the first channel that pops us has a blurry image, most people will intuitively switch to Channel 2. They gather comparison data. In this case it helps to make a judgement about whether the root cause lies with the channel itself or rather the TV.

How AI Helps:

• AI can enhance data collection by tailoring questions to specific cases (e.g., tangible vs. non-tangible products).

• It can also enrich data by searching through backend services (such as databases) to find corresponding evidence (for instance Log-Files to check for specific times when a problem occurs)

AI in Step 3: Developing Possible Causes

This is where AI can make the biggest difference.

Develop and phrase possible causes: Here it is important to relate back to the data to keep the number of causes reasonable but relevant to the problem.

How AI can help with causes:

• Data-Driven Suggestions: AI can successfully connect the dots in order to suggest high quality causes; the data captured about a problem, insights from previous issues, information from backend systems (tickets, pdfs etc.) and publicly available data. As a general rule: As more complete and accurate the data about a problem, as of higher quality are AI generated causes.

AI in Step 4: Determining the Most Likely Cause

The most likely cause is the one that best explains the data gathered.

Example: In the case below of “No electricity” (see image) the cause “Power outage in the city” can be excluded as the issue only affects one part of the house. Pursuing such a cause would be a waste of time.

How AI Helps:

• With the advancement of LLMs, AI becomes steadily better at reasoning. In Step 4 of the flow it could support by giving inputs regarding a probability of a cause – from clear likelihood suggestions to just providing additional context. Here as a general rule again: As more complete & accurate the data about a problem, as better could AI support in the evaluation of causes.

AI in Step 5: Verify the root cause

As part of verify the root cause, teams are actively testing the most likely causes. Often there are multiple ways in which this can be done. For instance: Causes can be observed, recreated in experiments or simply tackled by trying a fix. Whatever approach is the fastest, safest and cheapest, depends on the particular situation.

How AI Helps:

• AI can help with suggesting ideas on how a cause can be verified, either based on common knowledge or based on documentation in a knowledge-base.

• AI, to some extend, can also verify ideas itself. It could for instance run particular tests in the backend to verify assumptions.

A New Era in Problem-Solving

As industries grow more complex, the demand for smarter, faster, and more reliable problem-solving tools is undeniable. Causetec combines multiple elements to meet these demands and prepare teams for the future:

1. AI & ML Integration: Modern technologies ensure faster, more accurate problem-solving, reducing downtime and inefficiencies.

2. Intuitive Guidance: Causetec removes the need for extensive training by guiding users step by step through a proven methodology.

3. Advanced Tooling: Traditional RCA methods still rely heavily on paper or Excel sheets. Causetec integrates workflow management into a sleek, user-friendly interface.

   
   

The best time to prepare for the future is now. Request a free trial of Causetec today and experience the difference AI-driven RCA can make in your problem-solving.