What are the key points for successful AI implementation? (Part 3)

What Are The Key Points For Successful AI Implementation? (Part 3)

Hello everyone, My name is Kenshin Fujiwara and I am the CEO and founder of HACARUS Inc. 

Through this series of blogs, I will discuss a wide range of topics of AI, from the history of AI to practical tips for successful application of AI projects. I hope my blog posts will help you gain a better understanding of AI and solve your business issues.

In the last blog, we studied potential pitfalls with AI implementation. In today’s blog, we will continue to discuss other key points for successful AI implementation, in particular, how to keep moving forward an AI project under highly uncertain circumstances.

Accumulate Small Successes

In an AI project that is highly uncertain, it is important to keep moving forward toward a goal. By accumulating small successes, you can progress even when you have little experience as an organization.

For companies in the manufacturing industry, it is a good idea to start with a small goal like halving the man-hours required for a particular process. To reach this goal, AI development will follow an iterative process of trial and error, with detailed tuning. By setting reasonable goals, you can afford to make mistakes, tolerate them, and move forward step by step, making corrections as you go. 

An alternative idea is to run several small projects simultaneously. In this case, multiple focus areas are established and implemented with a portfolio of AI implementations. If this is the path taken, it is important to understand from the beginning that some of the projects will fail. At the same time, you should aim for success in several of these projects.  

Overcome KKD Barrier

In addition, by proceeding with the project using small issues, the project manager will be able to overcome the KKD (Knowledge Discovery in Databases) barrier by fully understanding the issues of the business. For example, if the company needs X days to collect enough data, the AI should be designed to make predictions at a slightly quicker pace. Similarly, some tasks may occur so rarely that they don’t need to be performed using AI.

This kind of data cannot be noticed by looking at data alone and it is very useful in determining the priority of work. This information also helps with the POC (Proof of Concept) and pilot testing stages. 

Understand the Process of Manual Visual Inspection

In the manufacturing industry, there are highly skilled “masters” who find defective products by using manual visual inspection. These employees often hone their skills through years of experience and it is difficult to verbalize how they make their decisions. 

To better understand this process, it is important to conduct in-depth interviews. Using this data, we can better sort out the data of good and defective products. Sometimes, there are cases where the status of a product is disputed between the inspection team. In these cases, a majority vote is sometimes taken to come to a conclusion. 

By better understanding the business that you are trying to apply AI into, it becomes easier to verbalize what you are looking for by the end of the project. This approach also enables smooth communication with the outsourced AI vendor during the POC phase. 

Start Small Projects

I have noticed that if the person in charge of the project lacks the confidence to move forward, they tend to become deflated. In these cases, it’s good to start small. While the business impact might seem meaningless in the beginning, prioritize small projects for success. This will give them confidence and lead to greater success in the future. 

At this stage, it is important to resist the urge to outsource the entire project. Instead, personnel should be trained in-house so that they can assist with the problem identification step. Following up, the scope of the problem identification step should be set appropriately to increase the probability of success.

In the next blog, I will continue to discuss other key points for successful AI implementation. For your updated knowledge and insight about AI technology, subscribe to our newsletter or visit HACARUS website https://hacarus.com.

 

Subscribe to our newsletter

Click here to sign up