Introducing AI to your Organization_468 x 60 copy 2.png

The world is being transformed by artificial intelligence (AI). Many businesses and industries are implementing this technology to improve predictability, streamline processes, and stay ahead of the competition. But introducing AI software and practices isn’t always easy.

In this course, you will learn how to successfully integrate AI into any organization. You’ll learn how to determine whether your organization could fully benefit from AI, as well as the technical capabilities necessary for deployment. You’ll learn how to perform business analysis and present a case for implementation.

Further, you’ll learn project management best practices to help install AI successfully. You’ll use the Scrum framework to build and deploy AI and to maintain the system when in production to collect data and model scenarios.

AI Project Management: Deploying and Maintaining AI for Business

# Why do AI projects fail?

Organizations that are new to AI should understand that deploying it isn’t always straightforward. They may only see an ROI (return on investment) after several AI projects have been deployed and when the AI team is more mature and proficient at creating a successful product.




It is very common for many organizations to believe that they have the data necessary to succeed. Still, the data is not readily available when they dig a little deeper. Maybe because the data is in multiple systems, extracting that data is extremely difficult. Or, perhaps they haven’t captured all of the necessary data points.


# A quick AI hypothetical

Imagine working on an AI project in which the team attempts to perform predictive maintenance on a specific device. The manufacturer estimates the device’s lifespan to be between 12 and 14 years.





Unfortunately, the team only has eight years of maintenance and repair data and electronic records. Earlier records were stored on paper in four-ring binders. A limited amount of predictive maintenance can be done with the available data.

In the case above, the AI models will fail in production because of model drift. The results may look promising when the models are being trained, but they perform poorly in production.



# What happens without a monitoring team?

Perhaps there hasn’t been a team actively monitoring the model’s performance. How do we avoid this? After the model is in production, we still require data scientists and machine-learning engineers to support it. This needs to be factored into the project timeline.

The end goal is not to create an accurate AI model. The end goal is to continuously monitor the model in production and ensure that it performs accurately according to the boundaries set by the business.

When a software application is deployed into production, engineers will measure latency, disk usage, CPU usage, and so on. AI models in production require a similar level of monitoring.


# Sometimes, it’s not the AI’s fault

Another reason AI projects fail is that the primary business problem wasn’t related to AI. Let’s look at an e-commerce example. An AI-based recommendation engine has been deployed, and it hasn’t made a significant difference to the site’s revenue. What went wrong? 


Let’s find out why AI projects fail and how to prevent that from happening.

__default

AI Project Failure

Why do AI projects fail?