newsroom

For AI to truly take off in an enterprise, it cannot be siloed within the ivory tower of the IT department, examined only by a team of specialized data scientists. To truly succeed, AI must extend to the front line workers who can augment their own domain knowledge with AI-driven recommendations and insight.

Undercover AI

Forward-thinking businesses need to unlock the black box of AI so that front line employees can understand how and why an AI is making its recommendation. There’s also the challenge of making it an intuitive tool that can seamlessly integrate into an employee’s regular tasks.

IBM has focused on software to help make AI enterprise-friendly, and our continuing Watson integration has made more of that software available on IBM Power Systems, such as:

• Intelligent Video Analytics (IVA), which can be used to build AI applications that analyze video footage and trigger an alert when an anomaly is observed. It has pre-built AI models designed to make it easy for an organization’s data scientists to deploy, and even easier for users outside of the IT department to use to augment their abilities
• Product Quality Insights (PQI) with Visual Insights uses AI-powered visual inspection to help identify defects, reduce costs and improve throughput. It uses computer vision to inspect parts, components, and products and identify defects by matching patterns to previously classified defects.

Both of these tools combine with our PowerAI Vision software. This comprehensive AI toolkit for images and videos is designed to allow an organization to train their AI for their unique problems, giving IVA and PQI broad applicability across a variety of use cases.

AI on the factory floor

Let’s look at how these tools could be applied in a hypothetical manufacturing company.

In manufacturing, workers’ safety is of paramount importance. For instance, workers should wear protective gear such as safety googles and helmets. However, factories are often so large that it can be difficult for managers to check that everyone is wearing the proper safety equipment.

Using PowerAI Vision with IVA, this manufacturing company could train an AI to monitor video feeds and flag potential safety violations. This alerts managers, who could then take steps to make sure the worker dons their safety gear.

After worker safety comes the importance of product quality. When manufacturing facilities need to create products to exact specifications, even the tiniest imperfection may lead to product failures in the field. The naked eye might miss such a flaw, but by using PowerAI Vision and PQI to train a model designed to spot defects, workers can augment their abilities.

The key takeaway is that the AI users aren’t data scientists; they are front line workers who can benefit from AI insights.

Putting confidence in AI

Regardless of who is using an AI application in an enterprise, one truth is universal: you need to be able to trust the results and be confident in the recommendation. No quality control inspector is going to scrap a multi-million dollar piece of equipment that AI says is defective unless they can back up that decision. That’s why we are building explainable AI features into our record-setting SnapML machine learning framework.

SnapML uses GPU-accelerated linear models that can scale across multiple GPUs and multiple servers. These models include logistic regression, linear regression, and support vector machines, and are easily interpretable, giving a clearly weighted value to each factor of a decision. These models resemble tree branches, traceable up and down individual branches to see where inflection points affected how the model came to its final conclusion.

Get started today with AI

I am delighted to announce that we are making it even easier for companies to get started with AI by introducing two solutions:

• Power AI DevBox by Raptor: An IBM POWER9 and NVIDIA GPU-equipped desktop workstation designed as a developer system for AI applications. Developers can use the 30-day free software licenses for PowerAI Vision to get started with AI right out of the box. This software is also available for free[i] via our academic program.
• IBM AI Starter Kit: A set of GPU-accelerated IBM Power Systems AC922 servers with IBM Watson Machine Learning Accelerator software pre-installed. This includes major open-source AI software frameworks and management software for managing multiple data scientists and the AI training tasks that they run.

These two simple and inexpensive hardware infrastructure options are designed to help developers, data scientists and enterprise leaders discover hardware systems and software tools to start developing their first AI application.

[i] You must be a member of an academic organization to qualify for the software discounts offered in the academic program. You will be required to provide proof of your academic affiliation to register for the program.

The post Deploying AI to the front line appeared first on IBM IT Infrastructure Blog.