Industry Insights

Syndicated News

Four Steps to an AI Strategy for Your Business

Creating a strategy for artificial intelligence is top of mind for many executives. As competitive pressures heat up in virtually every industry, leaders are looking to AI to help accelerate their business growth. While the hype around AI is exciting, there is a great deal of confusion around what exactly AI can do, what outcomes businesses can expect from AI and where AI can have the most impact. 

At Box, we believe there is a tremendous opportunity for organizations to use artificial intelligence to extract untapped value from all the unstructured data — images, videos, documents, audio, etc. — that makes up about 80% of all data in the enterprise today. From financial records to product designs to marketing assets to contracts, content is the lifeblood of how businesses work, and AI can help business harness the power of this data and turn it into a competitive differentiator. 

So, how can you go about finding ways to apply AI in your business? After working with customer to solve a wide variety of problems across industries and lines of business with AI, we've pulled together the following guide for how to build an AI strategy for your business content:

1. Start with the right problems

As with any new technology, it's easy to immediately jump to the most ambitious (and complex) implementations of AI. But this is the exact opposite of what you should do. According to a 2018 McKinsey report on use cases for AI, only about 16% of AI use cases are completely new innovations. This means the highest potential for AI in your business lies in improving the way your organization already operates. 

An excellent place to start is with repetitive or manual tasks that involve straightforward decision-making. Often, these low-skilled tasks and processes involve manually processing large volumes of data to either classify the data or to extract details from the data. The decisions made and information gathered as part of these workflows might seem simple but can have significant downstream impacts on how a business operates.

A great example of this is how legal teams manually review and organize contracts. Often, organizations need to organize contracts by type, value and whether it contains particular terms. This type of analysis is essential to protect the business from non-compliance and help unearth revenue opportunities. But legal teams are highly skilled (and expensive) resources, and their time is better spend on more impactful tasks than manual data classification and extraction. This basic yet essential decision-making is a great scenario to apply AI. 

As you start to identify similar tasks or workflows, you can begin to map how AI will fit in. Take note of the type of data you'd be working with (PDF documents, images, user analytics, etc.) and the kind of information you'd be looking to extract. 

2. Define the business outcomes

Much of the conversation around AI is about what AI can and cannot do, but it's crucial to view AI as a tool and not as a final solution to your business problem. The insights provided by AI are just one part of the overarching business process, and as a business leader, you'll need to understand how those insights are going to help drive business value.

In a blog post from Cassie Kozyrkov, Chief Decision Intelligence Engineer at Google, she says that "a common mistake businesses make is to assume machine learning is magic, so it’s okay to skip thinking about what it means to do the task well." Depending on the use case, the impact of AI can vary substantially. You could use AI to accelerate a business process, to provide a more personalized experience to the customer, to help a knowledge worker be more productive or to reduce risk from error-prone classification processes. Having a good understanding of what business outcomes you hope to achieve with an AI implementation will help you measure success and sequence potential use cases. 

In short, don't add AI for the sake of adding AI: use AI as a tool to reach a higher-order goal. 

3. Collect and organize your data

Possibly the biggest hurdles for adopting AI are (1) having proper access to information and (2) having your data in the right format for AI processing. Without these two pillars, it's nearly impossible to reap the benefits of AI.

For many organizations, data fragmentation across different systems and tools blocks AI initiatives. For AI to work correctly (and to provide the most benefit to your organization), you need as much data as possible. Having your data in a single centralized platform, such as a cloud storage platform or data warehouse, is critical for AI analysis. It's important to also consider the openness of the cloud platform you're working with; to leverage AI providers in the cloud, you'll need a platform that has APIs to send and receive data with these AI technologies. 

Moreover, many of the AI solutions available today are designed to handle "clean data," meaning that the data must be available in a format that is readable by the algorithm. If your organization is looking to use AI, for example, to gain insights from recordings of calls, you'll need to have high-quality recordings that allow for objective analysis of the data. If the recordings are muffled, low-quality, broken, etc., the AI processing won't be able to provide the right level of analysis. You'll need your data in a clean, consistent format for AI to recognize patterns in the data.

4. Choose the right technology

Once you've defined the problem you want to solve, aligned on the potential outcomes and ensured your data is ready for AI, it's time to determine what type of AI to apply. 

The good news is that, similar to the introduction of infrastructure as a service (IaaS) to provide scalable, cloud-based services for necessary infrastructure, many of the leading cloud providers are taking the same approach to AI by providing machine learning services like computer vision or speech-to-text conversion as hosted services. You can think of this as machine learning as a service (or MLaaS). 

There has been substantial innovation in the MLaaS space, with machine learning technologies of all sorts being made available to help accelerate AI initiatives. From leading cloud providers like IBM, Google, Microsoft and Amazon to more niche and specialized providers, there are dozens of algorithms to choose from, all with varying degrees of performance, quality and specificity. Some of these providers even provide AI algorithms designed for a specific industry or use case, like computer vision algorithms specific to car accident photos for insurance claims. 

There may be several services that meet the requirements for your use case, but choosing the right AI technology involves far more than assessing just the raw capabilities of the technology. You'll want to test what the algorithm does, ideally with a sample set of data, and see examples of the AI technology in practice in the form of case studies and customer references. You'll also want to explore the security, privacy and compliance standards of the technology. This could be especially important if you are in a heavily regulated industry like financial services or healthcare. And last, you'll want to make sure that the technology is developer-friendly, with robust documentation, SDKs and sample code. An IT development team will need to integrate this technology into your IT stack, and working with developer-first technologies will accelerate that. 

View original content: Here

Related Box News:


Are Governments Providing Improved Digital Experiences During a Global Pandemic?