Artificial intelligence (AI) is a fast-evolving technology, and it may be very hard to catch up with early AI adopters. Yet despite its obvious competitive advantage, organizations are finding it difficult to exploit the advantages of AI in practice.
Why? AI can deliver true value to serious adopters in the form of greater performance, which in turn can contribute to greater profits. So it comes as no surprise that many organizations have been quick off the mark to identify areas of application. But starting out with AI is far harder than the hype around it would have you believe.
To get the most out of AI it is recommended to use a combination of both learning and rule-based methods. The problem is that most organizations don’t have enormous data pools. The technology’s dependence on robust digital foundations and the necessity to train it using clean data means there are no shortcuts. Deep learning systems are thus not the first choice on the menu, but organizations should know that “mere” algorithms can be utilized to draw meaningful conclusions from even the smallest data sets.
There are endless AI deployment examples. Human resources can use AI to simplify recruitment. Removing human bias from the recruitment process encourages and highlights an inclusive workplace. Production can tap into the power of AI for predictive maintenance and planning. Customer service can exploit AI to gain real-time insights across customer touchpoints, to optimize agent availability and wait times.
Harvesting insights from unstructured content
Cognitive services and content analytics open up numerous fields of application that help companies to make targeted use of their data and to gain insights from unstructured content.
Sentiment analysis can be used to optimize customer service, for example. Sentiment analysis using AI interprets and classifies emotions – positive, negative and neutral – within text data, which can stem from social media to e-mails, blogs and online comments. Over time, AI has become much cleverer at interpreting the tone of sentences and not just single words. This is a very powerful marketing and customer service tool for organizations. By knowing what is being said about them, organizations can improve consumer engagement and better optimize their branding messages.
Named Entity Recognition (NER) is another area where AI can help in extracting information to identify and single out named entities and categorize them under pre-defined headings. Customer support departments can use NER to pull out complaints on certain products and their locations. News and publishing houses can use NER to categorize articles in defined hierarchies for faster searches.
To identify risk and fraud, AI can find patterns and anomalies in documents such as invalid contract clauses and fraud attempts. Banks and some large online retailers are already utilizing AI and machine learning as a direct response to more sophisticated cybercrime.
Blending AI with information and process management
Many of the methods being used work with structured data or images, but not so well with text as unstructured content. However, around 80% of all company data such as letters, contracts or e-mails are unstructured. This represents valuable information, but it must be analyzed, prepared and merged with structured data to become usable for AI with high data quality. Unfortunately, data is often isolated in silos, which complicates access and use. This is why AI should be directly combined with information and process management.
An ECM solution can greatly simplify this by bringing together data from different sources such as databases, servers, ERP, CRM, Microsoft SharePoint and file directories. The idea is to prevent redundant data storage and create order in documents through versioning and metadata. With cognitive services thus integrated into the platform, the AI services are automatically available for all applications.
In addition, to meet the extraordinary appetite of an AI system, a context-sensitive software that can manage, store and scale data horizontally is required. This is where an ECM solution comes into its own.
Knocking down the walls between content and data
AI is making groundbreaking changes in the way companies carry out processes and interact with their customers. For it to deliver on its promises, however, it needs accurate and properly managed data. ECM has a huge role to play here and should be central to any AI strategy for a company that wants to get ahead of the curve.
Latest posts by Gregor Joeris (see all)
- RPA Takes On the Operator Role and BPM Enters as Engineer - September 1, 2020
- Is Your Contract Management Strategy Ready for the New Normal? - August 17, 2020
- Deploy AI Now to Get Ahead of the Game - April 14, 2020