Automation in reporting: breaching the cognitive ceiling with natural language generation

Robowriters are software robotic agents that write reports the same way you or I would; only a little bit better in terms of speed, consistency and in interactivity potential.

Employing Natural Language Generation (NLG) algorithms, Robowriters plug into an organisations information sources. They then derive pertinent facts from the data and convey this information in natural language narratives that rival human edited text.

With major news organisations such as Forbes and Associated Press using Robowriters, chances are you have already consumed machine-produced narratives. The potential of Robowriters is endless; there is no reason why such technology can’t be adopted into any major reporting process across any industry.

The crux of Natural Language Generation, and its potential

At the heart of Robowriters is a family of Artificial Intelligence algorithms that are broadly categorised as Natural Language Generation (NLG). While implementation differences are of course pertinent, NLG is implemented as a process designed to optimally transform non-linguistic or semi-linguistic facts in a natural language communication that serves specific communicative goals.
This is significant from the perspective of a business’s journey to Integrated Automation. NLG enables the advancement from a robotic to a cognitive automation strategy by moving the dial on performing cognitive tasks that require human judgement – automation of tasks is no longer limited to basic rules based repetitive tasks.

Public-facing NLG applications range from financial reporting to the tailor-made obituary for Marvin Minsky, an AI pioneer. In organisations, NLG supports narrative generation for operational reporting, client-facing personalised content, fraud detection, regulatory reporting, decision support and performance optimisation. The limit is our imagination.
Some example narratives on anonymised data, generated by NL:


What is the value of NLG?

Communication is assessed by its efficiency in transmitting necessary information or to induce and support action. At the intersection of data growth and picky information consumption patterns, successful internal and external communications are becoming increasingly hard due to:

  • Information Overload: Data proliferation accelerates as monitored digital channels permeate everyday interactions and transactions. At the same time, the proportion of collected information used to support insights is decreasing. We are hoarding data but on average failing to get real value from it.
  • Complex Organisational Structure: Modern organisational designs are seeing their user base of reporting broaden as decisions are becoming increasingly data-driven. Proportionally, diverse users’ backgrounds need to be serviced with communication that is appropriate for the user’s data literacy and contextual knowledge. Proliferating an organisation with insights and information in the form of natural language will be important in making data-driven decision making a nimble, accessible process, removing reliance on the consumer to understand underlying data or information in its raw form.
  • Cognitive Bias: Seeking out pre-conceived insights from a set of data, as opposed to performing objective analysis – increases as humans strive to make sense of increasingly complicated datasets and data models. As data growth transcends volume and impacts variety and velocity of data, the risk of cognitive biases and its limitations in the production of valuable insights become more material.
  • The rise of the Bots: With business processes automated using Robotic Process & Desktop Automation, and statistical learning methods applied on huge volumes of data, narrative descriptions and explanations of bot actions and Machine Learning (ML) insights will be important in enabling control and integration of robotics and AI in ‘human-in-the-loop’ process designs.

NLG’s key features are addressing the above challenges in designing a future-proof reporting architecture. The graph below illustrates the relevance of the key NLG features in addressing these challenges.

What reports do I automate, and when?

Most reports can be automated, but in some cases, this may not result in noticeable value gains for the organisation. Return On Investment (ROI) maximisation is dependent on the maturity of NLG integration in reporting processes.
Initially, ROI is maximised by automating high volume and effort stand-alone reports. As simple use cases are operationalised, advanced use cases become tenable by leveraging the already built NLG analytics models and other robotics and cognitive automation ecosystem initiatives (e.g. Analytics, Machine Learning, Natural Language Processing).

The chart below illustrates indicative (but not exhaustive) types of reports and their relative prioritisation across the three key horizons outlined above.

NLG is one of many next steps to advance your organisation towards Integrated Automation, breaching the ceiling of automating rules-based tasks and entering the world of processes that require cognitive capability.
At the centre of processes and decision making, data that lies in your systems has a story to tell, and there are increasing challenges in ensuring its value is surfaced to create impacting value for your organisation.
And as early adopters continue to leverage NLG as a solution and provide real examples of competitive advantage, adopting NLG key’s skills to meet these challenges has become a reality that is within the reach of organisation’s digitisation agendas, with reporting being the first hotspot to realise early transformation.

Want to stay up-to-date?

Stay on trend and in the know when you sign up for our latest content