Artificial intelligence (AI) is everywhere in our personal lives: from predictive text to home heating, route-planners to face recognition apps. Take smart speakers for example: in 2020, over 136 million smart speakers were shipped worldwide, estimated to increase to over 409 million by 2026. Putting that in context, in the US 35% homes have at least one smart speaker!
A world without AI is almost impossible to imagine; it transcends so many aspects of our daily personal activities, that we readily embrace new digital technology that can make our lives a little easier. And with machine learning, software gets better and better at predicting and mimicking human behaviors, for example, Netflix suggestions, re-marketing advertisements, or suggestions for finishing whole sentences as you type.
Why is it, then, that deploying intelligent automation in the workplace can feel more of a threat than an opportunity? In business, the most valuable asset of any organization is data, and when it comes to collecting, processing, sharing and storing data, automation can power productivity. In a recent study by IBM Institute for Business Value, for organizations scaling intelligent automation, 90% of executives say it actually creates higher-value work for their employees.
Managing and handling today’s inconceivable and increasing volumes of data, at unbelievable speeds, is where machines can supplement the work performed by humans. Streamlining and automating workflows is where we can let AI do some of the work for us, freeing up time for innovation and creativity to gain a competitive advantage.
Workflow automation has traditionally been driven by RPA (Robotic Process Automation), which mimics rudimentary human tasks. The key objective of automation is to boost efficiency and productivity, helping to generate time and cost savings. RPA takes us only so far, but with advancements in the field of artificial intelligence we can now automate more complex tasks by simulating human behaviors too.
By leveraging cognitive computing, we can automate information or data analysis, automate the recognition of patterns, and perform relevant actions. Machine learning can also be integrated within systems, enabling predictive decision-making and deductive analytics, based on historical data trends, to approximate outcomes that would be expected of humans.
For example, a Loan Auditor is expected to go through hundreds of pages, identify relevant documents, gather the pertinent data, and make decisions on that information. The same workflow can be configured leveraging AI technologies, such as natural language processing (NLP), computer vision, deep learning, and machine learning, which is not possible with RPA.
Rather than solely relying on either humans OR machines, building solutions that amalgamate the two enables tedious or repetitive tasks to be automated, while allowing humans to maximize added value. This amalgamated approach helps bridge workflow gaps that can only be partially automated, by ensuring human intervention where required, better known as Human in the Loop. It integrates AI technologies into a workflow but ensures enough opportunities for humans to interact and influence its direction. The challenge of implementing a composite workflow is in the creation of a seamless interface between man and machine. For maximum user engagement and interactivity, this interface is pivotal to the success of workflow automation, for example, collaborations, notifications, UI/UX design, etc.
Apart from workflow automation related to document processing, machine learning is also leveraged in processes which involve analyzing vast volumes of structured and unstructured data. For example, a Quality Analyst for a call center can listen to and audit a limited number of calls in a given time. Whereas, AI can be used to automate the transcription of large volumes of calls, after which NLP will highlight insights or flag issues.
Similarly, in the healthcare industry, with an increase in telemedicine solutions for patient care, a Health Care Practitioner (HCP) might prescribe medications based on the presenting symptoms only, not having immediate access or time to review their full medical history. This is especially the case where a different HCP is consulted at each appointment. Using AI we can now enable a clinical decision support system, which can validate a prescription against any known allergies or conflicting medicines, immediately flagging this to the HCP.
In business, the vastly changed digital landscape has not equated to a reduction in employees. So integrated, trusted and relied upon has it become, that we simply take its availability for granted. Instead, it has supported the development of new roles and professions, new expertise and skills, and freed up time for innovation and creativity by taking care of manual, repetitive tasks. In fact, Deloitte reports 71% of organizations that are already scaling and implementing intelligent automation, have a clear vision and ambition for intelligent automation.
But even as machines gradually encroach more into our working lives, humans will always be in the loop. Humans possess fundamental reasoning skills, being able to imagine ‘what if’ scenarios and prompt ‘why’ questions. While AI can help us arrive at answers to these questions, through predictive, descriptive, diagnostic, and prescriptive analytics, it ultimately depends upon the data that we provide.
With human brains possessing approximately 86 billions neurons, and an estimated 100 trillion connections, even neuro-scientists admit we have a long way to go to solve the mysteries of the brain. So, it stands to reason that machines also have a long way to go to perfectly mimic human instincts, emotions and intuition that truly drives innovative breakthroughs.
Bottom line: As we move forward, AI elements will be the norm for intelligent, data-driven, continuously improving automation solutions, adding true value when used in conjunction with human centric services.
(The article has been co-authored by Gautam Ghai, Co-Founder & CEO and Sukhmani Gill, Director, AI & ML, SourceFuse Technologies and the views expressed in the article are their own)