How is AI-powered automation changing business for good?

Automation was the driving force behind the Industrial Revolution and was what allowed us to multiply our productivity and free up leisure time. Now that AI has grown in its application and funding, people and businesses are looking into how it will achieve a similar thing. The discourse has gone from if to how we will integrate intelligent systems into operations. 

Automation in the sense of rote task execution is just a glimpse of its potential (and, in fact, could be a weak point of AI as it’s often not repeatable). Instead, it’s human-like agents and cognitive capabilities that are the method of increasing productivity. That is, like humans, to scale up intuition and problem solving, rather than just completing routine, repetitive tasks.

What is AI-driven automation?

AI’s blessing (and sometimes curse) is that it doesn’t pertain to a rules-based system like traditional programming. With natural language processing and other forms of machine learning, it can handle unstructured data and non-linear tasks. Its ability to learn and adapt makes it closer to the domain of human cognition than machine.

AI driven automation can help unify disparate systems and processes where fixed rules aren’t possible. One example might be a health insurance broker being able to streamline its document-heavy workflows, perhaps acting as the first layer analysis and prioritization. For a software developer, it may be a way to write easy bits of code like an intern might, and just like an interns work, would need to be checked over.

This cannot typically scale indefinitely, then, because a lot of AI’s functions are first layer, meaning it’s still constrained by human assessment to tick it off. The same goes for reading MRI scans, where it can actually outperform humans, though still isn’t trusted to have zero human analysis, meaning it’s currently limited in scale. But, in many instances, human second-layer checks take less than 50% of the time it would with full responsibility, meaning productivity can often double. 

How AI automation is revolutionizing core business functions

One significant impact has been in delivering hyper-personalized customer experiences (and it’s no surprise that the stakes of individual interactions are lower than an MRI scan or the codebase for an app). Because it can analyze vast datasets of customer interactions, AI models enable a level of personalization that was previously unattainable with chatbots. This can lead to greater loyalty and conversions, and will soon be good enough to be passed off as a human interaction. This is truly scalable, in part because the stakes of individual errors are lower, but also because the human feedback for “rate how we dealt with your problem” provides a reinforcement learning opportunity for the AI to constantly improve - and guardrails on offence can be placed in programmatically on top of the AI.

Businesses where the feedback on the success of the AI is direct, like in financial market forecasting, are areas where it’s excelling. A McKinsey report found that 78% of organizations now use AI in at least one business function. And, even in businesses who do not implement it, it’s still having a positive effect - an employee may use AI to find the correct Excel function, which is faster and more bespoke than Google.

Reskilling the workforce for an automated future

Concerns about job displacement often overlook the more nuanced reality. Many of the above examples include a collaboration between humans and AI, and an understanding on AI’s limitations and where it’s prone to error. While some jobs may be replaced, many will be a matter of training and learning how to increase productivity with AI.

The World Economic Forum's Future of Jobs Report projects that automation (in general) may displace 85 million jobs, yet it could in fact create 97 million new roles too. The challenge will not be in the scarcity of work, but a skills gap, particularly with the technology adapting so quickly. 

To bridge this, organizations are looking at workforce transformation. According to one report, 85% of employers plan to upskill their current employees in response to these changing needs. 

Where we’re headed

Looking forward, AI automation is heading towards greater integration and sophistication. The concept of hyper-automation, which is the orchestration of multiple AI and RPA tools, is becoming an experimental goal. This looks at automating entire, complex business workflows. 

As AI's role in decision-making expands, the demand for transparency is pushing the development of Explainable AI. This is to help show the logic behind automated decisions so it’s not a black box. The issue with this is that unsupervised learning models can inherently be abstract, and new compliance laws surrounding this could be a hit to AI’s potential. 

Data privacy and algorithmic bias remain another obstacle too, and so more money is being poured into cybersecurity and compliance to counter the growing influence of AI within operations.