Betting On Bots
The number two global concern today is “Will I have a job tomorrow?” The first one, of course, is, “What are the chances of me having COVID-19?” At the moment, 16.3 million in the US have lost their jobs, in India 130 million are set to lose theirs and overall, 81% of the world’s workforce has had its workplace fully or partially closed. Behind the depressing news of growing unemployment, a drastic phenomenon is unfolding. The nature of jobs is changing. An increasing number of businesses are examining automation as a means to respond to the crisis. Research shows a dramatic increase in the intent to buy RPA as growing waves of COVID-19 smother the world. RPA, or Robotic Process Automation, lies at the heart of software bots. RPA is designed to execute narrow and specialized tasks without human intervention. Between early January and around mid-April this year the increase in average buyer intent for RPA has been 162 per cent. It’s an irony that bots will affect millions more jobs in the near future than COVID-19. But automation will keep people safe from COVID-19. How you ultimately think of RPA and its advanced cousin, Cognitive Process Automation, depends on which question is more important to you: “Will I have a job tomorrow?” or “What are the chances of me having COVID-19?”
Bots in the workplace are going to be a reality. Last year, Forrester made a prescient forecast. It said that “One in 10 startups will begin life with more digital workers than human ones.” In this instance, a Digital Worker is not the same as a bot. Many bots lined up like ducks in a row holding hands make a Digital Worker. Understanding the difference between RPA-based bots and full-fledged Digital Workers provides us insights into a world that is just round the corner and the moral dilemma it presents.
RPA-based bots are like co-workers. We love them because they make life easy by doing all the boring jobs for us. They free up our time to do the tasks that require human interaction. They give us the opportunity to be creative and take risks (which bots can’t!). For example, bots may be used to extract attachments from supplier emails to your organization and put them into a folder called “Invoices”. Another bot may open the invoices and identify the supplier ID and purchase order number. Yet another bot may look up the purchase orders in the organization’s procurement system and match them to the one mentioned in the invoice and so on. These are repetitive tasks that can dull the mind, soak in hours of productive time and often result in errors. Bots can do these tasks with precision and at speed. When you come back from your coffee break, all the invoices in the inbox will have been neatly extracted, stored, matched with existing purchase orders and ready for you to review and for payment processing.
But be wary of the next coffee break. When you come back, there may be no job. The organization can line up a string of discrete task-oriented bots in a continuous workflow – from receiving invoices to creating vouchers for line items, verification of line items, payment advice to the bank, updating payment records and sending confirmation mails to the supplier -- to create a Digital Worker that replaces an entire role, say an Accounts Payable executive. By applying Artificial Intelligence and Machine Learning a collaborative digital co-worker quickly turns into a competitor in the workplace.
This also tells us that it is impossible to have an army of bots running around, so to speak, without digitizing the business. Digital transformation is a pre-condition to being able to leverage automation and bots. The challenge here is not to build and maintain (and even manage the problem of selecting bots to retire when their ROI declines) but to bring a platform approach. This means an enterprise would typically subscribe to a bot store, log in, and select the bots it wants to use, create workflows and run the bots for the duration required— all this without downloading an application or changing enterprise systems. For enterprises accustomed to owning their key assets, this could be a difficult decision. To ensure investments deliver the highest ROI, an enterprise will need to discover processes that can be automated and prioritize them. While this is tricky, most enterprises would begin by selecting processes that address customer and employee pain points. Better still, why not use a technology partner to surface processes that produce the highest ROI?
The opportunities unlocked by roles-based automation are well known. But let’s list them quickly to bring perspective to the idea of end-to-end automation. First, they improve business resilience and keep the lights on (every enterprise would have arrived at this conclusion after being bludgeoned by COVID-19). It reduces the dependency on employees, lowers recruitment costs and eliminates retention costs. In the same chain of thought, it reduces the cost of training and re-training employees. The upsides, especially in today’s environment with a severe talent deficit, are undeniable.
Things are only going to get better from here. Extremely sophisticated AI is on the way to turn bots into amazingly intelligent components of the business. As an example, Elon Musk-backed OpenAI is ready to sell its extraordinary GPT-3-based text generation system. Powered by a neural network, Generative Pre-Trained Transformer-3 or GPT-3 uses a massive 175 billion parameters to generate text that sounds convincingly human. It took the tech world by storm last month by writing synthetic poems, op-eds, philosophical essays, articles and even functional code with minimal prompts. My favourite exploit was asking GPT-3 to express famous movies in emoji. The results were not particularly stunning or amusing. They were innocently accurate. But they did demonstrate that GPT-3 understands the meaning that emojis embody (I bet there are a bunch of emojis that you haven’t figured out as yet; I certainly am foxed by many). GPT-3-based bots could soon be writing mails marketing your products to prospects, they could be helping write elegant annual reports, or composing angry mails to the traffic department that has wrongly fined you for breaking a traffic light—the possibilities are endless.
The point is this: Bots, Digital Workers, Digital Colleagues, Digital Concierges—call them what you will—are going to get incredibly smarter and replace many of our roles. Is this good or bad?
Let’s try and bring perspective to that question. About 110 years ago, Ford Motors put out the Model T and created a new job called “driver”. Today, we have come a long way from the Model T, even though its engine was noted as one of the 10 best engines of the 20th Century.[i] Autonomous vehicles are on the way. They will eliminate the role of “driver”.
Autonomous vehicles are superbots. Many of us would want to see cars with Level 5 autonomy on the streets in our own lifetime. They will take away millions of jobs but will ensure a better world. For starters, we will have fewer accidents. Think about it: Today, around 1.25 million people die globally each year in road accidents, and another 50 million suffer non-fatal injuries. Drunken driving will become a quaint phrase—and thank God for that. Best of all, it will put an end to the need for traffic police, many of whom act on inaccurate information and (honestly) on prejudices. With autonomous vehicles following embedded legal and regulatory requirements, there will be no need for traffic police. I won’t even go down the road that talks about savings in tax payer’s money that go into maintaining the traffic police force.
You may like to remember that “automobile drivers” in 1908 made obsolete the role of horse carriage and wagon drivers. We looked at that as “progress”, didn’t we? Similarly, bots will take away existing roles, but new ones will turn up, making the world a better place. COVID-19 appears to be setting the course for that future in an irrevocable manner and in double-quick time.
