Intelligent Digital Robots or RPA 2.0

We live in unprecedented times of exponential growth of technology. With AI solutions knocking on every doors, it is time to think how it will influence the nature of jobs we do.
business intelligence, deep learning, digital robots, machine learning, process automation, rpa,

1. Robots


In the late 18th century Western world went through Industrial Revolution, changing from hand production methods to machines. Since then the world has never been the same as we went through faster and faster changes within job structure and society as a whole. Lots of traditional jobs were lost, but even more came to be.


We haven’t seen a revolution at such a scale until late 20th century when our civilization has adopted digital devices, which became ubiquitous at the beginning of 21th century. Suddenly we are living in the future, where robots – both digital and physical (e.g. cleaning Roombas, autonomous Teslas, Boston Dynamics robots or industrial arms) – are coexisting with us.


Digital robots came to be already in the 90’s with the rise of Internet. Nowadays scraping and spamming robots on the web are common. More business oriented robots are those existing in Robotic Process Automation systems (RPA for short), which are programmed to perform repetitive tasks in offices – from copying data to spreadsheets, to performing computations and passing results through emails. You just show them task by task how you do it and they mindlessly repeat that over and over again. It takes time to custom those robots, but if you have repetitive, tedious task done daily by dozens of people it’s most often worth the time. Just like it’s worth to have people replaced by automatic arms in factories, so that the whole factory space is safer while tedious, physical tasks are delegated to machines.


This first wave of digital robots is a digital equivalent of mechanical arms in factories – they don’t have any ‘vision’ or ‘sensing’ and they are able only to perform the same type of action they were programmed to do without adjustments to changing conditions or any sign of creativity.



2. Artificial Intelligence


Artificial Intelligence was debated ever since computers came to be because the idea of a computer slowly replacing humans seemed tantalizing. Due to hype and too high hopes, AI as a computer science domain experienced two winters (reduced funding, bad press) for most of 70’s to 90’s.


Situation changed in the last years of 00’s, when machine learning algorithms outperformed classical solutions in vision-related tasks. This was caused by the growth of computing power and access to large databases (digitized content). Machine learning itself is an approach to solving problems by letting algorithms adjust themselves on incoming data. A programmer doesn’t have to code all the necessary functions and features. It’s enough to put only the initial architecture in place and let a machine/computer learn on examples.


Rise of AI changed everything – from the way we do shopping (mostly online) to navigation (Google Maps) to even life expectations (every process has to be smooth and pleasant). Machine learning paradigm of sketching a draft and letting a machine figure out the rest won over the classical programmistic paradigm of hard-coding every function and feature into the structure.



3. RPA 2.0


With the rise of AI, digital robots can finally ‘see’ and ‘reason’ on incoming data. They don’t have to blindly follow rules, but can improvise on the spot. Digital Robots can finally become Intelligent.


From the technical point of view, machine learning – especially deep learning and reinforcement learning – allows a great robustness in RPA systems. You don’t have to program by hand the repetitive tasks at your office job. It is enough to let the machine run in the background, analyse what you do for a couple of hours or days, and come up with potential automation solutions.


However close that sounds to Artificial General Intelligence, which is still far away, these kinds of intelligent digital robots will soon flourish in all business verticals (and in general knowledge verticals), where one operates on highly limited dictionaries/ontologies – e.g. tax interpretations, financial reports or legal agreements.


We are currently at the very beginning of this new revolution which will limit tedious, boring tasks and allow people to focus on truly creative jobs. The change will occur in jobs which seemed at first out of reach for automation: consultants, bankers and lawyers. Within the next 5-10 years those occupations will greatly change. They will be more about human interactions and empathy than data manipulation. Office tasks like preparing slides, drafting a legal agreement or auditing will go away. This will truly be an equivalent of Industrial Revolution, but this time in the digital world. And that’s only a small part of what AI can give us in the future.


Prepare for the change!


written by Przemek Chojecki

business intelligence, deep learning, digital robots, machine learning, process automation, rpa,