AI: Taking Automation to Autonomous

AI: Taking Automation to Autonomous

Rapid developments in the field of AI have paved the way for multiple possibilities. One such possibility is to blend AI with automation technologies and to create a new world of ‘intelligent automation’. The buzz around GenAI is loud and ubiquitous. Much is being said about this branch of AI. However, in the world of mission-critical manufacturing facilities, applied automation must be ‘deterministic’. Everything that needs to happen must be firm and non-probabilistic. Hence, the AI technology must be much more developed and trustworthy. The outcomes from an AI algorithm that are fed to the automation system must be dependable and error-free, else the consequences can be disproportionate and, at times, disastrous. To ensure this, many enterprises are perfecting the more difficult branch of AI—machine learning. Machine learning is a subset of AI that allows computer-based systems to learn and improve without being explicitly ‘programmed’. It uses algorithms to analyse large amounts of data, learn from the insights and then make informed decisions. As machine learning models can learn from data without relying on rules-based programming (otherwise known as unsupervised learning), with sufficient ‘training’ and vast amounts of data, it can become deterministic. Machine learning came into existence as a scientific discipline in the late 1990s as steady advances in digitisation and cheap computing power enabled data scientists to stop building finished models and instead train computers to do so.

Change is coming so fast that human-in-the-loop involvement in all decision-making is rapidly becoming impractical. Looking three to five years ahead, we expect to see far higher levels of AI as well as the development of distributed autonomous systems (DAS).

These self-motivating, self-contained agents, formed as corporations, will be able to carry out set objectives autonomously without any direct human supervision. Some DASs will certainly become self-learning and self-enhancing. One school of thought views DASs as a threatening proposition.4 But by the time these systems fully evolve, machine learning would have become invisible in the same way technological inventions of the 20th century disappeared into the background. The older technologies get refined and rooted in the newer evolving technologies. As they mature, they are taken for granted. Upon maturity, they are expected to be harnessed and put to appropriate use. When automation moves towards autonomy, the role of humans will be to direct and guide such machine learning–based algorithms in the background to seek even more highly complex, multidimensional goals and objectives that are otherwise extremely difficult to achieve manually.

No matter what fresh insights such DASs bring out, only human managers can decide the essential and relevant questions, such as which future business problems a company or a manufacturing facility is really trying to solve. Just as humans need regular reviews and assessments, so do these ‘intelligent systems’. Their work will also need to be regularly evaluated, refined and perhaps even ‘fired’ or shifted to pursue entirely different ‘roles’ by AI-aware executives with real-life experience, wisdom gathered over time and industrial domain expertise. The solution will be neither machines alone nor humans alone but the two working together effectively.

Before we get to the fully developed DAS, there is one more stage—intelligent process automation (IPA). In simple terms, IPA ‘takes the robot out of the human’. IPA leverages a set of new technologies that combines fundamental process redesign with robotic process automation and machine learning. It is a suite of business-process improvements and next-generation tools that assist the knowledge worker by removing repetitive, replicable and routine tasks. It can radically improve manufacturing by simplifying interactions and speeding up processes. IPA mimics activities carried out by humans and, over time, learns to do them even better. Traditional levers of rule-based automation are augmented with decision-making capabilities thanks to advances in deep learning and cognitive technology. IPA promises radically enhanced efficiency, increased worker performance, reduction of operational risks and improved response times and customer journey experiences.5