Be AI-savvy! How robotic process automation (RPA) can boost companies’ operational efficiency

11 Oct 2023

Be_AI-savvy_Blog_1

Artificial Intelligence (AI) is currently all the rage. Chatbot and generative AI are surely the focus of attention. However, robotic process automation (RPA) is the ultimate solutions that can truly help businesses accelerate the process of digital transformation and boost their competitiveness. By mimicking human intelligence and automating decision-making processes, RPA enables companies to process a huge amount of data more quickly and more accurately, thus saving time and costs while unlocking employes’ potential. With RPA, companies can open a new chapter of business development.

Table of content

What is AI?
Going AI: an urgent task
Examples of AI in daily life
Three ways of learning AI
Enhance operational efficiency through RPA

What is AI and RPA?

Simply put, AI is a technology that uses natural language processing, perception and computer vision to enable a computer system or software to learn, think, create, and solve problems like humans. The academic world regards 1956 as the birth year of AI. That year, a number of scholars including John McCarthy, aka the father of AI, and Marvin Minsky hosted the first ever conference on AI at Dartmouth College in the US. During the event, the term AI was defined for the first time and the idea of algorithm and a related theoretical framework were introduced.

RPA, on the other hand, is a type of software that completes repetitive tasks on behalf of humans. In the process, no hardware is involved. RPA is designed to help humans carry out tasks, ultimately boosting productivity and corporate competitiveness.

Going AI: an urgent task

Without doubt, the current AI wave has been triggered by OpenAI’s AI chatbot Chat GPT. The company’s co-founder Greg Brockman has said on X (former known as Twitter) that within the first five days after ChatGPT was launched as a test version in November 2022, the number of its registered global users already reached more than a million. All of a sudden, other AI applications such as generative AI also garnered widespread attention. According to a survey, 64% of business owners are positive about the future of AI, believing that it will improve customer relationships and increase productivity. An IBM report indicates that 25% of businesses are adopting AI to address labour and skills shortages, and the AI adoption rate of Chinese enterprises is 58%, the highest in the world. All this suggests that Hong Kong companies should hurry up and adopt AI if they want to be integrated in the Greater Bay Area.

Examples of AI in daily life

When it comes to discussions on AI, chatbots OpenAI’s ChatGPT and Google Bard are very much in the limelight. But then many applications have already found their way into different aspects of our life, bringing us great convenience

Generative AI	Text, images or music are generated automatically based on prompts provided by the user. A board game company owner recently used AI to generate the painting Théâtre D'opéra Spatial and won an art competition. Though controversial, the image illustrates how high-quality AI-generated artwork can be mistaken as the work of a real artist.
Intelligence recommendation system	This system can gauge the user’s personal preferences and recommend them the right content based on their browsing history, preferences and behaviour. Common examples include hotels recommended by travel websites, movies or songs recommended by online music and film services, and interesting content recommended by social media platforms. Consequently, user experience can be improved.
Fraud alarm	AI can automatically collect intelligence online, analyse suspicious behaviours and issue alarms to companies. For example, when banks evaluate customers' credit risk and when network security surveillance companies examine whether logins are fraudulent, many use AI analysis technology to reduce risks of financial fraud and identity theft.
Healthcare	AI can help medical professionals improve diagnosis accuracy and bring down medical costs by studying patient cases and analysing images in huge quantities.

Three ways of learning AI

AI can be applied to a broad range of areas, but it does not exist without a context. Besides, there is no such thing as one size fits all. To attain a high level of precision for an AI application, a computation model has to be trained through data gathering and machine learning. To this end, three learning methods can be used: supervised learning, unsupervised learning and reinforcement learning.

Supervised Learning	This involves using an AI system to analyse labelled data and work out the relations between different data. Upon completing the training process, the system is able to make accurate predictions for new inputs. The supervised learning approach is best for making predictions. For example, when AI is used to identify data in a file or a table, the trained data must contain a great deal of tables in different formats, and columns have to be labelled in advance. This way, the AI system can grasp the differences between different columns as well as data that should be input in the columns. When the system encounters an unfamiliar table format not available in the database, it will be able to identify different columns in different parts of the table, classify them and store them accordingly.
Unsupervised Learning	This involves delegating an AI system to analyse uncategorised data and work out the correlations of different data for classification. After an analytical model is established, the system will be able to handle unclassified data input in the future. In case of small differences between two sets of trained data, the data can be compressed. For example, shopping platforms can conduct analysis based on customers’ gender, age and interests. The unsupervised learning method can also identify correlations among customers that go unnoticed by humans. Once training is completed, the AI system can recommend products of interest to customers with a high level of precision.
Reinforcement Learning	Business strategies and behaviours are optimised constantly through trial and error and feedback mechanisms, thus maximising rewards. Take self-driving as an example, an AI system can learn different driving situations and traffic signals to achieve safer and more efficient driving. In addition, reinforcement learning is often used to improve workflows, in that AI system can work out the most efficient or productive workflows on its own.

Enhance operational efficiency through RPA

Although AI systems are springing up like mushrooms, RPA remains the technology that can truly enhance companies’ operational efficiency. To be more AI-savvy, business owners can start with the details and identify the repetitive tasks that can be taken over by RPA , so as to address labour shortage, ease employees’ workload and free them to focus on more important tasks.

Document management and approval solutions	The content and content categories of paper documents or e-documents are analysed using a multi-function printer and Optical Character Recognition (OCR). A preset approval process will deliver the documents automatically and remind designated employees and supervisors to approve and file the documents. All this helps speed up a work process and minimise errors.
Workplace design services	These services analyse big data related to how company employees use office facilities. Motion detection equipment can automatically build an intelligent management and control model. For example, a smart lighting management system can help companies trim energy costs while addressing environmental concerns and meeting ESG (environmental, social, and corporate governance) standards, to which companies are attaching increasing importance.
AI robot	Through natural language processing models and speech recognition technology, AI robots can comprehend human speech and provide answers to specific questions. It can also carry out designated tasks according to instructions. Website chatbots, delivery robots in restaurants and smart visitor management systems in offices can all ease demand for labour by virtue of AI.