Our latest series looks at artificial intelligence in the management consulting industry. Cocoa Gao, CPA, CA brings her expertise and analyses a number of areas over the next five articles.
She starts off with an intro to AI and its current application (piece 1 - see below), sheds light on how to turn disruptions into opportunities by using AI (piece 2), illustrates how AI can be used in the back office (piece 3), offers insight into a new way of operating that is enhanced by AI (piece 4), and finishes off by assessing how AI is creating a new competitive landscape (piece 5). Stay tuned!
Our lives have been deluged with AI in many ways. We receive an immense amount of information on this topic in our everyday life, and constantly hear stories about how it will drive our (future) lives. The overwhelming amount of information we receive makes it hard for some of us to define what AI really is; AI simply becomes a buzzword.
So, what is AI?
AI can be defined as “A broad area of computer science that makes machines seem like they have human intelligence”.
AI was born in the 1950s. John McCarthy first coined the term ‘Artificial Intelligence’ in 1956 at the Dartmouth Summer Research Project conference. From then, AI has experienced a rollercoaster of successes and setbacks. AI development first flourished between 1950s and 1970s; with research focusing on specific intelligence, such as speech recognition and optical character recognition.
‘AI Winters’ came shortly after the initial excitement; failure to achieve scale, technical limitations, and complex real-world problems led to disappointment and disinterest in AI. Today, we are in ‘AI Spring’ again, and in contrast to before, the research now focuses on specific problems – machine learning, deep learning, high-speed internet and connectivity.
How are Robotic Process Automation (RPA), Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) different?
As we see can see from the image above, AI, ML, and DL are all related to each other. AI is actually the superset of DL and ML; DL is a subset of ML, which is a subset of AI. AI is data-driven technology which learns from previous outputs and extracts insights from data to perform intellectual tasks.
By contrast, RPA is a separate technology that helps humans automate their business processes. RPA automatically executes repetitive and rule-based tasks established by its initial programming; it cannot learn.
In short, RPA ‘executes’ while AI, ML, and DL ‘thinks’. Now, let’s dig deeper into each of the technologies.
Where is RPA used?
RPA takes the repetitive, tedious, and time-consuming tasks away from human attention so that employees can focus on more complex and value-added tasks. An RPA bot can input well-classified data into accounting software to generate orders, execute payments, and send confirmation emails automatically. It can also be programmed to collect social media statistics, fill forms, read and write into databases, and perform calculations without human intervention. After data are collected, RPA can make the data accessible to AI; AI then learns based on the data that is handed over.
How is AI used?
AI machines are usually classified into three groups – Narrow AI (takes over specific tasks), General AI (is equipped with capabilities that are similar to human intelligence), and Super AI (overtakes human intelligence).
All AI machines today fall under the category of Narrow AI. Input management is a key area where companies across industries have begun to use AI to increase productivity. For example, AI processes large volumes of information from emails, documents, and presentations by identifying the topics and sentiment from unstructured text. Before AI-based solutions, human intervention is needed to decide which departments should receive the information, which documents should be prioritized, and who needs to be involved.
How does ML add value?
The purpose of ML is to train the computing systems to 'learn' by themselves using the large amount of given data and to make predictions. Better predictions support better decisions, thus improving efficiencies and effectiveness.
Loan approval is an example of ML application. A computing system receives the loan application (i.e. input) and uses the applicant’s past data, such as income, credit history, and employment history to predict whether the loan can be repaid. Online ad placement is another example of how ML-based solutions can be used in the online advertising industry. Data like a user’s social media profile and browsing history are fed into the software. The software then learns from the input data using ML algorithms in the hidden layers and draws further insights from the user’s demographics and browsing history metadata. In the end, the software determines what types of advertisements the user will be interested in.
What is DL?
DL takes ML a step further – the system can learn from the real world and adjust its learning model automatically as it takes in new information. DL is usually done through neural networks. Different layers of systems are stacked together to mimic the function and structure of a human brain.
Chatbots, voice assistants such as Siri and Alexa, and facial recognition are samples of DL. Spoken words are captured by voice assistants through microphones. From there, DL systems analyze the context in which spoken words are captured in, compares them to the large datasets to find related phases, and then transforms the data into valuable insights.
What’s Next?
In the next sections, we will focus on how these AI technologies can be applied to the management consulting industry and what the implications are for the consulting world.
Further Readings:
Garbade, Michael J. “Clearing the Confusion: AI vs Machine Learning vs Deep Learning Differences.” Medium, September 14, 2018. [https://towardsdatascience.com/clearing-the-confusion-ai-vs-machine-learning-vs-deep-learning-differences-fce69b21d5eb].
Gesing, B., Peterson, S.J., and Mchelsen, D. “Artificial Intelligence in Logistics.” DHL Customer Solutions &Innovation (2018). [http://www.globalhha.com/doclib/data/upload/doc_con/5e50c53c5bf67.pdf].
Jajal, Tannya D. “Distinguishing between Narrow AI, General AI and Super AI.” Medium, May 21, 2018. [https://medium.com/@tjajal/distinguishing-between-narrow-ai-general-ai-and-super-ai-a4bc44172e22].
Marr, Bernard. “The Key Definitions Of Artificial Intelligence (AI) That Explain Its Importance.” Forbes, February 14, 2018. [https://www.forbes.com/sites/bernardmarr/2018/02/14/the-key-definitions-of-artificial-intelligence-ai-that-explain-its-importance/#570700e24f5d].
MC.AI, “Deep Introduction to Artificial Intelligence”, September 28, 2018.
Wei, Bobby Lee Chun. “RPA & AI (UiPath and Machine learning).” Medium, December 5, 2019. [https://medium.com/swlh/rpa-ai-uipath-and-machine-learning-8953d9a7c721].
