How does Artificial Intelligence bring tangible benefits in processing speed, accuracy and consistency? Why do many professionals now rely on it, including medical specialists who use AI to diagnose and make decisions about treatment?
To understand this, we need to know how Artificial Intelligence (AI) is being implemented.
The first phase of onboarding artificial intelligence is rather like the process of training an Assistant. You teach the new employee a few fundamental rules and hand over some basic but time-consuming tasks you normally do, so that you can focus on more important aspects of the job. The trainee learns by watching you perform tasks and by asking questions.
One common task for AI assistants is sorting data. An example of this is the ones Amazon and Netflix are using to help customers filter thousands of products and find the ones most relevant to them.
This kind of data sorting is used in more and more business decisions today. Consider the context of portfolio managers choosing which stocks to invest in. The information available is far more than a human can feasibly process. In addition, new information comes out all the time adding to historical data. The software can make the task more manageable by immediately filtering stocks to meet predefined investment criteria.
Anyone who uses Google might have noticed the prompts that appear as search phrase is typed in. Predictive text on smartphones offers a similar kind of user model related to what is sometimes called Judgmental bootstrapping. AI would use it to identify the choice an employee is most likely to make, given that employee’s past choices and would suggest the choice as a starting point when the employee is faced with multiple decisions.
An airline catering manager can predict what the airline’s passengers would order by analysing their past choices by using an AI. This auto-completion can be used to customize in-flight services for every flight using all relevant historical data, including food and drink consumption on the route and even the past purchasing behaviour of the passengers on the manifest for that flight.
The second phase is to have an AI system set for getting real-time feedback. AI can be trained to accurately forecast what a decision would be in a given situation. If a user makes a choice that is inconsistent with their history, the system can flag as discrepancy. This is helpful during a high volume of decision-making, when human employees may be tired or distracted. Research shows that humans have limited and imperfect reasoning capabilities, especially when it comes to statistical and probabilistic problems which are abundant in business.
AI systems were initially set up to usurp the autonomy in many companies which failed if deviated from the defined step and will report as fraudulent tranction. Here the AI’s choice is opaque and employees unable to question that choice even when mistakes have been made. That’s why it is used like a dialogue in which the algorithm provides nudges* according to the data it has, while the human teaches the AI by explaining why they override a particular nudge*. The Nudge in AI means, the use of user-interface design elements to guide people’s behaviour in digital choice environments. eg. Netflix in web series use the choice with the next episode.
This improves the AI’s usefulness and preserves the autonomy of the human decision-maker. So, a set of rules were designed to interact with AI when developed by the companies to ensure organisational consistency in norms and practice. These rules might specify the level of predictive accuracy required to show a nudge* or a reason whether to follow AI’s instruction or refer it to a superior rather than accept or reject it. So, monitoring has been an important phase.
The employee performance feedback usually comes from their superiors during the annual performance appraisal and not at a time or format when the employee himself wishes to know. The only way to discover strengths and opportunities for improvement is through a careful analysis of key decisions and actions. This requires documentation of expected and achieved targets. AI can solve this problem. The AI can generate feedback for employees, enabling them to look at their own performance and reflect on variations and errors. A monthly summary analysing data drawn from their past behaviour will help them better understand their decision patterns and practice.
As a good mentor learns from the insights of the people being mentored, a machine learning ‘Coachbot’ learns from the decisions of a human. This is the third phase as a Coach. In this relationship, a human can disagree with ‘Coachbot’ and create new data that will change the AI’s implicit model. If a portfolio manager does not trade highlighted stock because of recent company events, they can provide an explanation to the system. With this feedback, the data that the system continually captures can be analysed to provide more useful insights.
The last and final phase of AI as a Teammate with the concept of the extended mind suggests that cognitive processing and associated mental acts such as belief and intention are not necessarily limited to the brain. External tools and instruments can under the right conditions play a role in cognitive processing and create what is called coupled system. This system improves through its interaction with individual users, analysing and even modelling experts by drawing on data about their past decisions and behaviours, a community of experts –Humans and Machines will naturally emerge in organisations that have fully integrated AI chatbots. For e.g. A purchase manager, with one click, at the moment of a decision could see what price someone else would give and could benefit from a customised collective of experts.
Any organisation can successfully implement AI if they train the AI system to give real-time feedback and introduce the AI as a coach and finally as a teammate. This will help both humans and AI in continuous improvement.
Deputy Chief of Human Resources