Cars, phones and washing machines, there’s a proven, timeless benefit to innovation. But as with Boeing Max 8, Fire Island and that Samsung phone that blew up in people’s faces, there can sometimes be a costly downside. With today’s fast pace we’re seeing a substantial rise in the detrimental effects of misapplied technological advancements. There has been a wave in the tendency to follow trends, even when those trends provide zero value. Out with the old, in with the new doesn’t work if the new isn’t adding any value to your business. Artificial intelligence is no exception.
To understand where AI can apply value to a business environment, we first need to understand what artificial intelligence really is.
From I, Robot to Blade Runner, AI has been depicted as human-like machines. This perception has leaked into society, portraying AI is an area of computer science focused on creating machines that work, look, think and react like people do. The definition may be factually correct, but if you can make a robot that toasts your bread and asks you how your day went, further than having a bit of a laugh, what’s the point?
Particularly in businesses, AI must be value focused, and the value AI can provide most, lies in its predictive powers. AI learns from past experiences (or imagined experiences) to affect and improve future behaviours in order to achieve a better outcome.
Qualities that drive value
In business, value is achieved when change delivers improved quality, reduced cost, or faster delivery. Where predictive models are concerned, value is gained when predictions are made with higher accuracy, for less overall money and at a faster pace, ensuring delivery keeps up with the market.
It’s common for businesses, pushed by regulatory changes or market shifts, to make decisions in the face of uncertainty, allowing room for errors and costly losses. Predictions are made based on the relevant information and throwing AI into the mix makes the relevant information far better and more actionable. More accurate predictions mean better outcomes, and this is where AI provides incredible value for businesses.
- Reduced Costs: Making a decision that requires intense data analysis requires experts, and here, AI can provide significant value. Assuming the output remains the same, reducing the number of experts involved in predictive data analysis reduces the overall cost of the process. By leveraging AI, predictions are cheaper, faster and more accurate. A natural consequence of this is that the number of predictions increases astronomically. While expert talent must be reassigned, in an environment that utilises AI that talent is not lost, but heightened by being utilized in more critical areas, like data collection and the analysis of critical cases in an environment with higher volumes of predictions.
- Improved Quality: AI processes data at a much faster rate than a human possibly could. This faster processing power enables a more accurate analysis of relevant information. As a result, AI provides a better quality of prediction. More importantly, AI is able to learn and improve itself, so the quality of the prediction becomes better with time and experience. The system gets better over time at identifying correlations and assigning different weights to different variables, this faster learning process gives AI a sustainable, long-term edge any business would jump for.
- Faster Predictions: The quicker you know the answer the faster you can make a decision. The faster your decision the quicker you can go to market with products or services giving your business a competitive advantage. Since AI learns as it grows, that competitive edge acts as a cycle, every win gets you further ahead of the game and as your predictions become cheaper and faster it gets harder and harder for your competitors to catch up. Customer loyalty also ensures that if client’s buy from you first because you got there first, they’ll likely buy from you again, keeping you at the forefront of your industry.
So, here are the 3 questions you’re going to ask yourself when you start to look internally for opportunities to leverage AI; are we going to save any money, is this going to improve the quality of what we present to our clients and users, and does our speed of delivery matter? If the answer is yes, then enhancing your predictive process with AI may be a good business decision.
Tools that will help you decide if implementing predictive AI will deliver improvements
Professors Ajay Agrawal, Joshua Gans and Avi Golfarb at the University of Toronto’s Rotman School of Management introduced the AI Canvas to help leaders deduce how AI could improve business decisions. Collecting information focused around each of the 7 sections will help you assess the benefits AI could bring to your business.
Prediction: Step one is centred around defining what it is you are trying to predict. For example, at Deal AI, a project run by CPQi, our goal is to predict the 30 day volatility forecast of the Dow Jones Industrial Average (DJIA).
Judgement: Any prediction comes with a degree of inaccuracy. It’s important to understand the cost of that error, and know when an expert should be introduced to analyse the output. For example, at Deal AI, the DJAI volatility is the output. The recent attacks to Saudi Arabia’s oil infrastructure created an emotional reaction to the stock market and an expert would be able to analyse the cost of that emotional reaction.
Action: What’s the action you’re planning to take? At Deal AI, we plan to make better investment decisions based on the 30 day prediction our AI suggests.
Outcome: Actions lead to outcomes. There are 6 basic outcomes that could occur based on our use of AI.
- The decision was made to buy, and 30 days later a profit was made that would not have been achieved without our AI engines.
- The same decision was made, but we made no profit and no loss.
- The same decision was made, but we lost.
- The decision was made to sell, and 30 days later a loss was prevented.
- The same decision was made, but we made no profit and no loss.
- The same decision was made, but we were prevented from gaining something that we could’ve gained had that decision not been made.
Input: What data is relevant to the predictive engine? Objective and hierarchal value needs to be assigned to each data point and those data points need to be defined for their predictive value. AI often changes the value of those data points based on its assessment of their predictive worth as a result of long-term exposure to fluctuations in beneficial market forecasts. For example, while I can’t divulge the details due to the confidential nature of our project, with regards to our DJIA AI model, a range of values with thousands of variables inform our system on a daily basis.
Training: Appropriate training is required in order to prevent false conclusions and overfitting, which requires an objective understanding of what initial data it should be given. Due to the confidential nature of our Deal AI project, I cannot comment on the initial data used.
Feedback: The outcome of our prediction is then compared against the financial value achieved from our prediction, to create feedback that will drive the evolution of your business and the AI engine itself. By analysing the difference, and exploring how that difference can be reduced, the AI system is able to improve itself and deliver more accuracy in future evaluations. I cannot comment on the feedback processes currently implemented on our DJIA prediction model at Deal AI as a result of confidentiality conflicts.
Significant value can be derived from AI, as long as due consideration is given to effectively evaluating where AI can produce significant economic value. There is a strong risk in investing in technologies that don’t deliver any economic value as a result of hype mentality. Spending millions on an “electronic nose” that detects cocaine residue using AI provides zero value if it cannot compete with the fundamentally lower cost of a sniffer dog. Better to keep the dog and spend the money on something a little more worthwhile, like predictive technology that can tell you where the financial market is going tomorrow.
Esteban De Bernardis
Source: Prediction Machines by Ajay Agrawal, Joshua Gans and Avi Golfarb
About the author: Esteban De Bernardis is a global executive with wide ranging experience in finance, strategy and M&A who has led international and dispersed teams for different organizations over several decades.
He has held several senior roles including CFO at CPQi and CFO at CSA Group, contributing to their successful global expansion into new markets and business segments. He holds a CA designation from Argentina, with a BA in business and has pursued postgraduate studies in international Business and Marketing. He also holds an MBA from Rotman (University of Toronto). Further, he is a Certified Mergers and Acquisitions Advisor (CM&AA).
Esteban has lived and work in three different continents. Due to his interest in AI he completed the Artificial Intelligence course at Rotman School of Management, UofT.