The Real Cost to Implementing Artificial Intelligence

There are many perspectives on how Artificial Intelligence is changing the face of numerous businesses. Due to these different outlooks, it is hard to find the actual impact it has on the market. Ajay Agarwal, a professor at Rotman School of Management, mentioned that AI serves an economic purpose that is potentially transformative – artificial…

By

min read

Artificial intelligence robot looking at data analytics

The Real Cost to Implementing Artificial Intelligence

There are many perspectives on how Artificial Intelligence is changing the face of numerous businesses. Due to these different outlooks, it is hard to find the actual impact it has on the market. Ajay Agarwal, a professor at Rotman School of Management, mentioned that AI serves an economic purpose that is potentially transformative – artificial intelligence helps to reduce the cost of prediction.

Ripple Effects of Falling Costs

When you talk about the cost of artificial intelligence, the thing you are implying here is how much cost does it reduce. An economist or businessman will always look at how artificial intelligence helps to control costs. As mentioned earlier, artificial intelligence reduces the cost of prediction, which is the first thing that any business does. Let us look at semiconductors as an example to understand the changes that a business undergoes when there is a drop in the cost of input because of technology. If you have looked at the history of semiconductors, you will realize that they reduced the cost of using arithmetic. When semiconductors were introduced, the following changes occurred:

  1. People began to use arithmetic in applications that already used some form of arithmetic as an input. Most military and government applications used arithmetic for different applications. It was only when people began to use arithmetic in calculations that businesses began to focus on calculations like demand forecasting.
  2. People began to use arithmetic to solve problems that were not arithmetic problems in the first place. For instance, people began to use arithmetic to solve for the development of a photographic image using film-based photography or chemistry. When arithmetic became cheaper, people began to use it to design digital cameras.
  3. When the value of arithmetic fell, the cost of other things changed. The cost of every complement of arithmetic went up while the value of its substitutes came down. The complements in the case of photography were the hardware and software used to build digital cameras. The value of these went up when people began to use more of it when constructing a camera, while the value of the substitutes, like a film-based camera, decreased because people began to use less of those substitutes.

Expanding the Powers of Prediction

Since most companies now use AI in their systems, they reduce the cost of prediction. This means that they will use AI for some prediction problems like inventory management since it is easier and faster to predict the outcome. They can also use prediction to solve specific problems that were never considered prediction problems in the past.

For instance, nobody believed that autonomous driving is a prediction problem. Engineers developed a vehicle that could move around in an environment like a warehouse or factory depending on the instructions they gave the machine. They can instruct the machine to stop when a person walks past the machine or move to another shelf if the first shelf is empty, and so on. It, however, was difficult to put these vehicles on the street because there were too many conditions that the engineer needed to input.

In today’s world, engineers believe that autonomous driving is a prediction problem. The only question that the vehicle needs to answer is “What would a human do in this situation?”. The vehicle can answer this question using AI. We give the machine cameras, light detection and ranging (LIDAR), and radar since AI does not have eyes. The AI will look at the input and see how a human being would react to a similar situation before it performs the action.

The vehicle will make many mistakes in the beginning, but it will learn from those mistakes. The AI model will update itself every time it makes an incorrect prediction. The AI will become so good at predicting that it will no longer need a human being to help it perform the action.

The Growing Importance of Data, Judgement, and Action

The price of prediction, like in the case of arithmetic, will follow the simple principles of economics. When there is a drop in the price of prediction, the value of the complements will go up while the value of its substitutes will come down. Human beings always make predictions, whether it is for business or their life. That being said, we are noisy thinkers which means that we make very poor predictions. As a result of this, the value of human prediction will decrease when the quality of artificial intelligence improves.

That being said, the value of every compliment for prediction will increase. One such complement is data since it is an essential element for prediction. Therefore, if the company wants to reduce the cost of prediction, it must improve the quality of data. This increases the amount it should invest in data. There are many compliments to the prediction that are not discussed in detail, and one such complement is human judgment.

People always believe that human decision-making follows only one step. That is due to the fact that nobody has ever spent the time to unbundle the underlying actions of decision-making. Now that we want machines to make decisions for us, we are trying to determine the underlying actions. Since machines are now being used to make predictions, the role of judgment in making a decision is becoming more evident. Since AI does not look at the judgment, the value of human judgment will go up when the value of human prediction decreases. AI will pass the prediction on to human beings who can make a decision based on their judgment.

Since most businesses make predictions about how their products will run or predict their return on investment, they can reduce the cost by implementing AI. Most of these calculations are performed by human beings, and the business will need to compensate every employee working on those calculations. Since AI will decrease the cost of prediction, every business should look at implementing different features of AI to increase profits. This will help them reinvest in the business and improve their products and services.

What other “costs” have you experienced in your journey?

Leave a Reply

Your email address will not be published. Required fields are marked *