The Power of Prediction: Exploring the Role of AI in Decision Making
Say goodbye to gut decisions and hello to data-driven decision-making.
In today’s fast-paced world, the ability to make accurate predictions and informed decisions are more valuable than ever. Artificial intelligence (AI) has the potential to revolutionize this process by using algorithms to predict future outcomes and guide decision-making. You might be thinking, “That’s great, but what exactly does AI bring to the table here?”
The answer can be said in one word: Prediction.
Why does artificial intelligence not give us Intelligence
It might be shocking to think that technology so powerful and transformative has the above limitation. Today, we use AI in a multitude of ways, but with one common goal: to make decisions. Let me explain in the form of an everyday example.
You just woke up and as you’re making a cup of coffee you wonder: “Will I need to bring an umbrella to school today?” Luckily, you have Amazon’s Alexa to help answer that.
Alexa, what’s the weather going to be like today?
“Today’s forecast is cloudy with a 30% chance of rain in the afternoon.”
There is no doubt Alexa’s AI technology is powerful — it’s able to take in vocal input from a person using speech recognition, process the input using Natural Language Processing (NLP), and return a comprehensive answer. However, Alexa does not have a “mind of its own”, nor will give a perfect answer every time you ask a question. Instead, it returns an answer in the form of a prediction, based on probability and previous data.
The big idea: At the current stage of technological advancement, AI has limitations in its capabilities, the main one being that it can be used solely to make predictions across all of its various applications.
Value from Prediction
Let’s go back to the start of the article:
“The ability to make accurate predictions and informed decisions is more valuable than ever.”
What do I mean by the word valuable in this sentence? The answer can be found through simple economics; when the cost of something falls, there is more of it. Thus, cheaper predictions mean more predictions as well as more complements to predictions. To analyze this further, we can look at past moments where the above statement is true.
- The usage of light: In the 1800s, it would have cost 400x as much to consume the same amount of light as we do today. Over time, the significant price drop in light changed the world into what it is today, a fundamental input of society, all due to technological change.
- Arithmetic and computers: When the cost of creating arithmetic programs became cheap in the 1950s, there was a sudden burst of new arithmetic implementations for computers that were not possible before. This was predicted by Ada Lovelace 100 years back.
The big idea: Today, it is evident that when an essential input like prediction becomes cheap through advancements in AI, it creates tremendous value for society.
What about Ethics?
So far, we have only touched upon the positive ways AI is changing our world. It is equally important to be cautious and informed on the ethics involved in the decision-making functions of AI.
“All models are made by humans and reflect human biases.”
The main concern with AI ethics is unwanted bias being present in algorithms or systems. This can manifest in two ways:
- The input data used to train the AI model is biased. For example, consider an algorithm that is used to predict the job candidates that are most likely to be successful for a certain job. The training data is predominantly composed of men, which means the algorithm has a higher chance of being biased against women.
- The algorithm is not designed to account for certain variables — this is the “human” component present in a Machine Learning (ML) model
Upon identifying that bias in AI is due to the humans that create the models themselves, one can pose the question: “Are there ways to minimize or eliminate bias in ML models/algorithms?”
There most definitely are:
- AI systems must be transparent: it is crucial that all stakeholders understand and evaluate the decision-making process of the model. This includes being able to explain how an algorithm reached a decision and being able to identify any potential biases in the data or design of the algorithm.
- Accountability and responsibility must be considered: in the case of an incident, there must be a clear indication of where in the ML process the error occurred, how it occurred, why it occurred, and who is responsible for the design.
The big idea: It is crucial to consider the ethical implications of using AI in decision-making, and to take steps to ensure that algorithms are transparent and unbiased.
Looking into the Future
As AI continues to advance, it will play an increasingly important role in shaping the way humans make decisions. Those who understand not only the potential, but the value that AI brings will find it easier to navigate the opportunities and challenges that lie ahead. Whether you are a CEO, an AI enthusiast like myself, or simply someone interested in the future of technology, it is beneficial to have a solid understanding of the role that AI can play in decision-making.
This article was written for QMIND — Canada’s largest undergraduate community for leaders in disruptive technology.