An Economic Perspective on Artificial Intelligence

A Literature Review

With all the AI hype, it's tough to tell what products will be genuinely transformational beyond the power of Chat-4 and soon to be released BARD. 

In the last couple of weeks, I read a lot of academic papers on Artificial Intelligence. Some related to Behaviour Science and my thesis, and some...well, very random. 

I found a few exciting and visionaries, and I will share my high-level literature reviews with all of you in the upcoming week. 

The first article I found interesting it's called An Economic Perspective on Artificial Intelligence, published by Ajay Agrawal, Joshua Gans, Avi Goldfarb.

The article delves into the economic perspective of artificial intelligence (AI) and its potential implications for organisations and individuals. 

As prediction technology has become more common, Machine Learning is expected to replace human prediction tasks, including medical diagnosis, recruitment, and inventory management. While AI technology has made significant progress through Machine Learning models, it is vital to note that prediction alone has no value. 

Instead, complements such as data, action, and judgment are essential to make predictions valuable. So the models are as good as the data (!) - no surprises there. 

The article goes on to explain that better prediction technology could transform the way organisations operate, hire and grow.

For instance, Amazon currently uses predictions about the products customers want to buy, and these predictions are tools that help the company's customers complete tasks more efficiently. And influence consumer behaviour but a more detailed description in upcoming articles. 

As a reference, McKenzie estimates that 35% of Amazon sales come from those predictions (!!!). (Source: Quoted directly from the paper)

However, if the predictions become more accurate, Amazon could ship the item directly to the customer before it is even ordered, which could revolutionise how the company runs its logistics. 

Similarly, the article provides evidence that better prediction technology could significantly reduce recruitment costs and enable better-targeted advertising in the short-term and personalised recruiting messages in the long term. The paper is missing some references and data on the ethical part of using prediction technology in recruitment, but that's not one of the leading research areas for the authors. 

The research also highlight the importance of understanding the complements of prediction technology. Accessing, developing, or possessing these complements, such as judgment or the ability to understand payoffs to improve decision-making, will be critical in the future. 

The article concludes that competition in AI will be shaped by access to complements, strategy-making capabilities, and organisational flexibility. Therefore, it is vital to understand the economic implications of AI and its accessories to maximise its value. .

Overall, the article provides a comprehensive analysis of the economic implications of AI and highlights the importance of understanding its complements to achieve maximum value. And by economic implications I don’t mean JUST cost-management and replacing humans with machines

The authors don't touch on the ethical aspects of predictions and using behavioral data to train the model. That being said, this area of research will continuously grow. 

These types of insights, like the one presented in the paper, can impact how organisations perceive the Economics of Artificial Intelligence. How those insights will be used is a different matter, and as I am researching some of the Ethical considerations, I will publish reviews of what I find. 

See you next week!