I’m Henry Crankvis, and I’m the dean of the arduous School of Business and Economics here at Chapman University. Today, I have the great pleasure of talking to Professor Gary Smith, the Fletcher Jones professor of Economics at Pomona College, about AI and chatGPT.
Professor Smith has been teaching statistics and investments for many years and has known about the perils of data mining. Data mining is the process of looking at a bunch of data without any well-defined purpose in mind and searching for patterns. However, the problem is that any large dataset contains numerous patterns, and finding one does not necessarily prove its significance. This issue is magnified with AI, which is data mining on steroids.
AI, including chatGPT, goes through data, whether numerical, pixels, or sound waves, to define patterns and make decisions based on those patterns. However, the challenge with AI is that it is exceptionally good at finding patterns, and if it operates inside a black box, it becomes challenging to judge the validity of its recommendations.
ChatGPT, which stands for Generative Pre-training, is a subset of AI that trains on text and generates text. It has been trained on billions of words of text and can generate realistic conversations. However, it lacks an understanding of the meaning of words and often produces unreliable information. It can make up details that sound convincing but are entirely false.
While chatGPT may have some benefits, such as entertainment value and providing recommendations, its limitations and dangers outweigh its advantages. The opportunity cost of smart minds working on a text generator that produces unreliable information is significant. Additionally, chatGPT’s ability to generate disinformation at a rapid pace poses a threat to society.
From a business perspective, the benefits of chatGPT are limited. Its unreliable nature makes it unsuitable for critical decision-making. The risks of relying on chatGPT for important tasks outweigh any potential benefits.
In conclusion, chatGPT and other large language models are part of the AI universe. They share the same flaws as AI in general, particularly the inability to distinguish between fact and fiction. Until computers can differentiate between reality and nonsense, we cannot fully trust AI. It is crucial to understand the limitations and dangers of chatGPT and approach it with caution.
