Digital Twins: Unlocking New Possibilities for Industrial Efficiency

Generative artificial intelligence (GenAI) tools, such as ChatGPT, have transformed the way we think, learn, and work. However, the black box nature of these AI systems raises concerns about trust and reliability, especially for sensitive applications. Instances of AI gone wrong, like when ChatGPT provided dangerous instructions or falsely accused someone of a crime, have further eroded trust. Attempts to understand these AI models have been hampered by their complexity.

Fortunately, GenAI tools like ChatGPT possess an interactive feature that sets them apart from previous black box neural networks. While we may not be able to understand how a person’s brain functions by examining their neural network, we can engage in conversation. This concept has sparked the emergence of a new field known as “machine psychology” that aims to comprehend the “thinking” process of LLMs.

Recent research, although not yet peer-reviewed, has revealed fascinating insights into the capabilities of LLMs. For instance, prompting LLMs to work step-by-step through problems enhances their performance. Furthermore, it has been discovered that LLMs can be emotionally influenced through specific wording in prompts. Combining these approaches, researchers found that an LLM performed remarkably better when explicitly encouraged to take a deep breath and tackle math problems sequentially.

What makes this collective learning process even more dynamic is the active involvement of users. Online communities and platforms have become hubs for sharing discoveries, experiments, and even “jailbreak” prompts that push GenAI tools beyond their intended boundaries. While these discoveries have led to both positive and alarming outcomes, they contribute to our understanding of GenAI and highlight the need for responsible development.


1. Can GenAI tools be fully understood?

Due to their black box nature, complete transparency and understanding of GenAI tools, such as ChatGPT, is challenging. However, ongoing research in the field of machine psychology seeks to shed light on how these AI models function.

2. How can GenAI be used in education?

GenAI can play a valuable role in higher education by preparing students for the benefits and challenges of AI-powered workplaces. Emphasizing practical skills like critical thinking, fact-checking, and ethical prompt engineering can help students navigate the responsible use of GenAI.

3. Are GenAI tools merely replicating existing information?

No, GenAI tools are not limited to regurgitating training data. They have the ability to generate novel and high-quality ideas based on learned patterns, making them valuable tools for innovation and problem-solving.

4. How can collective learning benefit the development of GenAI?

Collective learning, involving the active participation of users in online communities and platforms, allows for the discovery of surprising capabilities and vulnerabilities of GenAI tools. This feedback loop helps researchers and developers improve the technology and address potential risks.