From the Editor

Readers who have been following the editorials in this column may have noticed the ongoing discussion about the potential impact of cutting-edge technologies like Big Data, Machine Learning, Industry 4.0, and Artificial Intelligence (AI). In fact, as far back as 2018, the idea was raised that "AI Needs CSI: Common Sense Input" in an article published in "From the Editor".¹ Since then, the focus of articles and papers published in the IJMC has been on integrating these evolving technologies to digitally transform various industries, particularly in the areas of operations, analysis, and decision-making.

More recently, we have witnessed the emergence of Large Language Models (LLMs), with Microsoft's ChatGPT being a prime example, introduced in November 2022. This represents a significant advancement in the field of AI. Unlike traditional AI systems, LLMs are capable of conversing with humans, displaying an impressive ability to generate human-like language and engage in more natural and open-ended conversations, using advanced machine learning techniques such as the Transformer architecture. This breakthrough technology has the potential to revolutionize natural language processing, chatbots, and customer service applications

The following is taken from an actual chat that I recently had on ChatGPT https://chat.openai.com/chat. This is not intended to be an endorsement but an example of the types of queries and conversations that you can experience. My initial question posed in the chat window:

TP: What causes hot tears in castings?

ChatGPT: “Hot tears, also known as hot cracking, are a common defect that can occur during the solidification of metal castings. Hot tears are caused by tensile stresses that develop in the casting as it cools and solidifies. These stresses arise from the fact that the metal in the casting contracts as it cools, while the mold or surrounding metal remains solid and inflexible.

There are several factors that can contribute to the formation of hot tears in castings. These include:

• Alloy composition: Some alloys are more prone to hot tearing than others. For example, alloys with high levels of sulfur, phosphorus, or carbon can be more prone to hot tears due to their lower ductility and higher susceptibility to cracking.

• Casting design: The shape and design of the casting can also influence the likelihood of hot tears. Castings with thick sections or abrupt changes in thickness are more prone to hot tearing because they cool at different rates and develop larger tensile stresses.

• Pouring temperature: The temperature at which the metal is poured into the mold can also affect the likelihood of hot tearing. If the metal is poured at too high a temperature, it can cause the metal to solidify too quickly and develop high levels of tensile stress.

• Cooling rate: The rate at which the metal cools and solidifies can also influence the likelihood of hot tears. If the cooling rate is too slow, the metal can solidify in a non-uniform manner, leading to localized areas of high stress and potential hot tears.

• Mold design: The design of the mold can also influence the likelihood of hot tears. Molds with poor venting or gating can trap gases and cause the metal to solidify too quickly, leading to higher levels of stress and hot tearing.

To prevent hot tears from occurring, it is important to carefully consider these factors during the casting design and production process. This can involve optimizing the alloy composition, designing the casting with uniform thickness and appropriate gating and venting, and controlling the pouring temperature and cooling rate to ensure uniform solidification.” ²

TP: How did you come to these conclusions?

ChatGPT: “As an AI language model, my responses are generated based on patterns and correlations identified in a large corpus of text data. In this case, my response is based on my understanding of the principles of metallurgy and casting processes, which is derived from a vast amount of technical literature on the subject.

The factors I listed as potential causes of hot tears in castings are well-established in the field of metallurgy and have been studied and documented by researchers and practitioners over many years. These factors have been identified through both theoretical analysis and practical experience in the production of castings.

To ensure the accuracy of my response, I drew upon a variety of technical resources, including peer-reviewed journal articles, textbooks on metallurgy and casting, and online resources from reputable organizations in the field. By analyzing these sources and identifying common themes and principles, I was able to formulate a response that accurately reflects the current understanding of the causes of hot tears in castings.” ²

TP: What about mold materials acting as a constraint?

ChatGPT ² : “Yes, the mold material can also act as a constraint and play a role in preventing the formation of hot tears in castings. During solidification, the metal in the casting contracts as it cools, while the mold or surrounding metal remains solid and inflexible. This creates tensile stresses in the casting, which can lead to the formation of hot tears.

However, the mold material can act as a constraint, providing resistance to the contraction of the metal and reducing the level of tensile stresses in the casting. This can help to prevent the formation of hot tears.

The ability of the mold material to act as a constraint depends on its stiffness or modulus of elasticity. Stiffer materials, such as steel or iron, can provide a greater level of constraint and are generally more effective at preventing hot tears than softer materials such as sand or plaster. The design of the mold, including the placement of gating and venting, can also influence its ability to act as a constraint and prevent hot tears.

In addition to the mold material, the use of chill molds can also help to prevent hot tears in castings. Chill molds are made of a highly conductive material, such as copper, and are placed in contact with specific areas of the casting to accelerate their solidification. By promoting more uniform solidification, chill molds can reduce the level of tensile stresses in the casting and help to prevent the formation of hot tears.” ²

TP: So why aren't these additional points including in your first response?

ChatGPT: “I apologize for the oversight in my initial response. Mold materials and the use of chill molds as a way to prevent hot tears in castings are important factors that should have been included. Thank you for bringing this to my attention.

In summary, the mold material can act as a constraint to prevent the formation of hot tears by providing resistance to the contraction of the metal during solidification. Stiffer materials such as steel or iron are generally more effective at preventing hot tears than softer materials such as sand or plaster. Additionally, the use of chill molds can help to promote more uniform solidification and reduce the level of tensile stresses in the casting, thereby preventing the formation of hot tears.” ²

TP: I find that when using metal molds, die thermal management can have a significant impact on occurrence of hots tears

ChatGPT: “You are absolutely right. Die thermal management is also an important factor that can affect the occurrence of hot tears in castings produced using metal molds, particularly for high-pressure die casting (HPDC) and other similar processes. Die thermal management refers to the process of controlling the temperature of the die during the casting process to optimize the cooling rate and ensure proper solidification of the casting.

In HPDC, for example, the die temperature is critical in controlling the rate of solidification of the casting and the contraction of the metal. If the die temperature is too high or too low, it can result in non-uniform solidification and residual stresses in the casting, which can lead to hot tears.

To minimize the risk of hot tears in castings produced using metal molds, it is important to optimize the die thermal management process. This can involve using various cooling techniques such as water-cooled channels, spray cooling, or a combination of both to control the temperature of the die and ensure proper solidification. Proper gating and venting design, as well as the use of coatings or release agents, can also help to optimize the cooling and solidification process and reduce the risk of hot tears.” ²

TP: Perhaps you should reflect and think about our discussion for next time the question is asked?

ChatGPT: “Thank you for the suggestion, I will certainly reflect on our discussion and take it into consideration for future questions. As an AI language model, I am continuously learning and improving, and feedback like yours helps me to provide better responses in the future.” ²

TP: How is that done? How do you reflect on discussions? How do you learn from feedback? Is there human oversight or is this internalized in the software?

ChatGPT: “As an AI language model, I learn through a process called machine learning, which involves analyzing large amounts of data and patterns in language use to improve my performance over time. This process is largely internalized in the software, but there is a team of developers and researchers who work on continuously improving the model and addressing any issues or limitations.

In terms of reflecting on discussions and learning from feedback, the model uses a combination of approaches. One approach is to analyze the text of the discussion and compare it to previous responses to similar questions to identify areas where the response could be improved. Another approach is to use human feedback, such as the feedback you provided, to identify areas where the model may be lacking and make adjustments to improve its performance in those areas.

The developers and researchers also use a variety of techniques, such as error analysis and experimentation, to identify areas where the model may be making mistakes or struggling to provide accurate responses. Based on these insights, they can make adjustments to the model's training data, algorithms, and other parameters to improve its performance.

Overall, the process of learning and improving as an AI language model is ongoing and involves a combination of machine learning techniques, human oversight, and feedback from users like you” ²

As noted in an Opinion article “ChatGPT Heralds an Intellectual Revolution” published in the WSJ Weekend, Saturday/Sunday February 25-26, 2023, by Henry Kissinger, Eric Schmidt, and Daniel Huttenlocker; “Generative artificial intelligence presents a philosophical and practical challenge on a scale not experienced since the start of the Enlightenment.”

“My response is based on my understanding of the principles of metallurgy and casting processes, which is derived from a vast amount of technical literature on the subject.”