In today's digital age, technology has evolved to the point where it can help with almost every aspect of our lives. From basic tasks to complex projects, there is an AI tool for everything. ChatGPT is one such tool that can help engineers in their work. However, it is important to note that ChatGPT should not be seen as a replacement for the knowledge and expertise that engineers bring to the design and product development process.
While ChatGPT can generate responses for engineering calculations or answer questions about generic engineering knowledge, its responses should be met with a healthy dose of skepticism and fact-checking. For example, when asked about the calculation of the flow rate of a fluid through a pipe, ChatGPT may provide an answer based on the properties of the fluid and the dimensions of the pipe. However, it may not take into account the specific conditions of the pipeline, such as the pressure and temperature, which could affect the accuracy of the calculation. Therefore, it is important for process engineers in the oil and gas industry to verify the information provided by ChatGPT and incorporate their experience and expertise into the design and development process.
ChatGPT’s Limitations
1. The quality of the data it's trained on can affect its performance. While the model is trained on a vast amount of text data, this data may not be representative of real-world language use. The data may also be biased towards certain topics or demographics, leading to limitations in the model's ability to understand other contexts.
2. The complexity of language can also be a limiting factor. While ChatGPT can generate responses to a wide range of questions, it may not always understand the nuances of language use. This can result in misunderstandings or incomplete responses, particularly with idiomatic expressions or sarcasm.
3. It's also important to note that ChatGPT is a language model, not a computational one. Therefore, when asked a simple problem such as "10 + 10", it will approach the problem as a language problem rather than a mathematical one. Its response of "20" is based on the information it has been trained on and may not always be the most accurate or appropriate response.
As artificial intelligence (AI) begins to make its way into engineering, several hurdles must be overcome before we can rely on AI to design our structures and systems. Here are the main challenges we face:
Data Accessibility: To create robust AI models, we need a wealth of high-quality data. However, many engineering processes and technology platforms are poorly integrated, leading to fragmented data. Large-scale data collection and storage also raise concerns about security breaches and infringement on sensitive information.
Model Training: The complexity and diversity of data sets required for training AI models in engineering services make this process challenging. Data sets need to include information on materials, construction methods, design codes, and past project data, covering all possible scenarios. Engineers often spend considerable time manually combining data from various sources. Training an AI model effectively and consistently with this vast range of data is no small task.
Practical Implementation: Integrating AI into engineering design and construction requires identifying appropriate applications for specific needs. Choosing the right AI application will depend on the project and its objectives. Integrating AI into existing workflows can be difficult, as can ensuring compatibility with other software used in the industry.
As artificial intelligence (AI) continues to advance, its integration into various industries has become increasingly important. The engineering sector, in particular, can greatly benefit from leveraging the capabilities of AI models like ChatGPT. However, several challenges and limitations must be addressed to effectively integrate AI solutions into engineering workflows and systems. Here are five solutions that can help engineers integrate ChatGPT into their workflows and engineering systems in the oil and gas industry, focusing on calculation problems.
SOLUTION 1: Enhanced Data Integration and Pre-processing
Engineers use tools to collect and organize data into a standard format that makes sense to ChatGPT. This helps ChatGPT provide faster and more accurate solutions to problems, like a puzzle with all its pieces sorted out and ready to be solved.
Sample Prompt: “Calculate the total volume of crude oil that can be extracted from an oil well over a period of 5 years, given the well's daily production rate of 500 barrels per day and assuming a constant extraction rate. Use the well's reservoir size of 50,000 barrels of oil equivalent (BOE), oil viscosity of 30 centipoise (cP), and extraction efficiency of 50% to improve accuracy of the calculation..”
SOLUTION 2: Customized Training for Domain-specific Calculations
To make ChatGPT smarter at solving tricky engineering problems, people should teach it how to do math for specific jobs, like calculating how hot things get in pipes. This helps ChatGPT get better at solving hard problems in things like oil and gas.
Sample Prompt: “What is the overall heat transfer coefficient for a heat exchanger with a heating surface area of 50 m², that is used to heat crude oil from 20°C to 60°C, using hot water that enters the exchanger at 90°C and exits at 70°C? The crude oil has a density of 800 kg/m³ and a specific heat of 2.5 kJ/kg°C, while the water has a density of 1000 kg/m³ and a specific heat of 4.2 kJ/kg°C. The fouling factor for the exchanger is 0.0004 m²°C/W.”
SOLUTION 3: Seamless Integration with Existing Engineering Software
Engineers need to connect ChatGPT with other engineering tools so they can share information and work together easily. This helps ChatGPT solve problems like figuring out how liquids flow through pipes. And, it makes it easier for engineers to start using ChatGPT alongside their current software.
Sample Prompt: “Determine the flow rate of natural gas through a 500 meter long pipeline with an inner diameter of 8 cm, operating at a pressure of 500 kPa and a temperature of 20°C. The gas has a density of 0.8 kg/m³ and a viscosity of 0.02 Pa·s. Export the results in a csv format that can be easily used in Excel calculations.”
The human element remains one of the biggest challenges. AI systems must be designed for easy interaction and suited to their intended application. The success of ChatGPT’s natural language processing approach and its widespread adoption demonstrate the potential of AI when average users can easily communicate with it.
With a $10 billion investment by Microsoft, the road ahead for ChatGPT has been significantly widened, the speed limit removed and paved with gold. While it may have a long way to go before it can be considered a tool in the engineers' toolbox alongside handbooks, calculators, CAE software, and Google, the investment brings with it vast potential for growth and development. This funding will enable ChatGPT to continue to evolve and enhance its capabilities as a language model, ultimately bringing new and exciting possibilities for engineers and other users in the future. The investment serves as a strong vote of confidence in ChatGPT's potential and highlights the significant role it could play in the years to come.
SOURCES:
Tara, R. (2023, January 31). A report on ChatGPT’s usefulness for design engineers. Engineering.com
https://www.engineering.com/story/chatgpt-has-all-the-answers-but-not-always-the-right-ones
Siebenale, K.B. (2023, March 28). Will ChatGPT AI Revolutionize Engineering and Product Development? Here’s What to Know. Ptc.com
https://www.ptc.com/en/blogs/cad/will-chatgpt-ai-revolutionize-engineering-and-product-development
Rawling, T. (2023, April 4). ChatGPT Challenges in Engineering Data Calculation. Calctree.com.
https://www.calctree.com/blog/chatgpt-challenges-in-engineering-data-calculation#section-3
Comments