Useful Prompts for ChatGPT and Thoughts About What Is Next in PLM AI

Useful Prompts for ChatGPT and Thoughts About What Is Next in PLM AI

ChatGPT is generating a lot of buzz in different industries and I’m sure you’re asking yourself how ChatGPT can help you today to optimize your product lifecycle management (PLM) efforts. In this blog, I will share some of my thoughts and give you examples of how you can use ChatGPT today. For some of my earlier thoughts about AI, LLMs, and ChatGPT, check out my article – PLM, ChatGPT, and Large Language Models.

Welcome to Prompt Engineering

Although ChatGPT is still very new, it created a revolution in the way we can think about generating language output, which can be extremely useful for what you do. To use ChatGPT, you need to know how to ask ChatGPT about what you need… which means natural language is the best programming tool for ChatGPT today.

This brings immediately the question of what is the best way you give instructions to ChatGPT about how to perform specific tasks to bring the best results. Prompt engineering challenges is a process of developing of prompts to give you the best results.

The goal is to create effective prompts that can help an NLP model generate accurate and relevant responses. This involves understanding the nature of the task the model is meant to perform, as well as the characteristics of the dataset and the model architecture being used.

Effective prompt engineering can lead to significant improvements in the performance of NLP models, particularly in terms of their ability to generalize to new and unseen data. It can also help to reduce the amount of training data required to achieve good performance, which can be particularly beneficial in settings where data is scarce or expensive to acquire.

Can ChatGPT help you to become better in PLM today

It is important to understand the strengths and limitations of ChatGPT today to give it the best use. It combines your understanding of how ChatGPT works and also understanding of your product lifecycle management (PLM) tasks. Here are some tasks and activities I found useful and will recommend you use ChatGPT to create a better outcome:

  1. Summarizing and generating requirements for your product
  2. Writing descriptions of functions
  3. Finding guidance for typical processes
  4. Finding guidance for typical documents
  5. Finding guidance for typical product structures

It is important to understand the limitations of ChatGPT in order to avoid situations when ChatGPT can misguide you or provide you with the wrong information. Keep in mind, ChatGPT “knowledge” stops somewhere around September 2021. While ChatGPT can generate text very efficiently, it will not be able to perform tasks such as design, managing processes, and analyzing specific data sets. Here are typical tasks where ChatGPT is unlikely to help you or can misguide you.

  1. Create 3D design (unless you’re writing a code for CAD)
  2. Create a detailed product specification
  3. Calculate product cost
  4. Find a specific supplier for components
  5. Generate graphic output

Keep in mind that AI is actively developing these days and both established vendors and startups in this field are looking at how to get more out of the use of large language models and other AI techniques to help manufacturing companies in the development process.

10 Useful Prompts To Experiment and Next Steps

To find the best use of ChatGPT today, your first task is to think about the language representation of your tasks. If anything you want to do can be translated into text/language output, then you can find ChatGPT useful for your tasks today.

Here are some examples of useful prompts:

  1. Write [language name] code to create a [geometrical entity] for [CAD system]
  2. Create a BOM for [name your product]
  3. Create an ECO approval table for [name people and roles]
  4. Find typical suppliers of [name component] for [name product] in [country]
  5. create a traceability matrix for [product name] design and requirements

I found some of the usages of these prompts can give you a very specific outcome that can be practically useful to speed up your process. When a programming language is getting involved (eg. AutoLISP or Onshape Feature Script), your outcome can be really good. Also, when you need to create a documented outcome (eg. BOM), you can be also successful. But when a specific task with precise logic needs to be performed, ChatGPT sucks and won’t help you much.

As a next step, if you think about how ChatGPT can help you, you need to figure out how your task can be translated into a specific text (language) input and output. This is where you will get the most existing strength of ChatGPT today.

What is my conclusion?

It is very exciting to see what AI tools can achieve, but it is still a very early time to think about PLM tools that can be powered by AI tools. I’m continuing to research ChatGPT, AI, and other knowledge management tools. In my next Beyond PLM blogs, you will find some ideas about where AI can take us in the future for PLM consulting, product development, and manufacturing.

While tools like ChatGPT give us powerful language generation capabilities, the breakthrough will be when we will be able to contextualize data for specific company data. Check what I do with OpenBOM AI copilot research. The data is one of the most important IPs of every manufacturing company. How to keep the data private and use AI tools for decision-making to help develop products faster, better, and cheaper is the next goal for all PLM AI research.

Just my thoughts…

Best, Oleg

Disclaimer: I’m co-founder and CEO of OpenBOM developing a digital-thread platform with cloud-native PDM & PLM capabilities to manage product data lifecycle and connect manufacturers, construction companies, and their supply chain networks. My opinion can be unintentionally biased.


Share This Post