ChatGPT helps improve Harris’ comments on AI

Vice President Kamala Harris shakes hands with the future. (Fake photo by Midjourney)

I gave ChatGPT a sample of Vice President Kamala Harris’ comments about AI and worked with its GPT-4 bot to improve the message. The result, I believe, shows that AI is nothing to be afraid of — and in many cases can aid in human communication.

Everybody who wants to communicate well needs a second opinion. Harris seems overwhelmed much of the time. An AI assistant could help her reduce the stress of creating material for public appearances. I know I use AI as a proofreader and occasional collaborator. Our vice president should consider doing so, too.

Here’s how I worked with ChatGPT.

User — Look at the following text and when you’ve done so, reply “Read.”: I think the first part of this issue that should be articulated is AI is kind of a fancy thing, first of all, it’s two letters, it means artificial intelligence but ultimately what it is is it’s about machine learning. And so the machine is taught and part of the issue here is what information is going into the machine that will then determine, and we can predict then if we think about what information is going in, what then will be produced in terms of decisions and opinions that may be made through that process.

ChatGPT — Read.

User — Please rewrite to enhance clarity and eliminate confusing word choices and sentence structure. Retain first-person viewpoint.

ChatGPT — Firstly, it’s crucial to understand that AI, which stands for artificial intelligence, is essentially about machine learning. It may sound complex with its abbreviated terminology, but fundamentally, it’s a process where machines are taught certain information. The key aspect to focus on here is the type of information fed into these machines. This input significantly shapes the decisions and opinions that the machine will ultimately generate. By carefully considering what information is being introduced, we can anticipate the potential outcomes of the machine’s learning process.

User — Simply even more.

ChatGPT — AI, short for artificial intelligence, is all about teaching machines. What we need to focus on is what we’re teaching them. The information we put in influences the decisions they make. If we know what’s going in, we can guess what will come out.

User — Eliminate duplicated words.

ChatGPT — AI stands for artificial intelligence and involves instructing machines. The input determines their decisions. Knowing the input allows us to predict the output.

User — Rewrite with fewer long words.

ChatGPT — AI means teaching machines. What we teach shapes their choices. Knowing what we teach helps us guess their choices.

User — Rewrite and avoid the word “guess.”

ChatGPT — AI involves teaching machines. The lessons they get shape their choices. By understanding their lessons, we can predict their choices.

User — Rewrite and shorten.

ChatGPT — AI teaches machines. Their lessons shape choices. Understanding lessons predicts choices.

User — Rewrite and change “get” to “given.”

ChatGPT — AI instructs machines. Lessons given shape choices. Knowing lessons foresees choices.

User — Rewrite and explain who is referred to in the last sentence.

ChatGPT — AI trains machines. The lessons they’re given mold their actions. When we know these lessons, we can anticipate their actions.

User — Use another word instead of “know.”

ChatGPT — AI guides machines. The lessons they receive define their actions. By comprehending these lessons, we can predict their actions.

User— Rewrite to make less confusing.

ChatGPT — AI is a way to train machines. The lessons we give them shape what they do. If we understand these lessons, we can predict what they’ll do.

User — If you can do so without altering the meaning, change “what they’ll do” to “their output.”

ChatGPT — AI is a method to educate machines. The lessons we provide determine their actions. If we understand these lessons, we can predict their output.

User — Rewrite the following to sound better in a speech: AI is a way of educating machines. The lessons we provide determine their actions, so by understanding the lessons, we can predict the output.

ChatGPT — In the realm of AI, we essentially teach machines. The essence of their actions is formed by the instructions we supply. By fully grasping these lessons, we gain the power to foresee their output.

User — Rewrite to be more friendly and conversational.

ChatGPT — So, you know, AI is like our way of schooling machines. What they do? Well, that’s pretty much based on what we teach them. And if we really get what we’re teaching them, we can kinda guess what they’re going to spit out.

User — Rewrite this to sound less academic: In the realm of AI, we essentially teach machines. The essence of their actions is formed by the instructions we supply. By fully grasping these lessons, we gain the power to foresee their output.

ChatGPT — When it comes to AI, we’re really just teaching machines. The stuff they do? It’s all based on what we’ve told them. If we truly understand what we’re teaching, we can better predict what they’re going to do next.

User — Find a non-clumsy way of telling what AI stands for.

ChatGPT — When we talk about AI, we’re actually referring to ‘Artificial Intelligence’. It’s just a fancy way of saying we’re teaching machines to think and act on their own.

User — Rewrite to eliminate redundancies: When talking about AI, we’re referring to “Artificial Intelligence.” It’s a fancy way of saying we’re teaching machines to think and act independently. If we truly understand what we’re teaching, we can better predict what they’re going to do next.

ChatGPT — AI, or “Artificial Intelligence,” is all about educating machines to act independently. By fully grasping the lessons we impart, we can anticipate their next move.