I’ve refrained from commenting much on ChatGPT except for a few LinkedIn posts. For the most part, I think it’s an amazing technological advancement for large language models (LLM). That’s not to say it doesn’t come with controversy and a few ruffled feathers. So, what’s the deal with ChatGPT anyways? Is it an online writing killer?

The short answer is no. The longer answer is that ChatGPT will be another tool in your writing and modeling toolkits.

ChatGPT intelligent-assisted programming

One of the most interesting applications of ChatGPT and LLMs is the intelligent assisted programming ability. You give it a prompt for a programming task in a specific language and it can spit out the code. This is amazing and a huge time saver.

This feature, in my opinion, will have the highest impact on companies in the technology field. Programmers, coders, and machine learning modelers will save so much time prototyping code and getting near production-ready code.

Of course, the caveat here is that this code should be reviewed, tweaked, and optimized. It’s not perfect, but if it can get you 80% of the way there then it’s worth it.

ChatGPT assisted writing

Another impressive feature is the ability to write content from a single prompt. It can write a first draft of an article or essay. Of course, this ability strikes fear in teachers and professors everywhere. Students will “cheat” by using ChatGPT to write their homework; if my children are a gauge, it’s already all over our local high school.

Of course, what the cheating students don’t realize is that it’s not perfect and right now everyone can tell that you used it. ChatGPT is not a replacement for your own voice and style. Until you train a custom model on all your written work, it won’t be able to mimic your writing style.

The strength and weaknesses of prompt engineering

When ChatGPT was first released by OpenAI, every one of us tried to break it by prompt engineering. Some people were quite successful in breaking it by having it generate biased content. Of course, this is a flaw or error in the system that will be fixed.

That said, prompt engineering will be key in writing good content or code. How you “ask” the LLM will become important to get the right output.

ChatGPT is only as good as the data it was trained on, and let’s face it there’s a lot of biased crap out there that it probably picked up on. This is an ongoing danger as these LLMs get bigger and consume more training data. There’s a lot of crap out there on the Internet.

Still, we can remain vigilant for bias by applying critical reasoning to the results. While people fear ChatGPT will replace them or make their jobs redundant, I believe it will open up new opportunities for everyone.

We’ll need more critical thinkers and reviewers of what it generates and content creators can set themselves apart by developing their voice and style.

Overall, I believe these LLMs are a good thing. They’ll help us unlock troves of information and knowledge that’s been hidden for years for our benefit.