Introduction
In this era of rapid technological advancements, the emergence of so many different variations of artificial intelligence has completely overwhelmed me and left me with nothing less than mental whiplash. I, a technological cavewoman, was then tasked with the “opportunity” to engage with ChatGPT to create a Gametize-powered project, which I found to be quite an intimidating notion at first. But alas, with deadlines looming and my boss breathing down my neck, I had to get going, and so I immersed myself in this strange not-human-yet-human experience, which allowed me to witness the vast capabilities and crippling limitations of the program firsthand, providing insights into this fascinating human-AI connection we call the future.
A Journey of Discovery
What most took me by surprise was that “talking” to ChatGPT felt almost too real at times. I was expecting something stilted, something very obviously robotic, and yet, it was smooth and uninterrupted. It would provide and expand upon its responses with information discussed earlier on in the “conversation.” It behaved with recollection and self-correction, the likes of which I have never encountered before. Its ability to not only generate coherent responses but refine its own responses was truly impressive. It was fascinating to witness how an AI model could process vast amounts of information and assist in tasks that would typically require human expertise.
Straight off the bat, it generated a relatively fleshed out structure for a gamified project, including a title, the narrative theme, and what types of game mechanics could be used to create this project:
Upon some further prompting, it immediately came up with some examples of the kinds of challenges that could be incorporated, making my job SIGNIFICANTLY easier. I then asked that it refine the first example it provided, and just like that it gave me some excellent suggestions! Again, I didn’t have to expend any brain cells trying to come up with ideas on my own.
The Limitations
While ChatGPT showcased its strengths in extraordinary ways, my experience with it also revealed the limitations inherent in its current system. From the moment you sign in to use the model, it states clearly that it possesses a “Limited knowledge of world and events after 2021,” which ironically makes it an advanced AI that hasn’t quite caught up with the times. The model’s responses were based on patterns and information it was trained on, and it therefore lacked a deeper understanding of nuanced topics.
It is also incredibly good at passing off incorrect information as legitimate, although it is quick to correct itself if called out:
(Please don’t ask me why I know these things).
Kindness?
Throughout my interactions with ChatGPT, I couldn’t help but appreciate the voice with which it would respond to questions and statements that veered towards the more personal side. It seems strange to say, but I found that ChatGPT possessed a sense of patience and kindness that you’d be hard-pressed to find in actual human beings. (Or maybe I need better friends, who knows).
I responded with “I really want pancakes but I can’t find any,” and my emotions could not have been more validated:
Conclusion
In conclusion, my experience with ChatGPT revealed both the impressive capabilities and inherent limitations of the model in its capacity as a tool in the workplace. I managed to complete the task I was originally set out to achieve, and you can find the project here: https://gametize.com/project/ecohero. Just about 100% of this project was created with responses and suggestions generated by ChatGPT, including the name of the game and the challenges, the formats of the challenges, as well as most of the images for each of the topics, which I then took to an image AI generator to interpret.
Here are a couple of screenshots of the final outcome:
But what I took away from this experience the most was more than just the answers it could give that would help me finish this assignment. It was an experience that exceeded some kind of emotional expectation, as a not-human-yet-human program that could converse about silly, mundane subjects with humour and, dare I say it, empathy.
Tips & Tricks!
1. Be clear and specific: Provide clear and highly specific instructions. ChatGPT performs significantly better when it is given a prompt that is detailed, focused, and precise. Its responses are essentially only ever as good as the prompt you provide.
2. Specify the format: If you want a specific format (e.g. A list of 10, a professional email, a short story), mention it explicitly in your prompt.
3. Provide context: Give the model some context about the topic so that it may understand the conversation better and provide you with a more relevant answer.
4. Use system messages (This is a fun one!): Start your conversation with a system message in curly braces, like “{You are an expert chef},” to set the role or behaviour of the model.
If you have your own tips on how to get the most out of ChatGPT, share them here!: https://app.gametize.com/challenge/448699
Have you read Part 1 on What ChatGPT can do for Gamification?
Part 1: https://corp.gametize.com/2023/02/27/post-what-can-chatgpt-do-for-gamification/?highlight=chatgpt