It was in December last year when a good friend of mine introduced me to ChatGPT. Like a well-meaning matchmaker, he explained how the chatbot could save me time spent on research work. He forwarded me over WhatsApp an example of a prompt he fed ChatGPT: “Create a world of about 25 unique species fighting for one energy source for 100 million years. The story should include love, war, and redemption.”
What ChatGPT came up with read like an elementary sci-fi novel by a novice. The short story included monotonous sections of an Introduction, The War, The Main Characters, The Final Battle, The Aftermath, and even a Conclusion.
The storyline certainly wasn’t gripping, and the narration was rough around the edges. The main characters didn’t even have names, for crying out loud. But the speed at how quickly the bot composed the story got me very curious.
“I’ll be out of a job,” I muttered, horrified.
“No, you won’t. The bot still needs humans to ask it questions. Don’t be scared, embrace it. Our role is to leverage AI and integrate it into our work. And not everyone knows how to ask the right questions,” he quipped.
I wasn’t quite convinced. I believe that machine learning grows at an exponential speed, and it wouldn’t be long before this chatbot starts churning out bestsellers (and quite possibly, snagging writing jobs from me). As a matter of fact, some individuals have already used ChatGPT to create e-books and offer them for sale on Amazon.
Over the next weeks, I interrogated ChatGPT (which is free to use for now) relentlessly and tested it with all sorts of questions that I could think of.
“How does oxygen stratification affect shrimp farming?”
“What are the top 10 technologies that will impact the arms and ammunition industry in the future?”
“Write me a 100-word story about how crypto will rule the world and how AI will super-power all industries.”
When ChatGPT responded with a convincing story in less than 4 seconds, I cautiously typed my next question with bated breath and almost timidly hit the ‘Enter’ key.
“Will AI rule the world and take over jobs performed by humans?”
Plausible. Job displacement. These keywords in the first paragraph were enough to cause me further anxiety.
“Who ARE you?” I typed.
While it is not a search engine, ChatGPT seems to have the ability to churn out straightforward responses to all kinds of questions, which can be very helpful for a writer. For example, ChatGPT goes straight to the point in answering how oxygen stratification affects shrimp farming. On Google, I had to sieve through several web results and peruse numerous articles for an answer. Don’t get me wrong, Google has dramatically changed the way the world accesses information. But it appears that while serving up tons of content in response to web searches, Google has left the sleuth work of finding the right answer to us, the users. ChatGPT, on the other hand, provides tailored answers that are forthright and to the point. Still, that doesn’t quite give ChatGPT a firm upper hand.
Preply (an online language tutoring marketplace) analyzed the performance on both ChatGPT and Google to determine which was smarter. Both platforms displayed different proficiency levels depending on the type of questions served, although ChatGPT scored higher when it came to mature and wiser-sounding responses. While its agile and straightforward replies may sound more thoughtful, some ChatGPT’s answers lack current relevancy compared to Google. For instance, if you would like to find out the best shows streaming on Netflix at the moment, ChatGPT is unable to give you the most up-to-date answer - for now.
It is worth noting that this was before ChatGPT has access to the Internet. I’m pretty sure that when it does connect to the Internet, ChatGPT has a very good chance of dethroning Google as the go-to web search platform.
For now, I am inclined not to take ChatGPT’s every word for it. In fact, I almost always perform the same question search on Google for authenticity. You see, ChatGPT does not have direct access to the Internet to retrieve data in real-time. Although it is trained on large amounts of publicly available text data from the Internet and other written materials, it is unable to access events or information from the Internet after September 2021 - its knowledge cutoff date. However convincing ChatGPT may sound, it does generate wrong answers at times. As it is trained to respond with words it statistically predicts what should follow next, ChatGPT can often make up facts, or worse, guesses.
I asked ChatGPT if it lies. I found its answer to be refreshingly candid for a chatbot.
I wasn’t expecting a bot to be so frank and forthcoming with its flaws. It spoke with such plainness about its capabilities that one can’t help but feel empathetic should it make a mistake. ChatGPT’s ‘honesty’ if I may term it, felt like an accountability virtue that is sorely lacking even in some humans. Although its reply can also be viewed as a disclaimer, it is hard for one to fault a bot that puts an indemnity up front and not in fine print.
ChatGPT was equally honest when asked if it can draw intelligent conclusions about trends.
In my humble opinion, this response authenticates the very core of AI intelligence. Intelligence lies in the human element. While AI can draw inferences from multiple data sources, the input still needs to come from humans. In interpreting perceptive conclusions about trends, AI lacks the breadth and depth of human intelligence and personal experience. After all, it is artificial intelligence.
When it comes to writing, ChatGPT or AI, in general, can certainly generate content effortlessly and provide much assistance in terms of research, story ideas, content, and even proofread an article on a whim. However, it lacks the unique individual writing styles and subjective opinions possessed only by human writers. The perceptivity, creativity, and flair of a human writer cannot be imitated nor replaced by an AI bot. A true blue writer has a compelling story to tell, personal experiences to share, wisdom to impart, and a legacy to leave behind.
To a certain extent, yes. As consumers of content, we are exposed to a multitude of information online, relevant and irrelevant. With the entry of ChatGPT which can generate human-like text, content creation will continue to grow in leaps and bounds. And herein lies the danger: when the responses are so human-like, there’s the risk of mistaking an inaccurate piece of information as the truth. This also begs a bigger question: will humans apply vigilance and responsibility in using such AI tools?
For now, I am embracing the advancement of AI and I look forward to seeing it grow in applications and transform industries. After all, when the big tech companies want to make it better, it can only get better.
P/S: This article was written entirely by a human. And by the way, we are looking for human writers. If you are a word-wielder confident of slaying ChatGPT at Wordle, drop us a message at firstname.lastname@example.org.
It’s about rethinking traditional employment archetypes. Can we progress from an economy built on full time employment habitually enslaved by unemployment fears, to one where individuals have greater autonomy and are self motivated to do work that inspires them? And as a result, benefit the economy as a whole?
You can’t own full time employees. But you can build a winning team with talent management companies. As businesses demand more, external talents are emerging as the sure forerunners of an agile workforce. At Chance Upon, we partner businesses to get a head start over competition by creating collaborative work between companies and the right talents.