Behind the Hype of ChatGPT
AI-induced panic. That’s the status of numerous writers at the moment as artificial intelligence seems to be poised to replace human content creators. But is all of this trepidation justified? Perhaps. But more than likely, at least for the time being, the answer appears to be ‘No’—primarily because software like ChatGPT, Google Bard, and Hubspot’s Content Assistant are still very much in the infantile phases of development. And their current limitations, in addition to the vast amount of sources that passages are compiled from, greatly impact how the generated copy can be used.
For example, I recently received a general email from my daughter’s school stating that submitting an essay, research paper, or any formal homework assignments written using ChatGPT is essentially plagiarism. From an educational administration POV this is problematic because much of the material is directly excerpted from other pieces already published (or archived) online. Thus every user must keep in mind that the majority of the writing produced by these tools is borrowed—without consent or compensation—from the initial scribe. This is a valid concern many visual artists also expressed last fall when “magic avatar” images downloaded from AI photo apps like Fotor, Lensa, and ToonApp started trending on social media.
To further explain the pros and cons of these chatbots, I interviewed MIT grad and Fancy Comma founder Sheeva Azma:
1) In your opinion, should writers be anxious about our art form becoming obsolete?
I am not worried at all. We learn to write and speak as humans so we can communicate with our fellow humans. In the 21st century, we can use artificial intelligence to communicate better, and I am excited about it.
Innovation drives problem-solving for the greatest challenges we face, and innovation in writing is a huge part of that. Think of all the great, life-changing books you’ve been able to read because they’ve been published at scale thanks to printing technology.
2) What potential benefits of this type of technology being more accessible?
- Promote artificial intelligence within society and boost our level of understanding of AI. It’s been one of those areas in computing where only experts could understand it. Now that it’s getting to a point where non-expert humans can interact with it, that’s very exciting.
- Create a blueprint for improved chatbots that can tackle some of the limitations of ChatGPT.
- Expose the technical problems involved in developing a chatbot, such as content moderation on social media (which is a huge problem).
3) What’s the biggest misconception currently circulating about ChatGPT (and similar AI tools)?
The biggest misconception, to me, is that ChatGPT replicates human knowledge to the extent that humans can. ChatGPT is powerful, but it is limited. When you try to reduce the human language module (which develops over more than a decade) to a set of algorithms, as with ChatGPT, there are going to be a lot of places where the AI falls short. ChatGPT has been known to make up its own facts (and sources), for instance. People that use ChatGPT would benefit from understanding the technology’s deficiencies.
Azma also stresses this type of AI is still reliant upon content modifications which must be done by a person. She points out that Facebook is one company that has been doing this for a while now because it is not something natural language processing tools can do. “Fact-checking is a necessity,” Azma says. “Plus, being human allows you to add nuance and catch built-in biases (racism, sexism, etc). A machine usually can’t tell what is inappropriate content, allowing this type of harm to slip through the cracks.”
However, she does admit to no longer thinking her gig as a science writer-for-hire is fully recession-proof. “The future of most technological advances tends to be driven by consumer demand,” Azma states. But before you start majorly stressing, remember that most of us have already been engaging with the beta-level version of such tech for years: It’s responsible for the 24/7 service provided via the online chat boxes we have come to expect as customers of several large and mid-size brands like Amazon.
Ultimately this means, we are better off figuring out how such tech can be a resource rather than a rival. ‘Doubling down’ on your areas of expertise never hurts as well since a bot can’t customize a teaser blurb for a new restaurant or hotel opening next month, or craft a review of a book or movie released two days ago. Thus own your niche, keep being the content pro you are—and charge accordingly.
(*Want the cliff notes version on how ChatGPT actually works? Read this post by Azma on the Fancy Comma blog.)