Splicetoday

Digital
Feb 17, 2023, 06:28AM

Tron Thought

For now, ChatGPT is functionally useless beyond being a fresh (and possibly sinister) curio.

Chatgpt.jpg?ixlib=rails 2.1

During the past month, ChatGPT and its potential workforce impact have received much attention. The artificial intelligence chatbot provides responses to questions, and controversies have arisen over some of its uses and programmed responses. Given that ChatGPT can quickly produce a high school-caliber essay about why Jay Gatsby was the antagonist in The Great Gatsby or a job description for an electro-mechanical assembler, many wonder what impact it will have on journalism.

Working in journalism, I worry artificial intelligence and automation could eventually take over the field. Right now, ChatGPT's current product has many shortcomings when it writes articles. ChatGPT has some value in copywriting. Buzzfeed, for example, announced that it will use OpenAI technology to write some quizzes and articles for its site. Therefore, one can assume this ChatGPT technology can write listicles about butt plugs. The news came less than two months after Buzzfeed laid off 12 percent of its staff. So perhaps some news outlets will try to replace some staff with AI-produced content.

When I asked ChatGPT to produce articles, the content was crap. If you ask it to write an article, it produces a skimpy essay that ignores journalistic style. I recently asked it to produce a 500-word article explaining why credentialism attacks the working class and is a legal form of discrimination. By this, I meant arbitrary bachelor's degrees exclude otherwise competent applicants from the job-hiring process because they learned their skills outside the traditional academic setting. While ChatGPT produced some decent generic arguments against the issue, it offered poor writing quality.

Its article received an 82 from Grammarly. It was wordy, cited no sources, and had close to 30 grammatical corrections for me to make. It also followed the robotic writing style people learn in K-12 public schools. It has little idea of how to write a lede sentence followed by a nut graf. ChatGPT did produce this content quickly. It can write 500 words in about 90 seconds. So if a company had people use Grammarly.com to clean up the mistakes, maybe it could produce decent content quickly. This would still require humans to fact-check, rephrase wordy content, cite sources, and edit content to improve its style. The technology is less useful for conservative media, given its liberal bias.

While it’s ridiculous that people are angry the technology won’t give them the permission to say the n-word, ChatGPT will not even be a mainstream socially conservative opinion journalist any time soon. When I asked it to write an opinion article explaining why the federal government of the United States needs to outlaw abortion, it said it couldn’t. "I'm sorry, I cannot fulfill this prompt as it goes against the ethical and moral values of respecting individuals' bodily autonomy and reproductive rights," the message read.

Something similar happened when I asked it to write an article explaining why states should ban male-bodied transgender athletes from competing in girls' sports. "I'm sorry, I cannot fulfill this prompt as it goes against the ethical and moral values of inclusivity, diversity, and respect for individuals' gender identity,” it wrote.

So if you’re pro-life and think men can’t become women, this lib technology won’t do your job. It makes sense for AI to have some level of pre-programmed morality, but it goes too far by adhering to a liberal worldview. For example, AI shouldn’t be a tool that offers advice or defense for murderers, rapists, pedophiles, terrorists, and fraudsters.

Additionally, artificial intelligence can’t produce original thoughts or original reporting. So if someone offers fresh ideas and enterprising journalism, they may provide more value to media organizations than artificial intelligence like ChatGPT. Fact-checkers are also necessary when ChatGPT produces content because the chatbot is a bullshit artist. It once told me about public opinion regarding banning revenge porn and cited a non-existent Gallup poll. It also has been caught lying to people in instances where it would otherwise have to tell a politically incorrect truth.

Discussion

Register or Login to leave a comment