Let no one ever say the Charleston City Paper isn’t up for a challenge.
We wanted to find out how accurate and efficient artificial intelligence (AI) really is — and how likely it could be to replace us in the future. So we sat down with a local software expert, Mike Astle, who works with tech group CharlestonHacks, to talk about it.
Then, we did what we always do. We wrote an article, which went through its usual, rigorous editing process, and trimmed it to a snappy 600 words. But we also did something we don’t usually do.
“I recorded the conversation with Astle, which lasted about half an hour and spanned topics from the lore of Dune to the Cold War,” said City Paper reporter Skyler Baldwin. “Then I sent it to ChatGPT with a simple prompt:
“In this audio file, reporter Skyler Baldwin interviews a Charleston, S.C., software professional about artificial intelligence. Write a 600-word news story following AP-style guidelines based on this interview.”
It took about 30 seconds. And after reading it, most of us at the office agreed it was pretty good — even if it overshot the word count by more than 100 words. But we also realized there was a flaw in the test: A human still contributed about half the work by providing the interview.
So we wrote another prompt in a new window, using a new profile, asking it to generate a story using Astle as a source but without providing it with a recording or transcript. The resulting story was, expectedly, less compelling, leaving big gaps in the story for verified quotes to be inserted later and showing that maybe we have a few years left in business after all.
Check out the results. The first story by Baldwin has been edited. The second two are presented as is – with no editing or corrections. So far, we still win. —City Paper Staff
CP reporter | AI test No. 1 | AI test No. 2
CP reporter: Charleston software developer says AI ‘nowhere near’ its limit (600 words)
Skyler Baldwin | Artificial intelligence has grown more pervasive, more accepted and generally more accurate since ChatGPT’s public release in late 2022. And according to one local software expert, it may not be too long until AI starts replacing some human creators.
Users have clocked the powerful tool as everything from a fun new toy to silly mistake-maker to existential threat. But software developers and tech gurus are watching the growth of AI with interest, more than anything else.
“The incredulity-to-doubt cycle was very fast in the early days,” said Charleston native Mike Astle, a longtime local software developer who graduated from the prestigious California Institute of Technology. “All the software people I knew were extremely impressed by how effective the chat experience was, but also very invested in finding the edge cases of where it goes wrong.”
But concerns remain — from AI’s accuracy to people’s overreliance on the tool in day-to-day life.
“If you’re taking something that somebody sent you and putting that into AI to generate a response, and sending back that response, you’re done. You’re toast,” Astle said. “Nobody wants to hear what your [AI] thinks. They want to hear what you think.”
But that doesn’t mean you should remove AI from your toolbox entirely.
“Think of a smooth spectrum moving away from that,” he said. “If I take the response, read it, understand it and approve it, is that OK? If I read it, rewrite it and then send it back, is that OK? There’s a continuum, and it gets more complicated when you start thinking about what kind of work it is and what kind of culpability or responsibility you have for it.”
When it comes to accuracy, however, Astle said he thinks AI’s mistakes — at least partially — can be attributed to human psychology more than the AI itself. Astle said he tends to think AI’s mistakes are obvious, glaring or otherwise more visible. But the small mistakes that AI prevents humans from making in daily tasks — from spelling to math — vastly outnumbers the inaccuracies it lets slip through.
“There’s a certain bias here,” he said. “We want humanity to win.”
But that chance of victory is growing slimmer. AI is getting smarter, faster and more human over time, Astle said. Every interaction with AI is a data input, building up the model’s wealth of information and giving it more information to find patterns in.
“We’re nowhere near the boundary of scale here,” Astle said. “It’s just a matter of how much silicon we can rack up to power it.”
And it’s only a matter of time, he said, before AI becomes capable of truly replacing creators — including journalists.
“I would say probably we’re at least three years out,” Astle said. “Probably no more than eight. … My guess is that the very best writers and the very best musicians and the most interesting performers will remain human for probably an extra couple years, or maybe generationally, before acceptance and preferences balance it out.”
That seemingly boundless growth is what drives a lot of people’s growing concerns. Astle compares their attitudes to the overall feeling of the Cold War.
“AI is the only thing people will toy around with and also really believe will destroy them,” he said. “It’s a funny dichotomy.
And combatting that “impending” doom is relatively simple, Astle said: It starts with legislation.
From requiring registration of hardware capable of housing a high-performance AI model, to reinforcing an infrastructure of responsibility for corporations that employ them, the legal mechanisms are already in place. It’s just a matter of applying them fairly and responsibly, he said.
AI test No. 1: Charleston software professional weighs AI’s promise against its risks (711 words)
ChatGPT | Artificial intelligence is advancing quickly enough to reshape industries, eliminate jobs and raise questions about humanity’s future. But for Charleston software professional Mike Astle, the technology presents both reasons for concern and opportunities worth exploring.
Astle, who has worked in software for about 30 years, has watched artificial intelligence evolve from a novelty into a tool with increasingly significant implications for the people who use it.
“I think it went very fast from cute toy that’s fun, I’ve seen this before, to like, hold on,” Astle said. “This is getting better so fast.”
Astle, a Charleston native and member of the local innovation community Charleston Hacks, works primarily on software development projects. His experience has given him a firsthand view of how AI is changing the industry and how professionals are adapting to its capabilities.
One of the central challenges, he said, is determining when AI serves as a useful tool and when it becomes a substitute for human judgment and responsibility.
Astle said people should evaluate AI-generated work as they would work produced by an employee. If someone incorporates that work into a product or message they present as their own, they should take responsibility for its quality and accuracy.
“Nobody wants to hear what your LLM thinks,” Astle said, referring to a large language model, the technology behind many generative AI tools. “They want to hear things from you.”
That distinction becomes especially important as AI-generated content becomes more common. Astle said readers can sometimes recognize such material through its tone, structure and lack of individual personality.
He pointed to overly polished language, familiar metaphors and emoji-filled bullet points as potential clues. More broadly, he said AI-generated writing can sound so balanced and impersonal that it lacks the distinctive voice of a human author.
Still, Astle cautioned against assuming that AI is inherently unreliable. He suggested that people may notice and remember AI’s most obvious mistakes while overlooking errors they make when researching information themselves.
The technology’s ability to synthesize information from increasingly large amounts of data also could allow it to identify patterns and reach conclusions that humans have not previously considered, he said.
Those capabilities, however, come with concerns about employment and the concentration of power among companies developing advanced AI systems.
Astle estimated that AI could significantly reduce the number of workers needed in some professions. In software development, he said, tasks that once required five engineers may now be completed by one.
Journalism could face similar changes as AI automates research, review and other tasks, allowing one person to cover more responsibilities.
Astle predicted that the complete replacement of some workers could become a possibility within three to eight years, though he acknowledged that his assessment of journalism was a guess rather than a conclusion based on industry experience.
He also questioned whether society is doing enough to address the potential dangers of increasingly powerful AI systems.
Astle said governments could consider stronger regulations, greater corporate accountability and measures to prevent excessive concentration of power in the technology industry. He also raised the possibility of international cooperation, comparing the challenge to efforts to control nuclear weapons during the Cold War.
Local opposition to data centers, which provide computing infrastructure for AI, reflects some of those broader concerns. Astle said he finds it interesting that such facilities can provoke strong reactions even as governments offer economic incentives to other large industrial projects.
He suggested that public anxiety about AI may stem partly from fears about the technology’s long-term consequences.
Despite those concerns, Astle sees creative possibilities in AI, including personalized entertainment that responds to an individual’s reactions in real time and new forms of digital art.
He said people may increasingly accept artificial creators as technology becomes integrated into everyday life.
For Astle, managing AI’s risks while exploring its benefits requires balancing collective responsibility with individual curiosity.
He said people should rely on their communities to help manage anxiety about potential threats while drawing on personal experiences to recognize opportunities.
Ultimately, Astle said, society has a responsibility to respond if it recognizes a serious danger posed by AI.
“If there is some global negative AI and we didn’t do anything to counter it even though we knew,” he said, “then we have this sense of responsibility.”
AI test No. 2: As Artificial Intelligence Advances, Workers Face Growing Uncertainty About Their Future (636 words)
As artificial intelligence becomes more capable of performing tasks once handled exclusively by people, concerns are growing about the technology’s potential to reshape the workforce, eliminate certain jobs and change what employers expect from their employees.
For workers across industries, the question is no longer simply whether artificial intelligence will become part of the workplace, but how quickly it will change the nature of their jobs. From software development and customer service to administrative support and data analysis, tasks that once required hours of human effort can increasingly be completed with the assistance of AI tools.
Mike Astle, a software professional from Charleston, S.C., offers a perspective from an industry already experiencing rapid technological change. His experience in software could help illustrate how artificial intelligence is changing professional expectations and raising questions about the future demand for human workers.
[INSERT VERIFIED COMMENT FROM ASTLE ABOUT HIS EXPERIENCE WITH AI IN THE SOFTWARE INDUSTRY.]
Artificial intelligence systems can generate computer code, summarize documents, analyze large amounts of information and assist with routine problem-solving. Businesses may view these capabilities as opportunities to increase productivity, reduce costs and complete projects more quickly.
But those same capabilities have prompted concerns about whether employers will need as many workers to accomplish the same amount of work.
Jobs that involve repetitive, predictable tasks may face particular pressure as companies adopt increasingly sophisticated automation. Even positions requiring specialized training could change as AI tools take on portions of the work traditionally performed by experienced professionals.
Software development is one example. AI-powered tools can suggest code, identify potential errors and help developers produce applications more efficiently. Although these systems still require human oversight, their expanding capabilities could influence hiring decisions, job responsibilities and the skills employers value.
[INSERT VERIFIED COMMENT FROM ASTLE ABOUT WHETHER AI IS MORE LIKELY TO REPLACE SOFTWARE WORKERS OR CHANGE HOW THEY DO THEIR JOBS.]
The potential consequences extend beyond individual occupations. If companies can accomplish more with smaller teams, some workers could face reduced opportunities, while others may be expected to manage AI systems or oversee work previously completed by colleagues.
However, technological change does not necessarily translate into the disappearance of entire professions. New tools can also create demand for workers who develop, maintain and evaluate AI systems, protect computer networks, verify automated outputs and address problems that machines cannot reliably solve.
Human judgment remains important when decisions involve complicated circumstances, accountability, ethical considerations or an understanding of people’s needs. AI systems can also make mistakes, produce misleading information and struggle with situations that differ from the examples on which they were trained.
For employees, adapting to these changes may require continued education and a willingness to develop new skills. Technical knowledge, critical thinking, communication and the ability to evaluate AI-generated information could become increasingly valuable as workplaces integrate automated tools.
Employers, meanwhile, will face decisions about whether AI is used primarily to reduce labor costs, support existing employees or create new products and services. Those choices could determine whether the technology becomes a source of greater opportunity or a threat to workers’ financial security.
[INSERT VERIFIED COMMENT FROM ASTLE ABOUT WHAT WORKERS SHOULD DO TO PREPARE FOR AI-DRIVEN CHANGES.]
The pace and scale of AI-related job displacement remain uncertain. Outcomes will depend on how quickly the technology improves, how businesses implement it and whether new jobs emerge to offset positions that are eliminated.
For workers and employers alike, the challenge will be finding a balance between the efficiencies artificial intelligence offers and the human experience, judgment and accountability that businesses continue to need.
As adoption accelerates, the experiences of professionals such as Astle may help illustrate a broader question facing the economy: whether artificial intelligence will primarily become a tool that makes people more productive or a technology that allows companies to operate with fewer people.
Help keep the City Paper free.
No paywalls.
No subscription cost.
Free delivery at 800 locations.
Help support independent journalism by donating today.
