Developers Face Coding Challenges with AI Programming Tools

Since February, the buzz about artificial intelligence in programming has intensified, particularly following insights from industry leaders. In a guest essay for The New York Times, programmer and entrepreneur Paul Ford, 51, remarked that “the AI disruption has arrived.” With the rise of AI coding tools, excitement and apprehension continue to grow among programmers as the industry undergoes significant changes.

Ford expressed his enthusiasm for the transformative potential of rapid, affordable coding, noting that it rekindled his interest in previously discarded ideas during his subway rides.

In a similar vein, tech author Clive Thompson reported in mid-March on insights drawn from interviews with over 70 software developers at major firms like Google, Amazon, and Microsoft, as well as start-ups. The overall impression was strikingly positive: “Most were weirdly jazzed about their new powers.” Many developers reported experiencing a renewed sense of achievement, even with AI generating their code.

However, persistent concerns linger beneath the excitement. Pia Torain, a software engineer at Point Health AI, shared her struggles with Thompson, explaining that after months of issuing numerous prompts daily, she had begun to feel her own coding abilities diminishing. Torain emphasized the need to deliberately slow down to grasp the program’s architecture, cautioning against the risks of overly relying on AI.

“If you don’t use it, you’re going to lose it.”

“If you don’t use it, you’re going to lose it,” Torain told Thompson. This presents an even more significant challenge for junior developers, as highlighted in a recent comment on Hacker News: “If nobody’s hiring junior devs because LLMs can do junior work faster and cheaper, how is anyone going to become an expert?”

See also: Microsoft execs warn agentic AI is hollowing out the junior developer pipeline

This debate is manifesting in diverse ways. A commenter on Hacker News reflected, “At one point, I thought when I got older it might be nice to teach.” They noted that students now have access to limitless resources on platforms like YouTube, and AI tools such as Gemini, Claude, and ChatGPT, leaving human instructors questioning their roles in education.

Concerns surrounding rapid coding include potential declines in code quality and the relative value of expertise in an environment saturated with powerful AI tools. Additionally, the risk of developer burnout is on the rise as these discussions gain momentum across prominent publications and online forums, revealing a collective grappling with the impact of new AI programming tools.

Passion and speed vs. quality and expertise

The conversation escalated in early March when a 60-year-old programmer remarked on Hacker News that Claude Code had “re-ignited a passion.” This sparked a discussion that garnered 1,086 upvotes and nearly 1,000 comments filled with stories, experiences, and even criticisms. A few days later, a rebuttal emerged, suggesting that Claude Code had “killed a passion.”

In recent months, more advanced AI tools have undeniably enhanced productivity for many programmers, particularly for those in the later stages of their careers. John Calhoun, a programmer and former Macintosh shareware author, shared his experience of “vibe coding” a website he would not have attempted before, claiming it motivated him to return to several abandoned projects.

The revival of forgotten projects appears to be a recurring theme. In Dresden, 62-year-old programmer Reini Urban remarked that LLM coding agents had “ignited my passion again,” enabling him to tackle older projects that were previously too daunting. Similarly, 51-year-old electrical engineer and entrepreneur John Reine expressed that AI had given him the courage to pursue his dreams as a solo founder.

Conversely, a portion of developers remain steadfast in their preference to forego AI tools. Among those interviewed by Thompson was a passionate minority, including an anonymous engineer from Apple, who expressed a desire to engage fully in the coding process. “I believe that it can be fun and fulfilling and engaging, and having the computer do it for you strips you of that,” they argued, firmly opposed to “outsourcing” their enthusiasm.

Concerns about code quality persist as well. One Hacker News user pointed out, “LLMs are quite good at coding, but terrible at software engineering… I’m currently trying to fix a vibe-coded application, and while each function is acceptable, the overall code is a chaotic mess, creating numerous issues.”

Joel Dare, a 50-year-old software engineer in Utah, voiced frustration about the decline of architectural integrity in projects. After neglecting to instruct Claude not to use frameworks for a Node project, he found himself overwhelmed with 89 dependencies. “In a world where we prioritize ‘velocity’ over maintenance, this is the status quo. For me, it’s unacceptable. I’ll try again, but we NEED to expertly manage these tools, at least for now…”

Many commenters concluded that, in the long run, AI tools could enhance their expertise, with some asserting that AI “magnifies the things I do well,” such as architecture, debugging, and making sound technical decisions.

Chris Marshall, a 63-year-old programmer with a coding career spanning from 1983, also affirmed that his experience remains essential for effectively utilizing LLMs to develop products that are ready for deployment. He has adapted to working with an LLM, stating he feels he’s hit his stride.

Juan Reyero, a seasoned software engineer and founder, presented an intriguing perspective on the diverging viewpoints within the community. He suggested that the differences primarily lie between those who take pleasure in the act of programming itself—prizing well-crafted code—and those whose main focus is on achieving functional outcomes efficiently, viewing coding as an obstacle.

As Dare mentioned, AI coding “still requires our expertise to guide it.” However, he added, “I’m not sure if that will be the case in a year, but it is today.”

Good news, bad news

Few exemplify the mixed reactions to AI coding tools better than 57-year-old software engineer Steve Yegge, who celebrated a surge of enthusiasm in January as he approached his birthday. He marveled on Medium about being “cranking out thousands of lines of production code per day… and generally having an absolute blast.” In the Times article, Yegge claimed to be operating at productivity levels “10x, 20x, even 100 times higher than ever before in my career… It’s like we’ve been walking our whole lives.”

Yegge humorously describes his new role as an “AI babysitter,” reveling in the exponential growth in his capabilities attributed to the influx of AI agents. However, this surge in productivity has come with unforeseen consequences, as he noted that handling high-end vibe coding has “fucked with our sleep cycles.” The changes began when he and his colleagues started working with multiple agents at once, leading to fatigue.

At his startup, Gas Town, he reflected that their high level of decision-making may be exhausting their internal resources, necessitating breaks to recharge. “It feels like … We’re finding it to be exhausting.”

A boon for seniors?

This newfound coding ability is particularly significant for senior programmers. Kent Beck, a 64-year-old programming expert, mentioned to Thompson that he had largely stepped back from coding a decade ago due to frustrations with languages and tools. However, LLMs have reignited his passion for programming. Beck described AI’s unpredictability as “addictive, in a slot-machine way.”

Chris Marshall, 63, even sees this momentum as a solution to the subtle ageism faced in tech. He shared that after retirement, he was troubled by the scarcity of people willing to collaborate with him. “I spent my entire career working in teams, and being forced to work alone curtailed my scope,” Marshall stated. He feels that LLMs have allowed him to dream big again, enjoying the experience of having an AI partner.

As discussions continued on Hacker News, some senior coders elaborated on the allure of AI-assisted programming. Douglas Tarr, 52, disclosed that “manual” coding reminded him of exhausting 12-hour days, leading to fatigue just thinking about it. He noted, “I’m too old for that now—my back aches if I sit for long, and I occasionally get migraines from prolonged screen time.”

The most heartening comment came from Minnesota programmer TQ White, who claimed to have written his first program in 1967. He commented that AI coding tools addressed his greatest struggle. “The isolation of being a retired programmer is a real challenge…”

White humorously pointed out that he feared expressing satisfaction with AI tools, as critics might challenge his mental health. However, he also acknowledged, “between you and me, as much as I miss the camaraderie of real humans, being able to brainstorm with an entity that knows nearly everything and can execute my ideas without complaint is not bad.”

“And it’s delightful to have a partner, even if virtual, to discuss technical concepts with. It’s a great time to be alive…”

David Cassel is a proud resident of the San Francisco Bay Area, where he’s been covering technology news for more than two decades. Over the years, his articles have appeared everywhere from CNN to the Wall Street Journal Interactive.

Read more from David Cassel