The Hidden Cost of Cheap AI: Undermining Entry-Level Jobs and Future Expertise

In 2007, Luke Arrigoni embarked on his career journey as a junior software developer, earning a respectable $63,000 in his first role. Fast forward to today, and Arrigoni, now an AI entrepreneur running Loti AI, a company specializing in detecting unauthorized deepfakes of Hollywood stars, observes a stark contrast: AI tools capable of writing code superior to his early efforts are available for a mere $120 annually. This dramatic disparity in cost—human labor versus AI utility—doesn't sit well with him.

Arrigoni voices a growing concern within the tech industry and beyond: the seemingly irresistible affordability of AI tools is creating a powerful economic incentive for companies to bypass entry-level human positions. His fear is that this trend could effectively truncate career paths before they even have a chance to begin. To counteract this, he proposes a counterintuitive solution: make AI systems more expensive. "If you make the AI systems more expensive, then you have an economic incentive to hire someone that is starting out," he argues, suggesting a fundamental shift is needed in how we value both AI capabilities and human potential.

A Perennial Anxiety Gains New Urgency

The notion of AI transforming or outright eliminating jobs is not a new one; it has been a recurring theme of anxiety in discussions about technological advancement for years. However, this concern is taking on a new level of urgency with the accelerating demand for sophisticated AI agents. These advanced AI systems are no longer confined to simple tasks; they are increasingly capable of performing complex functions previously exclusive to humans, such as conducting sales calls or writing intricate software code. This expansion of AI capabilities into domains once considered safe havens for human expertise is intensifying the debate about the future of work.

Despite these anxieties, the current situation isn't uniformly bleak. Data from hiring platforms like ZipRecruiter suggests that, at least for now, the landscape for aspiring professionals hasn't completely collapsed. Their estimates indicate that summer internships in the US this year have rebounded to roughly pre-pandemic levels. This offers a degree of temporary reassurance, suggesting that the immediate impact on entry-level opportunities might not be as catastrophic as some fear.

However, the future remains uncertain. Speaking at the Snowflake Summit in San Francisco, OpenAI CEO Sam Altman offered a perspective that, while framed around technological progress, fueled the underlying concern. He likened current AI tools to interns, implying that the next generation of AI would be akin to more experienced workers. Altman went further, claiming that some managers are already beginning to supervise "a bunch of agents" in much the same way they traditionally oversaw "relatively junior employees." This vision, while highlighting the increasing sophistication of AI, underscores the potential for direct replacement of human roles.

OpenAI has publicly discussed potential mitigation strategies for the anticipated jobs crisis, such as investing in reskilling programs to help workers adapt to the changing landscape. What they haven't proposed, however, is increasing the price of their services as a means to deliberately slow down the transition from human labor to AI-driven work. This absence of pricing intervention is precisely what keeps entrepreneurs like Arrigoni on edge.

Even when accounting for premium features and add-ons, the cost of AI coding agents remains a mere fraction of the salary required for a junior engineer. Arrigoni's core worry is that if inexperienced workers are unable to secure these foundational roles, they will be deprived of the crucial opportunities needed to acquire the skills, knowledge, and practical experience necessary to eventually lead teams in the future—teams that may comprise both humans and machines. The potential long-term consequence is a deficit of experienced professionals capable of navigating and managing increasingly complex technological environments.

"Less Than Human": The Economics of AI Pricing

The pricing of AI tools has been a dynamic and somewhat unpredictable factor since the public launch of ChatGPT in 2022 ignited the current AI boom. Initially, many AI companies offered free tiers to encourage widespread adoption, and the cost of basic services has generally decreased over time. While top-tier plans offering the most advanced features have seen price increases, these costs have typically not reached levels that guarantee profitability for the AI providers, nor have they significantly deterred rapid adoption by businesses eager to leverage AI's perceived advantages.

Startup executives and pricing consultants point to intense market competition as a primary driver of these low prices. "Their only way to win is mass adoption," explains Ajit Ghuman, CEO of pricing strategy company Monetizely. This competitive pressure forces AI companies to keep their prices comparable to rivals, prioritizing market share and user growth over immediate profitability. Ghuman suggests that without significant disruptions, such as major shortages in essential resources like electricity or GPUs, or the emergence of a single company dominating the AI market, it's unlikely that AI prices will see substantial increases in the near future.

Decagon, a San Francisco-based startup developing customer service chatbots used by major retailers and tech companies, exemplifies this pricing strategy. They charge $1 or less per conversation for their chatbot services, a price point roughly half the cost of employing human support staff for the same task. Jesse Zhang, Decagon's CEO, acknowledges that in some instances, their chatbot might even outperform a human agent. Yet, he believes their clients would not be willing to pay more than the human equivalent. "The reason to invest in AI is efficiency," Zhang states. "You're going to be less than human labor. That's kind of like the point of technology."

Zhang notes that Decagon is profitable on a per-conversation basis, excluding certain overheads, but declined to comment on the company's overall profitability. However, with significant venture capital backing, including $100 million raised from prominent firms like Andreessen Horowitz and Accel, Decagon has the financial flexibility to prioritize aggressive growth and market penetration over immediate profit margins. Zhang reflects on the pricing strategy, saying, "Whether we could be pricing more, it's always like a 'what if?' But in general we're pretty happy right now." This sentiment highlights how venture funding can enable companies to maintain low prices to capture market share, even if it means delaying profitability.

"So Cheap": The Disproportionate Value of AI

Erica Brescia, a managing director at the investment firm Redpoint Ventures, experienced a moment of realization regarding AI agent pricing recently. She was struck by the $250 annual price tag for Google's new AI Ultra plan, finding it astonishingly low. "All this is so cheap," she recalled thinking, noting that the price seemed "disproportionate to the value people are getting." In her estimation, a price at least double that amount would be more reflective of the utility provided. This observation underscores a key tension: the market price of AI tools appears significantly lower than the value they deliver in terms of productivity gains and cost savings compared to human labor.

Brescia's perspective is informed by her previous role as chief operating officer at GitHub, a company that played a significant part in establishing early benchmarks for AI pricing in the developer community. GitHub's Copilot coding assistant, launched at $10 a month in 2022, predating the widespread public awareness triggered by ChatGPT's debut by several months. Brescia explains that GitHub deliberately chose an accessible price point to encourage a critical mass of users. The primary objective was to gather extensive data to refine and improve the service. Microsoft, GitHub's parent company, was willing to absorb losses on the new tool to achieve this strategic goal of data acquisition and product enhancement. Brescia now estimates that a price 100 times higher than the initial $10 would more accurately reflect the immense value Copilot provides to software developers in terms of increased productivity and reduced coding time.

Today, GitHub Copilot's pricing tops out at $21 a month, and many similar AI coding tools have followed this lead, establishing a competitive landscape where low prices are the norm. Zed, a company that has received $12.5 million in funding from Redpoint and other investors, launched its AI-assisted code editor in May with a minimum price of $20 a month. This pricing aligns closely with the precedent set by Copilot, reinforcing the market expectation for affordable AI developer tools.

Nathan Sobo, CEO of Zed, anticipates that AI companies will eventually need to increase their prices because the current models are not sustainable in the long term, particularly as the costs associated with developing and running large AI models remain substantial. However, he also expresses a desire to keep AI agents relatively affordable compared to human labor. His rationale is that accessibility allows more people to use these tools to augment their work, leading to better software development and potentially creating new types of jobs. "I want as much intelligence at my disposal at as low a cost as possible," Sobo states. He adds that this desire includes enabling junior engineers to utilize this technology affordably, suggesting that low cost can also be a tool for empowering human workers rather than solely replacing them.

Decagon's Jesse Zhang echoes this sentiment regarding AI coding tools, acknowledging that while his company might be willing to pay slightly more for such services, a significant price jump, like to $2,000, would likely be prohibitive. He emphasizes that "the hunger for good engineers is infinite," suggesting that even with powerful AI tools, the demand for skilled human talent remains high, particularly for complex tasks and strategic roles that AI cannot yet fulfill.

The Career Pipeline Problem: Where Do Future Leaders Come From?

The most profound concern raised by the low cost and increasing capability of AI agents is their potential impact on the traditional career pipeline. Entry-level jobs have historically served as crucial training grounds, providing new graduates and inexperienced individuals with the practical skills, industry knowledge, professional networks, and soft skills necessary to advance in their careers. These roles are where foundational competence is built, where individuals learn to collaborate within teams, solve real-world problems, and gain the experience needed to take on more senior responsibilities, including management and leadership.

If companies widely adopt cheap AI agents to perform tasks previously assigned to junior staff—such as basic coding, initial customer support inquiries, data entry, or preliminary research—the number of available entry-level positions could shrink dramatically. This reduction in opportunities could create a bottleneck, making it significantly harder for new talent to get a foot in the door and acquire the necessary experience. The long-term consequence is a potential shortage of experienced professionals down the line, impacting the ability of industries to innovate and grow.

Luke Arrigoni's worry stems from this exact scenario. If the initial rung on the career ladder is removed or made inaccessible, how will individuals gain the expertise required to become senior engineers, project managers, or even the leaders who will oversee future AI-driven operations? The skills learned in early roles—debugging complex code, handling difficult customer interactions, navigating corporate structures, collaborating effectively—are often tacit and acquired through hands-on experience, not easily replicated by theoretical knowledge or AI assistance alone.

The paradox is that while AI agents are becoming more sophisticated, they still require human oversight and management, particularly for complex or novel situations. The article mentions that some companies are beginning to hire "AI architects" to manage and refine agentic systems. But who will fill these roles in the future if the pool of experienced professionals, who developed their skills through traditional career progression, diminishes? The concern is that the very efficiency gained by using cheap AI today could undermine the development of the human capital needed to manage the advanced AI systems of tomorrow.

Challenges and the Need for Human Oversight

While the potential value of AI agents is high, their current limitations also play a role in their pricing and adoption. AI entrepreneurs suggest that agents could command higher prices if they were easier to set up, more reliable, and less prone to errors or unexpected behaviors. Nandita Giri, a senior software engineer with experience at major tech companies like Amazon, Meta, and Microsoft, illustrates this point. She states she would be willing to pay thousands of dollars annually for a truly effective AI personal assistant. However, she adds a crucial caveat: "But strict conditions apply—you can't get frustrated by using it."

Giri's personal experience highlights the current gap between the promise and the reality of AI agents. As a personal project, she attempted to develop an AI agent designed to help prevent psychological burnout. The outcome was unexpectedly drastic: "It just canceled all my meetings," she recounts. While technically a solution to a packed schedule, it was far from the nuanced, reliable assistance required for a professional setting. This anecdote underscores that despite their capabilities, current AI agents can be unpredictable and require significant human intervention, debugging, and oversight to function effectively in complex, real-world scenarios.

This need for human oversight is giving rise to new roles, such as the "AI architect" mentioned earlier. These professionals are tasked with designing, deploying, and managing AI agent systems, ensuring they operate correctly and safely. However, the question remains: where will the talent for these roles come from if the traditional pipeline of entry-level jobs, which provides the foundational technical and problem-solving skills, is disrupted? The concern is that by optimizing for short-term cost savings through cheap AI, companies may inadvertently create a future talent gap for the very roles needed to manage advanced AI.

Seeking Solutions: Pricing, Policy, and Prioritization

Addressing the potential long-term costs of cheap AI requires considering various perspectives and potential interventions. Economist Simon Johnson at the Massachusetts Institute of Technology points out that companies, driven by market forces, are unlikely to voluntarily factor in the broader social cost of career disruption when setting prices for AI tools. He suggests that government intervention might be necessary to counteract this market failure. Johnson proposes lowering payroll taxes specifically for entry-level roles as a mechanism to make hiring junior staff more economically attractive for employers. "The right lever to pull is one that reduces costs to employers," Johnson argues, suggesting that policy could help rebalance the economic incentives.

Luke Arrigoni, the entrepreneur who initially highlighted the issue, is taking a different, more direct approach within his own company, Loti AI. Rather than relying on policy changes or hoping for shifts in AI pricing, he has made a conscious decision to prioritize the steady hiring of junior engineers. Notably, his company has not adopted AI coding tools, despite their affordability and potential efficiency gains. Arrigoni explains his reasoning: if the feared "job apocalypse" driven by AI displacement comes to pass, "I don't want to be at fault." His approach represents a choice to prioritize the human career pipeline and the development of new talent, even if it means foregoing the immediate cost savings offered by cheap AI.

The debate over AI pricing and its impact on jobs is complex, involving economic incentives, technological capabilities, and societal well-being. While the efficiency and cost savings offered by cheap AI tools are undeniable and attractive to businesses, the potential long-term consequences for career development and the future workforce cannot be ignored. The discussion involves finding a balance between leveraging powerful new technologies for productivity gains and ensuring that opportunities remain available for the next generation of human professionals to gain experience, develop expertise, and lead the way in an increasingly AI-integrated world.

The Interplay of Competition, Value, and Sustainability

The current low pricing of AI tools is heavily influenced by the intense competition among AI providers. Companies are in a race to acquire users and data, believing that mass adoption is key to refining their models and achieving long-term dominance. This competitive pressure drives prices down, often below the true cost of development and operation, particularly when factoring in the immense computational resources required to train and run large language models and AI agents. Venture capital funding further enables this strategy, allowing startups like Decagon to prioritize growth over immediate profitability, offering services at prices that human labor simply cannot match.

However, this model raises questions about sustainability. As Nathan Sobo of Zed notes, current pricing models may not be viable indefinitely. The significant costs associated with AI development, infrastructure, and ongoing research and development will eventually need to be recouped. If prices do rise in the future, it could potentially shift the economic calculus for businesses, making human labor relatively more competitive again, or at least narrowing the cost gap. Yet, the market expectation for cheap AI, set by early movers like GitHub Copilot and reinforced by intense competition, makes significant price increases challenging without risking customer churn.

The perceived value of AI also plays a critical role. While AI agents can perform specific tasks efficiently, their current limitations in terms of reliability, adaptability, and the need for human oversight temper their value proposition compared to a fully capable human worker. As Nandita Giri's experience illustrates, an AI agent that cancels all meetings, while perhaps a humorous anecdote, highlights a lack of nuanced understanding and control that a human assistant possesses. Until AI agents achieve a higher level of reliability and autonomy, their value might be capped in certain complex roles, potentially justifying lower prices but also necessitating continued human involvement.

The tension lies in the fact that while AI's current imperfections might justify lower prices, the *potential* value it offers in terms of efficiency and cost reduction is enormous, as highlighted by Erica Brescia and Jensen Huang's willingness to pay a high salary for an ideal AI agent. This gap between current price and potential value, coupled with the competitive drive for adoption, creates the economic pressure that threatens entry-level human jobs.

Navigating the Future of Work

The discussion about cheap AI and job displacement is fundamentally a conversation about the future of work and the evolving relationship between humans and technology. It's not simply about whether AI can perform a task, but at what cost—both economic and social—and what the long-term implications are for skill development, career progression, and societal structure.

The insights from various individuals in the article offer different perspectives on how to navigate this transition:

**Economic Rebalancing:** Luke Arrigoni suggests increasing the price of AI to make human labor more competitive. Simon Johnson proposes government intervention through payroll tax cuts for junior roles to achieve a similar outcome by reducing the cost of hiring entry-level staff. Both approaches aim to alter the economic incentives currently favoring AI adoption over human hiring.
**Prioritizing Human Development:** Arrigoni's decision at Loti AI to hire junior engineers and avoid AI coding tools represents a choice to actively preserve the human career pipeline, prioritizing long-term talent development over short-term efficiency gains.
**Augmentation vs. Replacement:** Nathan Sobo and Jesse Zhang express a desire for AI tools to be affordable to *augment* human work and create new opportunities, rather than solely replace existing roles. This perspective emphasizes AI as a tool for human empowerment and productivity enhancement.
**Addressing AI Limitations:** Nandita Giri's experience highlights the need for AI agents to become significantly more reliable and less frustrating to use before they can fully replace human roles requiring nuanced judgment and consistent performance. Improvements in AI reliability could potentially justify higher prices, but also might make them more capable of replacement.

The emergence of roles like "AI architect" suggests that the future workforce will likely involve humans working alongside and managing AI systems. However, ensuring a sufficient supply of skilled individuals for these new roles requires a healthy pipeline of talent gaining foundational experience today. The current economic pressure exerted by cheap AI threatens this pipeline, creating a potential future deficit in the very expertise needed to manage the AI revolution.

Conclusion: A Critical Juncture

The affordability of AI tools presents businesses with powerful incentives for efficiency and cost reduction. However, as highlighted by entrepreneurs, economists, and technologists, this short-term gain may come at a significant long-term cost to the human workforce. By potentially eliminating entry-level positions, cheap AI threatens to disrupt the traditional career pipeline, making it harder for new talent to gain experience and develop the skills necessary for future leadership and innovation.

The current pricing models, driven by intense competition and the pursuit of mass adoption, do not appear to fully account for this societal cost. While AI capabilities continue to advance, their current limitations still necessitate human oversight and management, creating new roles that require experienced professionals.

Navigating this critical juncture requires careful consideration from businesses, policymakers, and individuals. Whether through adjustments in AI pricing, government incentives for hiring, or a conscious prioritization of human talent development, finding ways to balance the benefits of AI efficiency with the need to maintain a robust and experienced workforce is crucial for ensuring a prosperous and equitable future of work. The story of cheap AI is not just about technology; it's about the economic and social choices we make today that will shape the career opportunities of tomorrow.

Illustration of a coupon page with an AI robot agent dollar signs and the words bargain and save — Illustration: Jacqui VanLiew

External Sources:

Subscribe to Our Tech & Career Digest