As generative artificial intelligence transforms industries across the globe, its impact on the labor market has raised urgent questions about the future of work. A landmark report from the Center for an Urban Future reveals a paradox at the heart of New York City’s economy: while overall tech employment remains near record highs, entry-level tech job postings have plummeted by 49% since 2022. Rather than triggering widespread layoffs, AI appears to be putting particular pressure on the bounded, routine tasks that historically served as the training ground for early-career professionals.
In this conversation, Searchlight editor Robert A. George speaks with Eli Dvorkin, editorial and policy director at the Center for an Urban Future, to examine how AI is reshaping career pathways, why the transition disproportionately impacts entry-level candidates, and what actionable policy solutions can ensure economic mobility for all New Yorkers.
Searchlight: What’s the main headline here? Your report outlines the pressures that AI is placing on the tech sector and entry-level employment. Could you sketch out the lay of the land for us?
Eli Dvorkin: Absolutely. Let me start with some good news, which might be surprising given the general tone surrounding this topic recently. Despite all the fears around AI, the core challenge right now is not whether New York City has technology jobs. Arguably, the city’s tech sector is larger, stronger, and more diversified than ever. Overall tech employment in New York sits at around 208,000 jobs, which is nearly 24% higher than before the pandemic.
There is no indicator yet that AI is leading to anything close to the mass job extinction event that many have prophesied. If anything, our tech sector has proven remarkably resilient given all these macroeconomic changes and technology shifts, and it remains well-positioned for future growth.
However, the real challenge is who gets the opportunity to enter that sector in the first place. Our top-line finding is that while New York’s tech economy has not cratered—though hiring has certainly slowed dramatically since 2022—the first rung of the career ladder has snapped in half. Entry-level tech job postings in New York City have plummeted 49% since 2022. While New York still has a powerful tech economy producing significant numbers of middle- and high-wage jobs, and, crucially, holding onto the jobs it already has, what is breaking down right now is the pathway into it.
Searchlight: Let’s zero in on that. When we talk about entry-level jobs in the context of the tech sector, what specific kinds of jobs are drying up?
Dvorkin: In general, we’re talking about a wide range of computer and mathematical occupations. They aren’t the sum total of every job at a tech company—you could work in tech in HR, accounting, or facilities. But when we think about classic tech roles where the primary function is building technology, computer and mathematical occupations comprise the vast majority.
We are talking about software developers, computer programmers, data scientists, computer systems analysts, network architects, and cybersecurity experts. These occupations are the backbone of tech jobs across our entire economy, not just in pure tech firms, but everywhere from healthcare to finance. For our report, we shined a light on those computer and mathematical occupations because they represent the core software and data talent pool.
Searchlight: A few years ago, when other industries were facing disruption, the somewhat derisive cliché thrown around was “learn to code.” Are we basically seeing that even learning to code is no longer a guaranteed pathway into the tech world?
Dvorkin: That is definitely a core component of the challenge. Or rather, learning to code at what was formerly thought of as the entry level will no longer be sufficient.
There are now, and will continue to be, software engineers working in tech and other industries despite—and in some cases enabled by—the growth of AI. I don’t think those roles are going away anytime soon. The persistence of those roles across the economy suggests they aren’t directly threatened with full wholesale replacement, at least not yet.
What is happening is that the barriers to entry are rising. The bar for operating as a software engineer is going up and up, likely in large partbecause of AI. There still needs to be a human in the loop exercising judgment, maintaining quality control, and ensuring that code generated by AI agents is vetted by an experienced human who understands the system. But that oversight can now be done with far fewer humans than was previously required.
In part because the level of sophistication needed to provide quality control over AI-generated code bases is so high, the entry points into those careers are shrinking. The most automatable work in the sector was historically done by employees with the least experience. So while these professions aren’t disappearing, the barrier to entry is growing, expectations are increasing, and all of that places immense pressure on the beginning of the career ladder.
Historically, entry-level work involved bounded, routine tasks: writing basic code, cleaning data, troubleshooting routine problems, producing first drafts, or conducting preliminary analysis. Those are precisely the tasks AI can help an experienced worker execute much faster. A senior employee using AI can now handle those entry-level tasks faster than a junior employee could. That doesn’t eliminate the occupation, but it means a team that once required 10 people may now only need eight. The two positions that disappear are often the exact roles where someone would have learned the business in the first place. In short: AI can eliminate the training ground without eliminating the profession.
Searchlight: Eliminating the training ground without eliminating the profession—that is a crucial distinction. It aligns with what we’ve experienced in journalism. Tasks like aggregating links or doing baseline research that used to take hours or require an intern can now be jump-started using AI models in minutes. Which leads to a broader question. The public fear surrounding AI often centers on the idea that as each iteration becomes smarter, automation won’t stop at entry-level roles. If AI is constricting entry-level jobs today, is it legitimate to worry that the position above entry level will be squeezed next, and then the level above that?
Dvorkin: At the moment, the clearest impact in our data is overwhelmingly concentrated at the entry level. That doesn’t mean the scenario you’re proposing is impossible. There is a very real possibility that this conversation will look dramatically different five years from now. But without a crystal ball, it can be counterproductive to speculate beyond what the current data supports.
The data suggests both good news and bad news. The worst-case predictions of total job extinction from the past few years have not come to pass. That is due to a mix of factors: the actual speed of technological deployment, macroeconomic conditions like higher interest rates, and societal choices about what we value in the workplace.
What is clear right now is that the early impacts of AI adoption fall disproportionately on early-career roles. However, the data also reveals a nuanced picture when you look at how different occupations are affected:
- Automation vs. Augmentation: Job postings have fallen by 20% for entry-level occupations where AI automates more tasks than it augments. Conversely, for occupations where AI augments human work more than it automates it, entry-level job postings have actually increased by 3%.
- Low-Exposure Roles: Entry-level postings in occupations with minimal AI exposure have grown by 21%.
- Demand for AI Skills: Entry-level job postings explicitly requiring AI-related skills have increased by 20%.
So we are seeing a bifurcation in labor market demand rather than a single narrative. Where roles consist of easily automatable tasks, demand is dropping. Where AI enhances productivity, demand is holding or growing. The core issue is that employers face an individual incentive to hire workers who already have two or three years of experience. But if every employer acts that way, the broader entry-level ecosystem collapses, which ultimately harms employers as well.
Searchlight: How can policy address this breakdown in the entry-level pipeline? What specific recommendations does your report propose?
Dvorkin: Because we have a bottleneck at the front door—where recent graduates, training program alumni, and early-career workers are competing for fewer open roles—policymakers should step in. If the private sector isn’t offering that initial training ground, we have to create alternative pathways for workers to gain real-world experience, particularly those from low-income backgrounds or institutions such as CUNY and nonprofit training programs like Pursuit or Per Scholas.
We outline six proposals, including:
- Launch an NYC AI Service Corps: The city, in partnership with training providers and the private sector, should establish a paid six-month service fellowship. Emerging technologists and recent grads would work on real-world AI and data projects within city government, non-profits, and small businesses—sectors that have historically been slower to adopt new technology. Participants get the paid, hands-on experience that private employers demand, while public and non-profit entities gain technology capacity. We propose an initial city seed investment of $10 million, leveraged alongside federal funding like AmeriCorps, national workforce initiatives like RAISE US, and private philanthropy.
- Recruit the Next 100 Large Employers into Talent Partnerships: Intermediaries like the New York Jobs CEO Council currently work with roughly 30 of the city’s largest corporations to build direct hiring and internship pipelines from CUNY. However, New York is home to dozens of second-tier large employers that don’t traditionally build these structured pathways. City Hall should use its convening authority to bring the next 100 large employers into formal talent partnerships offering paid internships, apprenticeships, and direct hiring pipelines. This is especially urgent given that paid internship postings in New York City have dropped by over a third compared to pre-pandemic levels.
- Establish an AI Workforce Readiness Fund: New York should establish a dedicated fund to help workers and workforce organizations adapt to technological change. This fund would focus on three areas:
- Incumbent Upskilling: Helping existing middle-wage workers acquire AI-native skills so they can adapt as their job requirements change, preventing displacement in established middle-class careers.
- Rapid Retraining: Providing transition pathways for early-career workers in fields experiencing high levels of automation so they can move into growing sectors.
- Subsidized Work-Based Learning: Expanding access to subsidized internships and experiential learning to help non-traditional candidates gain a foot in the door.
To fund this, the city should actively pursue public-private partnerships, working with federal initiatives and major frontier model companies—such as OpenAI, Anthropic, and Google—to invest directly in workforce adaptability.
Searchlight: Is there an underlying tension between advocating for these public-private partnerships and the reality of city government operation? Could integrating AI within municipal operations lead to a smaller city workforce, much like in the private sector?
Dvorkin: Government should lead by example in two distinct ways: as a smart adopter of technology and as an exemplary employer.
Responsible AI adoption in city government should focus on cutting bureaucracy rather than cutting headcount. Civil servants often spend vast amounts of time navigating administrative hurdles, manual data entry, and compliance reporting. Automating routine paperwork and administrative tasks frees city workers to focus on direct, human-centered public service. AI should be used to improve the experience of working in and interacting with city government.
As an employer, government should serve as a model by modernizing civil service hiring pipelines and building transparent partnerships with local training institutions like CUNY. By demonstrating how to effectively recruit and upskill local talent internally, the city gains the leverage needed to ask the private sector to do the same.
Ultimately, economic and technological shifts are moving faster than traditional education systems. By building infrastructure like the AI Service Corps, expanded employer partnerships, and a dedicated Workforce Readiness Fund, New York City can create a more adaptable workforce ecosystem capable of navigating current and future technological transitions—whether that’s AI, quantum computing, or transformations we haven’t even imagined yet.
Searchlight: Thank you, Eli.