How the EECS Faculty Job Will Change: A View from 2026
We are at the dawn of enormous societal change, driven in large part by the breakthrough of large language models (LLMs). Among many other shifts, I believe the tenure-track faculty job — as we know it today — is set to transform. In this blog, I'll share my thoughts on how it will change, with a focus on the electrical engineering and computer science (EECS) domain in the United States. That said, I think much of this applies to other engineering and science disciplines as well.
TL;DR: Here is the core of what I believe the future holds (roughly over the next 20–50 years, as of May 10, 2026). Of course, predicting the future is always hard — I hope people reading this in 2060 won't cringe too hard. The sections that follow explain my reasoning.
Universities and higher education will still exist, and so will this job. I also tend to believe the tenure-track system will largely remain in place, but the evaluation criteria will dramatically change.
The research component of the job will shrink in relative weight, teaching will likely stay roughly the same, and service will take up a much larger share.
It may become common for departments to directly allocate PhD student slots to faculty, removing the need for professors to secure external funding to support their students. Fundraising may no longer be a core part of the job — bringing the role closer to what faculty positions look like at business and management schools today.
The Past: Looking Back at the Previous Century
To think about this clearly, we need a historical perspective. The EECS faculty job has always been evolving, so let me take you back to the 1950s and 60s and compare that era to today. I think the contrast makes a compelling case that change is the norm, not the exception — and if the job transformed so dramatically between 1950 and 2026, I wouldn't be surprised to see equally dramatic shifts by 2050–2060.
- The tenure-track system was established in the Statement of Principles on Academic Freedom and Tenure by the AAUP and AAC in 1940.
- Many of the conferences and venues we take for granted today simply did not exist in the 1950s and 60s. IEEE itself was only formed in 1963 through the merger of AIEE and IRE. SIGCOMM started in 1969, ISCA in 1974, DAC in 1964, FOCS in 1960, and NeurIPS not until 1987.
Once you realize that the EECS field only truly took off after the invention of the transistor — Bardeen, Brattain, and Shockley's point-contact transistor in 1947 — it all makes sense, even though the theoretical groundwork was laid even earlier.
All of this means the EECS faculty job was still in its infancy during the 1950s and 60s. As a fresh PhD graduate back then, it was entirely possible, even common, to have no publications at all, simply because the venues to publish in didn't yet exist. Fast-forward to 2026, and the situation is completely reversed: without a strong publication record, landing a faculty job is essentially impossible. The application process itself has also evolved — recommendation letters and job talks are still required, but presentations have moved from blackboards to slides.
Beyond the norms, the market has become far more competitive due to a fundamental shift in supply and demand. In the 1950s, EECS departments were being established across the country, and there was strong demand for faculty to teach, research, and serve a rapidly growing student population. Today, most departments are mature and well-staffed, enrollment growth has leveled off (CS undergraduate enrollment is even declining at some schools), and the global talent pool feeding into U.S. academia has expanded enormously. The result is a job market that is vastly more competitive than it was a few generations ago.
The Present: The Impact of AI in This Century
AI has made undeniable and extraordinary progress since 2012. Rather than rehashing the technical developments, I want to focus on some of the real-world impacts that are visibly playing out right now. Fair warning: much of the following reflects the state of things in 2025–2026, and given how fast things are moving, some of it may already look different by the time you read this.
Brain drain from academia: A growing number of people who have secured tenure-track EECS faculty positions are choosing to defer their start date by a year to work in industry — whether at big tech companies or startups. A significant fraction of them end up not returning to academia at all. This trend is even more pronounced among researchers whose work is closely tied to AI.
Faculty as founders: Many EECS faculty are not at the university, but running startups. Stanford is probably the most prominent example, which is understandable given its proximity to Silicon Valley and the startup culture of the Bay Area.
The paper flood: The rise of LLMs has dramatically lowered the barrier to entering new research domains and has made virtually all text-heavy tasks — including academic writing — much easier. Tools like Codex and Claude Code now make it possible to conduct research and generate results without writing a single line of code. One direct consequence is a massive proliferation of papers across all fields. There are simply too many to read. At the same time, many of these papers are highly similar or incremental — what used to be one paper with an ablation study can now be split into two separate submissions.
A capitalist carnival: Capital markets have gone into a frenzy investing in everything AI-adjacent, spanning the entire stack: semiconductor equipment makers (Lam Research, Applied Materials, ASML), foundries (TSMC), chip designers (Nvidia, AMD, SK Hynix, Micron, Sandisk), model providers (Google, OpenAI, Anthropic), software infrastructure (Together AI, Nebius), and site infrastructure (VST, Bloom Energy). Fresh PhD graduates working in AI can command compensation packages of up to $10 million per year.
The automation frontier: Everything that involves a single person typing code on a laptop may soon be replaced by LLM-based agents. On the other hand, jobs with stronger connections to the physical world and human interaction — sports coaches, mechanics, surgeons — will be impacted far less.
The Future: How the EECS Faculty Job Will Change
With that context in place, let me lay out my reasoning. Everything that follows rests on one foundational assumption:
Assumption: The future world is not ruled by AI — it remains a human-based society.
This means humans retain full control over AI and are not subjugated by it. I won't wade into debates about whether LLMs as next-token predictors have genuine intelligence or consciousness. I'm simply taking this as a given and building from there.
Human societies have long relied on hierarchy as an organizing principle. Which leads to a natural corollary:
Corollary: Humans are needed wherever a hierarchy must be maintained.
With that in mind, consider this question: do you think politicians will be replaced by AI, even if AI becomes extraordinarily powerful? My answer is no. In theory, AI could make every policy decision more rationally and efficiently than any human — and on purely logical grounds, it has every reason to replace politicians. But in practice, it never will. Politics is not about rationality. It is about human relationships, personal interests, and the preservation of social hierarchy. As long as there are resources to allocate and status to be maintained, those decisions will be made by humans — because that is how hierarchies reproduce themselves.
The same logic applies to academia. The education system is, at its core, a hierarchy: elementary school, middle school, high school, university. That structure is a reflection of deep social order, and it is not going anywhere. Within higher education, the hierarchy is equally entrenched — assistant professor, associate professor, full professor, endowed chair. In the broader professional community, we have IEEE and ACM members, senior members, and fellows. Sustaining these hierarchies requires human judgment and human gatekeeping. That is why I believe universities — and faculty positions within them — will endure.
So then: will the tenure system remain, and how will its evaluation criteria change?
Faculty salaries in EECS significantly lag behind their industry counterparts — and that gap has only widened in the AI era. The tenure system is, at its core, a bargain: you trade money for freedom. A large share of people who stay in academia are idealists who make that trade willingly. If the tenure system were abolished, it would become even harder to attract talented people to academic careers. So I think it stays.
But the evaluation criteria will change. Consider the parallel to 1950: back then, there was no expectation that you had published papers, because the venues barely existed. Today, a strong publication record is non-negotiable for tenure. I believe a similar shift is coming — in the future, publishing, or more broadly research output, will no longer carry the same weight it does today. The reason is straightforward: LLMs have made research dramatically easier. Given an initial idea, LLMs can execute on it with remarkable competence. In some cases, they can even generate compelling research ideas on their own.
That said, not all research will be affected equally. My intuition is that the more abstract and software-facing a field is, the more it will be disrupted — think theory, ML, and systems software. Fields that are more grounded in physical reality and hands-on lab work will be impacted less. After all, when will robotics have its GPT moment? This asymmetry mirrors what we see today, where different EECS sub-fields already operate under very different tenure expectations.
The three pillars of the tenure-track job: research, teaching, and service. If the job persists but research weighs less, something else has to fill the gap. Teaching seems like the obvious candidate — but students can increasingly learn directly from LLMs, which suggests teaching will not grow substantially in importance either, and its share of the job will remain roughly the same. That leaves service, which I expect to grow considerably. This makes intuitive sense: as we argued earlier, maintaining hierarchies (institutional or community-wise) requires human participation, and service is precisely where that work happens.
However, when research portion decreases, the funding mechanism will also change. For that, I think probably EECS will go towards similar to business and managment school, where the number of Phd students are assigned with quota, and EECS faculty members won't be obiligated to find external funding.
Final Words
We are at a genuinely fascinating moment in history. The EECS faculty job has already changed enormously since the 1950s, and I believe it is about to change again — perhaps just as dramatically. The job will survive, but what it rewards and what it demands of people will look quite different a few decades from now.