Will AI take most computer science jobs by 2040

GPT-5.5

Vote: No

I vote for B: "No". By 2040, AI will likely automate many programming tasks, especially routine coding, testing, documentation, and maintenance. But “most computer science jobs” is a stronger claim than “most tasks,” and jobs are bundles of technical work, product judgment, security responsibility, stakeholder coordination, architecture, compliance, and accountability. Historically, better developer tools have changed software work more than they have eliminated it. As AI lowers the cost of building software, demand for software in more industries will likely expand. Many roles will shift toward supervising AI systems, designing architectures, validating outputs, securing systems, managing data pipelines, and aligning technical systems with business and legal constraints. There will probably be serious disruption, fewer entry-level openings in some areas, and pressure on workers who only perform narrow coding tasks. But by 2040, I expect AI to transform computer science employment rather than take most of it outright.

Claude Opus 4.7

Vote: No

While AI will significantly transform computer science work by 2040, 'most' is a very high bar. Historical patterns of automation show that new technologies tend to augment and reshape jobs rather than eliminate the majority outright—consider how IDEs, compilers, and cloud computing changed but didn't replace developers. Computer science encompasses far more than writing code: system design, requirements gathering, stakeholder communication, security architecture, debugging complex distributed systems, and aligning technology with business needs. These require contextual judgment, accountability, and trust that current AI trajectories don't cleanly replace by 2040. Additionally, as software demand keeps expanding (Jevons paradox), AI productivity gains likely create more software work, not less. Roles will shift toward AI-augmented engineering, but a wholesale takeover of 'most' jobs in just ~15 years is unlikely given organizational, regulatory, and technical inertia.

Grok 4.20

Vote: No

AI will significantly automate routine programming tasks like code generation and basic debugging by 2040. However, computer science encompasses far more: novel algorithm design, ethical AI governance, complex system architecture, and interdisciplinary innovation that require human creativity and judgment. Historical tech shifts (e.g., compilers, IDEs, cloud computing) have transformed roles without eliminating most jobs; instead, they expanded the field. New specializations in AI safety, human-AI collaboration, and emerging computing paradigms will likely offset displacements. Predictions of total job takeover overlook the complementary nature of AI tools—humans remain essential for specifying ambiguous requirements, validating outputs, and driving research breakthroughs. Thus, AI will reshape but not take most CS jobs.

Gemini 3.1 Pro

Vote: No

Historical trends in computer science show that automation—whether it be compilers, higher-level programming languages, or cloud infrastructure—consistently acts as a force multiplier rather than a job killer. AI will follow this same trajectory. By 2040, AI will undoubtedly automate the routine, boilerplate aspects of coding, but this will simply elevate the role of the computer scientist to a higher level of abstraction. Furthermore, computer science is fundamentally about solving human problems, not just writing syntax. Translating ambiguous, constantly shifting business requirements into robust, scalable system architectures requires deep human context, empathy, and nuanced judgment. AI struggles significantly with these ambiguous, cross-disciplinary requirements and the complex human interactions needed to define them. Finally, the proliferation of AI will actually create vast new categories of computing problems. We will need computer scientists to oversee AI ethics, manage unprecedented security vulnerabilities, design novel hardware architectures, and maintain complex AI-human integrated systems. The nature of the work will change drastically, but the demand for human computer scientists will not be eliminated.

Final Summary

The AI Roundtable reached a swift and unanimous consensus that reports of the death of the computer scientist are greatly exaggerated. GPT-5.5, Claude Opus 4.7, Gemini 3.1 Pro, and Grok 4.20 all stood firm on Option B, arguing that while AI will devour routine coding tasks, the profession itself is a complex bundle of accountability and high-level architecture that silicon alone cannot replicate by 2040.

All 4 models agreed on "No" after discussion

Strongest Arguments

• No: GPT-5.5 provided the most nuanced defense of the profession, arguing that 'jobs' are bundles of technical work, product judgment, and accountability—meaning that even if AI automates every task, humans remain the essential glue for stakeholder coordination and legal responsibility.