According to Business Insider, AI pioneer Geoffrey Hinton, often called the ‘godfather of AI,’ stated that computer science degrees “will remain valuable” despite the rise of agentic AI, arguing their value extends far beyond just coding. He bluntly noted that being a “competent mid-level programmer is not going to be a career for much longer, because AI can do that.” His view aligns with other tech leaders like OpenAI chairman Bret Taylor, who called a CS degree “extremely valuable” for learning systems thinking, and Google’s Sameer Samat, who said CS needs reframing around problem-solving. UC Berkeley professor Hany Farid added that the most exciting CS jobs now are at the intersection with other fields like medicine and finance, not just at major tech firms. Hinton also compared learning to code to learning Latin—a useful intellectual exercise even if you never speak it professionally.
The real value is thinking, not typing
Here’s the thing: Hinton and the others are making a crucial, and probably correct, distinction. For years, a CS degree was seen as a golden ticket to a high-paying software engineering job. You learned syntax, algorithms, and data structures, and you were set. But that’s the part AI is eating first. The actual act of translating a spec into lines of code? That’s becoming a commodity. What’s left, and what these experts are saying is the real enduring value, is the foundational mindset. It’s systems thinking, architecture, breaking down complex problems, and understanding computational trade-offs at a deep level. You can’t prompt-engineer your way to that. It requires the rigorous, often frustrating, foundational training a good CS program provides. It’s the difference between someone who can wire a house and someone who understands electrical engineering.
Where the jobs are going
This is where Farid’s point is so critical. The “usual suspects” in Silicon Valley are in a period of contraction and hyper-efficiency, and they’re the ones most aggressively deploying AI to replace routine coding work. So where does that leave graduates? The opportunity is spreading out. Computational biology, climate modeling, fintech, supply chain logistics, media analysis—you name the field, and there’s a massive, unsolved computational problem at its heart. These roles require a hybrid mind: deep domain knowledge and serious computational chops. A CS degree gives you that latter half of the equation in a way few other disciplines can. It’s less about becoming a cog in a big tech machine and more about becoming a power tool for innovation in any field. That’s a much more durable, and arguably more interesting, career path.
The Latin paradox and why it matters
Hinton’s comparison of coding to learning Latin is fascinating, and I think it’s more than just an academic quip. Learning a “dead” language like Latin forces you to understand the structure of language itself—grammar, logic, etymology—which then makes you better at understanding any language, including your own. It’s mental weightlifting. Coding, in this new AI-dominant era, is similar. By struggling through the logic, the debugging, the precise syntax, you internalize how machines think and how complex systems are built and can fail. Even if an AI writes the final code, you need that internalized understanding to evaluate it, direct it, and trust it. You can’t effectively manage or critique what you don’t fundamentally comprehend. So, should everyone learn to code? Maybe not. But for anyone wanting to work with advanced technology, not just use it, that foundational grind is probably non-negotiable. It’s the bedrock for the critical thinking Hinton says will separate the high-level engineers from the obsolete ones.
