N※N

Geoffrey Hinton: CS Degree Still Valuable, but Coding Skills Matter Most in 2025

  //science-technology.news-articles.net/content/2 .. uable-but-coding-skills-matter-most-in-2025.html
  Published in Science and Technology on Sunday, December 7th 2025 at 5:28 GMT by Business Insider

• 🞛 This publication is a summary or evaluation of another publication
• 🞛 This publication contains editorial commentary or bias from the source

2025-12-07 179 x 224 / 10730 Bytes

2025-12-07 224 x 224 / 10320 Bytes

2025-12-07 223 x 224 / 6109 Bytes

2025-12-07 224 x 224 / 10308 Bytes

Geoffrey Hinton: Why a CS degree is still a “good bet” – but learning to code matters more in 2025

When the world’s most‑renowned artificial‑intelligence pioneer speaks, his words ripple through academia, industry, and the broader conversation about what it means to be a “software engineer” in an era of generative AI. In a recent Business Insider piece, Geoffrey Hinton—often dubbed the “Godfather of AI” for his foundational work on back‑propagation and deep learning—delivered a candid assessment of the CS degree, the future of coding, and the role of AI in 2025. The article, peppered with links to Hinton’s own research, to interviews with tech leaders, and to tools that are reshaping how we write code, boils down to one clear message: a CS degree is a valuable foundation, but the ability to translate that foundation into working code remains the most important skill.

1. Hinton’s legacy and the context for his remarks

Hinton’s career has been intertwined with the rise of neural networks. After earning his PhD at the University of Edinburgh, he went on to be a professor at the University of Toronto, where he pioneered the algorithm that would become the backbone of today’s deep learning models. The article references a recent interview on The New York Times that highlighted his view that “the best machine‑learning researchers are the ones who can code.” Hinton’s own research page, linked from the article, shows how his early work on Deep Belief Networks (2006) was the first step that made modern AI systems possible.

2. The CS degree: a solid foundation, but not the whole story

Hinton’s remarks echo a sentiment that’s been brewing in the tech community: degrees still matter, but they’re only one piece of the puzzle. According to the article, Hinton believes CS degrees provide students with:

1. A rigorous theoretical grounding – knowledge of algorithms, data structures, complexity theory, and systems design.
2. Problem‑solving habits – the ability to break a problem into logical steps.
3. Credibility – a credential that still carries weight in hiring, especially for research or roles that demand formal education.

However, Hinton stresses that the application of that knowledge is where the real value lies. He references a link to a Harvard Business Review article that found companies still heavily weigh coding proficiency during interviews, often over formal education. In a 2024 survey published by LinkedIn Learning, 62 % of hiring managers cited “practical coding skills” as the top factor when hiring a junior engineer.

3. “Learning to code” as the “real skill” of 2025

While a CS degree is valuable, Hinton’s central thesis is that “learning to code” should be the primary focus for anyone entering the AI field today. He draws a comparison between programming languages and other forms of literacy: just as we learn to read before writing, we must learn to code before building AI systems. The article links to a GitHub repository that contains a set of curated “first‑practical‑projects” for AI beginners, showing that many of the best resources today emphasize hands‑on coding over abstract theory.

Hinton also acknowledges the rise of code‑generation tools—GitHub Copilot, ChatGPT, and other large language models—that can produce boilerplate code or even entire functions. He cautions that reliance on these tools without a deep understanding of the underlying logic can lead to brittle or insecure systems. The article includes a side‑by‑side demo from a YouTube tutorial that showcases how a novice can use Copilot to draft a neural‑network training loop, then modifies the loop to optimize performance.

4. The future of AI: more human oversight, less “black‑box” coding

Hinton’s interview, which the Business Insider piece cites, also touches on the ethical dimension of AI. As AI models become more capable of generating code, the responsibility of ensuring that the code is safe, unbiased, and maintainable shifts back to human developers. He references a recent paper from the MIT Technology Review (linked in the article) that highlights the importance of “human‑in‑the‑loop” systems for ensuring that AI‑generated code complies with security standards.

This stance dovetails with Hinton’s emphasis on coding. If a developer can understand and audit the code, they can spot potential vulnerabilities that an AI might inadvertently introduce. In other words, the future of AI will still need developers who are not just proficient coders, but also critical thinkers.

5. Practical take‑aways for students and professionals

The article distills several actionable points for readers:

• Start coding early. Even if you’re pursuing a CS degree, dedicate time to hands‑on projects—build a chatbot, train a small convolutional neural network, or replicate a classic RL algorithm.
• Leverage AI tools, but don’t replace them. Use Copilot for scaffolding, but always review the logic.
• Blend theory with practice. Understand the math behind back‑propagation, but also implement it from scratch.
• Pursue interdisciplinary learning. Many AI breakthroughs come from combining CS with fields like neuroscience or economics—consider double‑majoring or taking electives in those domains.
• Stay ethical. Learn the fundamentals of algorithmic fairness and security to responsibly deploy AI solutions.

6. How the industry is responding

Several companies referenced in the article are adjusting hiring practices in line with Hinton’s view. Google’s “AI Residency” program now includes a code‑proficiency test that is the same for all applicants, regardless of educational background. Microsoft has launched a new internal course called “Code‑First AI” that focuses on building practical models from the ground up. Meanwhile, DeepMind has partnered with the University of Cambridge to run a series of workshops that emphasize “low‑level coding skills” for aspiring AI researchers.

These initiatives underline a broader industry trend: while formal degrees still provide a useful baseline, companies are increasingly demanding demonstrable coding abilities, especially for roles that involve building production‑grade AI systems.

Final Thoughts

Geoffrey Hinton’s remarks, as captured by Business Insider, remind us that the core of AI—and of software engineering—remains the same: build things. A CS degree equips you with the language and the theory, but without the practical skill of turning that theory into working code, you’re missing a crucial piece of the puzzle. In a 2025 that will see generative models able to write large chunks of code, the human coder’s role shifts from “write code” to “interpret, validate, and extend AI‑generated solutions.” In short, Hinton’s message is clear: keep learning to code. It’s the skill that will carry you through today’s AI boom and into tomorrow’s uncharted territories.