OpenAI’s Codex product lead, Alexander Embiricos, says that computer science degrees remain valuable, but students need to adapt to the changing landscape of software development shaped by AI tools.
Speaking[1] on a recent episode of the A16z podcast,[2] Embiricos[3] warned that students risk falling behind if they study computer science in environments that prohibit the use of AI. He emphasized that the future will demand more software creation and, consequently, more software engineers. However, succeeding in that environment requires fluency with AI tools.
Embiricos highlighted that applicants who demonstrate tangible outcomes, such as building projects and sharing them publicly, stand out during hiring. He noted that OpenAI, while not hiring many entry-level positions, prioritizes candidates who have actively created software or solutions.
The increasing role of AI in software development has prompted colleges to reconsider their computer science curricula. Thomas Cortina, associate dean and professor at Carnegie Mellon University, said that students are realizing gaps in understanding after relying on AI to complete assignments. This “reset” highlights the importance of learning the underlying principles of coding alongside using AI to accelerate development.
OpenAI chairman Bret Taylor has echoed the sentiment[4] (recently) that a computer science degree is still essential for cultivating systems thinking. Researcher Szymon Sidor also encourages students to continue learning to code in high school to build foundational skills.
When asked how he would reshape a computer science program, Embiricos suggested combining hands-on learning with AI integration. He proposes a mix of manual exercises to understand core principles and practical projects that produce tangible outcomes. This approach, he argues, helps students maintain mental flexibility while preparing them for a software landscape increasingly influenced by AI.
A related summary from Embiricos’ social media highlights two key points: first, software development is far from obsolete, as AI tools like Codex can increase software creation and demand for engineers. Second, pairing traditional coding skills with AI fluency enables developers to complete projects that might otherwise never have been attempted, either by accelerating initial development or achieving near-complete outputs that motivate human follow-through.
Embiricos’ guidance points to a broader trend in computer science education: learning to collaborate with AI is becoming as important as learning to code itself. Students who embrace AI tools while grounding themselves in fundamental principles are likely to have an advantage as software development continues to expand.
Image: Catgirlmutant / Unsplash / Notes: This post was edited/created using GenAI tools.
Read next: Is Learning to Code Still Worthwhile in an AI-Driven Industry?[6]
References
- ^ Speaking (x.com)
- ^ episode of the A16z podcast, (a16z.simplecast.com)
- ^ Embiricos (x.com)
- ^ echoed the sentiment (www.youtube.com)
- ^ Catgirlmutant / Unsplash (unsplash.com)
- ^ Is Learning to Code Still Worthwhile in an AI-Driven Industry? (www.digitalinformationworld.com)