Click on the “Turing Artificial Intelligence” above, select “Star” public account
The AI insights you want, delivered first
Reprinted from 51CTO Technology Stack, for academic sharing only. Removed upon request if copyright infringement.
Editor | Yun Zhao
“The software industry is at quite a delicate inflection point!”
“In the coming years, it will determine whether we trade understanding for speed, or turn understanding into a new moat.”
In his blog post “Software Engineering for the Next Two Years,” Addy Osmani, the Google leader internally driving the integration of Gemini, Vertex AI, and the Agent Development Kit into developer workflows, did not rush to give a definitive answer.
Instead, he breaks down software engineering into 5 critical questions.
To judge what software engineering will look like in the next two years, one must first see these 5 key issues profoundly affecting the software engineering industry.
In summary, as AI disrupts the entire software industry, continuous learning and maintaining creativity are the only ways forward.
It was noted that Richard Seroter, Google Cloud Senior Director and Chief Evangelist, also shared this insightful article.
Starting today, if you aspire to work in the software industry, if you are a graduate, begin building your portfolio of creative AI-integrated projects. If you are a veteran, hone your architectural taste and cross-functional mapping skills!
Here are the key insights compiled.
Software is at a Delicate Inflection Point
1. Are Junior Engineers Still Needed?
Another possible future is the opposite.
Addy’s Recommendations
(1) For Junior Engineers:
(2) For Senior Engineers:
2. The Skills Question: Are Programmer Fundamentals Obsolete?
But the opposite is also possible.
Addy’s Recommendations:
For Junior Engineers:
For Senior Engineers:
3. The Role Question: System Orchestrator
Developers may devolve into “auditors of AI output,” or evolve into orchestrators of AI-driven systems. In either case, the value is no longer just in writing code.
In the most pessimistic scenario, the creative role of engineers is severely compressed, leaving only the work of reviewing and overseeing AI: checking if automatically generated code is compliant, secure, free of bias, and then approving it for release.
Some engineers already feel they are becoming “code janitors” rather than creators.
But a more appealing future is:
Engineers become high-level system orchestrators—deciding system structure, task allocation, how AI and software components collaborate. You don't write every line of code, but you define the system's melody.
In this “agentic” development environment, engineers are more like conductors than musicians.
Recommendations
Junior Engineers: Cultivate system perspective, communication skills, documentation skills
Senior Engineers: Assume roles in architecture, standards, ethics, and guidance
Evolve from coder to conductor
4. Specialists Will Decline, Generalists Will Rise
Soon, as original niche areas are disrupted by AI, the market will embrace a new model: the “versatile” or T-shaped developer.
5. The Education Question: Is a Bachelor's Degree Still Necessary?
Is a CS degree still a golden ticket? Addy believes there are two possibilities: one, attending university is still necessary but increasingly insufficient, requiring extra supplementary knowledge.
The second is more extreme: traditional education will be systematically replaced. Addy notes that programming bootcamps, online certifications, self-study portfolios, and corporate training academies are becoming mainstream. Simultaneously, more leading companies (like Google, IBM) have already removed degree requirements for some technical positions. Data shows that by 2024, nearly 45% of companies in the U.S. plan to drop bachelor's degree requirements for at least some roles.
This is a global issue: universities update slowly, while industry changes rapidly. More companies are eliminating hard degree requirements, shifting to skills-based hiring.
Recent graduates report they never learned about cloud computing, modern DevOps, or AI tools during their university studies. If universities provide low-relevance education while demanding significant time and money from job seekers, they risk being seen as expensive gatekeepers.
Degrees remain the default credential, but curricula lag behind rapidly changing needs. Slow course update cycles and cumbersome approval processes hinder progress. Both students and employers feel academia is disconnected from industry, with professors teaching theory or outdated practices that don't translate to practical job skills.
AI itself is becoming a personalized tutor, lowering the learning barrier.
Addy’s Recommendations
Beginners: Prove Ability with Projects, Portfolios, Certifications
If you're in a traditional CS program, don't rely solely on it. Actively supplement with real project experience: build a web app, contribute to open source, seek internships or cooperative projects.
For hot topics not covered in courses, learn via online platforms yourself. Obtain industry-recognized certifications (GCP, AWS, Azure, etc.) to signal to employers “I can hit the ground running.”
If you're self-taught or from a bootcamp, focus on your portfolio: at least one substantial, well-documented project.
Actively engage in developer communities: contribute to open source, write technical articles, attend meetups and conferences, build connections via LinkedIn. Seek endorsements from experienced developers.
Use AI as your personal tutor.
For senior developers and managers:
Advocate skills-first hiring, internal training, and mentorship programs.
Maintain interaction with universities and alternative programs: join advisory boards, give guest lectures, provide feedback on curriculum gaps.
The Only Constant: Change
Interestingly, for all these questions (except “specialist vs. generalist”), Addy provides two versions of the outlook. The peculiar thing is that these different versions are not mutually exclusive.
Addy states reality will be more complex, likely encompassing multiple versions simultaneously.
Some companies will reduce junior roles, others will expand hiring in new areas;
AI will automate vast amounts of repetitive coding, while raising the bar for human intervention;
A developer might review AI-generated code in the morning and design high-level architecture in the afternoon;
...
The only certainty is: Change itself is the constant.
Continuous learning, maintaining skepticism, holistic thinking, strengthening uniquely human judgment and creativity, and applying technology to solve real problems are the only ways to navigate the uncertainty.
The best way to predict the future is to engineer it yourself.
Reference Links:
https://addyosmani.com/blog/next-two-years/
https://www.youtube.com/watch?v=IMHneaMO-dg
Selected Articles:
1. That Stubborn French Old Man Left, Taking Silicon Valley's Last Idealism With Him
2. The World Model Debate Between Fei-Fei Li and LeCun
3. Fei-Fei Li's Heavyweight Essay: Beyond Language Models, Spatial Intelligence is AI's Next Decade