[7] An Optimistic Take: AI's Augmentation of Software Professionals

No doubt you've heard it before: "AI will make programmers obsolete" or "AI will take your job." But is this narrative accurate? News stories like Cognition AI's Devin, albeit found to be categorically a hoax, have nonetheless fueled fears among software professionals about the future of their careers. However, the most current look at the state of AI in software development suggests a more nuanced and hopeful future.

The AI Job Impact Narrative

The narrative that AI will replace programmers is not new. This idea has pervaded movies, books, and news stories for decades dating back to the beginnings of the first wave of AI research in the 1950s following the Dartmouth Conference in 1956 (the event where the term "artificial intelligence" was first coined). In 1984, the Australian newspaper The Age published an article entitled "Researchers coming with some provocative hi-tech findings", in which the author details the opinions of two Stanford University professors, Henry Levin and Russell Rumberger, who studied employment statistics and concluded that "high technology is no panacea for unemployment... high technology will reduce the skills required by workers in the next decade because machines will do the work. Creativity and innovation by many individual workers will suffer." The article goes on to quote another Stanford computer science professor, Edward Feigenbaum, a prominent artificial intelligence researcher who at the time said:

 "One of the worst pieces of advice you can give youngsters today is to
 become a computer programmer. Automation of software development will
 take place in this decade and the 1990s."

Obviously, the likes of Bill Gates, Linus Torvalds, and countless other software professionals who have made their careers in the decades since would beg to differ. More recently, in 2018, McKinsey & Company, a global management consulting firm, predicted that 800 million jobs would be lost to automation by 2030. In countless media outlets since, the narrative has been repeated and amplified.

The proliferation of this narrative has led to widespread fear and uncertainty among software professionals. In 2019, Forbes published an article that cites a ZipRecruiter survey finding that "One in five job seekers (one in three for those between the ages of 18 and 22) fear they will one day lose their job to AI." Fast forward to 2023, and a representative Gallup web poll of over 5,400 Americans found that 75% of respondents believe AI will reduce the total number of jobs in the United States in the next 10 years.

However, we don't have to accept this narrative based solely on projections and surveys—- the data is in and the results are more optimistic than you might think.

The Data is In

Annually, Stanford University's Human-Centered Artificial Intelligence (HAI) Institute releases a report called the AI Index. The AI Index is a comprehensive report that tracks and measures the progress of AI research and development across a variety of metrics. Here are some of the key takeaways from the 2024 report:

• AI makes workers more productive and leads to higher quality work.

"AI has been shown to enable workers to complete tasks more quickly and produce higher quality work. A meta-review by Microsoft, which aggregated studies comparing the performance of workers using Microsoft Copilot or GitHub’s Copilot LLM-based productivity-enhancing tools with those who did not, found that Copilot users completed tasks in 26% to 73% less time than their counterparts without AI access" (AI Index, 2024, p. 272)

• AI is creating new jobs.

"The United States once again led the world in terms of total AI private investment. In 2023, the $67.2 billion invested in the United States was roughly 8.7 times greater than the amount invested in the next highest country, China ($7.8 billion), and 17.8 times the amount invested in the United Kingdom ($3.8 billion)" (AI Index, 2024, p. 247)

"Consistent with trends in private investment, the United States leads all regions with 897 new AI companies, followed by China with 122, and the United Kingdom with 104." (AI Index, 2024, p. 251)

A great recent blog post from Builder.io, a company that builds AI-powered tools for developers, also provides a comprehensive look at the current state of AI in the software industry and also seeks to dispel the myth that AI will replace software professionals. The post states:

By automating routine tasks and augmenting human capabilities,
AI tools are ushering in a new golden age of software engineering.

Developers who adapt and leverage these tools will find themselves
more productive, more creative, and more valuable than ever before.
The future of software engineering isn't about competing with AI —
it's about orchestrating it to build amazing things.

The U.S. Bureau of Labor Statistics also provides a positive outlook for the future of software development jobs, stating, "Overall employment in computer and information technology occupations is projected to grow much faster than the average for all occupations from 2022 to 2032. About 377,500 openings are projected each year, on average, in these occupations due to employment growth and the need to replace workers who leave the occupations permanently."

The U.S. Bureau of Labor Statistics data for employees in "custom computer programming services" from the Current Employment Statistics survey also shows a steady increase in employment over the past 10 years:

Historical Precedent: The Rise of Spreadsheets

In the late 1970s, the accounting industry was on the brink of a technological revolution. Before the advent of electronic spreadsheets, the accounting industry relied heavily on manual calculations and paper-based record-keeping. Accountants, clerks, and bookkeepers spent countless hours poring over ledgers, meticulously ensuring every number added up. This labor-intensive process was the norm for decades.

Then came VisiCalc, the first spreadsheet program, launched in 1979. At first, many accountants were skeptical, even fearful. The idea that a piece of software could perform calculations and automate tasks traditionally done by hand was both revolutionary and terrifying. They worried about job security and the potential devaluation of their skills.

In the book Leading Projects with Data, author Marcus Glowasz writes:

Nowadays, no accountant would imagine doing their work by hand, but back
then [in the late 1970s and early 1980s], many accountants were hesitant,
fearing that these tools would make their jobs obsolete. And, to some
extent, their fears were well-founded. Since 1980, around 400,000 accounting
clerks have become obsolete as spreadsheets took over their role. But
spreadsheets made accounting more effective and easier, which resulted in
600,000 new jobs in the accounting industry. (Glowasz, 2022, p. 138-139)

The introduction of spreadsheets did indeed lead to the automation of many tasks previously done manually. However, it also created new opportunities and job roles that didn't exist before. By no longer having to spend hours on calculations, accountants could focus on higher-level analysis, strategy, and decision-making, leading to the emergent and lucrative roles of financial analysts, controllers, and CFOs. The fear of job loss was replaced by a newfound optimism and excitement for the future.

Opinion: Enthusiasm for the Future

Personally, one of the more exciting pieces of news surrounding AI in the past year was the announcement of a new project from GitHub called GitHub Copilot Workspace. This project aims to allow developers to write their tasks in natural language and have Copilot generate the code systematically for them. As the blog post explains, the workflow for Copilot Workspace is as follows:

1. Write a task: Describe the task you want to accomplish in natural language in a GitHub issue or new GitHub repository.
2. Workspace creates the plan: Copilot Workspace generates a step-by-step plan to accomplish the task. The plan is entirely editable and can be tweaked to fit your needs.
3. Execute the plan: Copilot Workspace generates the code for each step in the plan. You can run the code in the workspace and edit it as needed until you are satisfied with the results. You can then create a pull request to merge the code into your project.

The implications of this project are vast in my mind. Using this tool drastically lowers the barrier to entry for new developers-- you don't even need to use an IDE as GitHub already has a web-based code editor. As the article states, "With Copilot Workspace we will empower more experienced developers to operate as systems thinkers, and materially lower the barrier of entry for who can build software... we are accelerating to a near future where one billion people on GitHub will control a machine just as easily as they ride a bicycle." If writing code becomes as easy as writing a natural language description of the task, the possibilities for what can be built are endless. In the same vein as VisiCalc and the rise of spreadsheets, imagine the lucrative new job roles that could emerge as programming's skill focus shifts from low-level syntax to high-level problem-solving and creativity.

As of the time of writing, Copilot Workspace is still only in limited technical preview, but projects like this serve to heighten my optimism and enthusiasm for the years to come in the software industry and the augmentation of our jobs as software professionals to greater creative freedom and productivity.

References

• 1984 The Age Article on Artificial Intelligence
• McKinsey & Company 2018 AI Automation Post
• Forbes 2019 AI Job Impact Article
• 2023 Gallup Poll on AI Job Impact
• Stanford HAI's 2024 AI Index Report
• GitHub Copilot Workspace Announcement
• Builder.io Blog Post on AI Job Impact
• U.S. Bureau of Labor Statistics
• Leading Projects with Data