It looks like AI is poised to do to computer programming and possibly other types of so-called knowledge work what automation has done to other jobs, from the factory floor and warehouse to the checkout aisle and call center. In those industries, the end result of widespread automation is the elimination of countless roles – and in their place roles that either require relatively little skill and knowledge, or a lot, for workers at both ends of this spectrum. are rewarded accordingly.
In other words, software is eating the software industry.
Economists call this “skill-biased technological change”. This happens when technology makes skilled workers more productive, while taking over the complex and difficult parts of more repetitive tasks, making those workers easier to train and more interchangeable.
Now, AI is automating knowledge work, and its implications for the half of the American workforce employed in such jobs are profound. It is true that these white-collar jobs have been evolving over the decades as technology has improved, but the elimination of medium-skilled jobs is bound to accelerate as AI becomes institutionalized in the workplace. This new technology has the potential to reshuffle the deck of winners and losers in America’s increasingly polarized economy.
Coding was the beginning of the generative AI boom that has captured the world’s attention since the release of OpenAI’s ChatGPT in November. While generative AI is generally thought of as a tool for making text, images, and even videos look as though humans created them, it can also be used by programmers to generate code. And also being done automatically to test it. Microsoft made GitHub Copilot – a programming tool that uses OpenAI’s technology – widely available in June 2022, five months before OpenAI made its ChatGPT bot public.
Tech layoffs due to macroeconomic trends over the past year have coincided with the advent of truly useful AI for coding. For many young coders, the timing is unfortunate. Data from workforce-analysis company Revelio Labs shows companies furloughing their newest employees, and in 2023, software engineers will represent the largest share of people laid off by tech companies. Meanwhile, the few technical job opportunities that remain are being picked up by the more experienced software engineers still in demand.
The rise of the kind of AI that could make more experienced programmers more productive could mean that companies that have paused hiring are waiting longer to start hiring again, And startups trying to survive the current downturn in venture-capital investing are choosing not to move forward with current placements. Schemes.
“If I were an investor, and my companies were looking at hiring hundreds of engineers, I would say, well, maybe instead you can use AI to be more productive,” Stack Overflow said. says Prashant Chandrasekhar, the company’s chief executive. Maintains a user-generated repository of questions and answers for programmers.
Stack Overflow’s latest survey of nearly 90,000 programmers, conducted in May, found that 70% are already using, or planning to use, AI tools in their work. Of those, a third of programmers said the main reason they use such tools is because the technology makes them more productive.
Many tech companies have pledged to continue growth despite the layoffs. Since the spring, investors have been pouring in stocks considered to be at the forefront of the AI boom, such as Nvidia, Advanced Micro Devices, Apple and Meta Platforms.
On the one hand, these companies will need to continue hiring developers with skills in building AI. These engineers are currently the highest paid and most sought after, says Josh Brenner, CEO of tech-job marketplace Hired, founded in 2013.
On the other hand, there is BrainTrust.
The worker-owned talent network represents approximately 360,000 freelance developers, and its more than 1,500 clients include Nike, Meta and Google. CEO Adam Jackson aims to double BrainTrust’s revenue over the next two years without adding a single employee.
The company is building its own internal AI, which is trained on all of its job postings, developer resumes, and successful matches between the two. Jackson says the system will handle much of the work of matching companies looking for developers with those freelancers. It will use the same techniques that make other generative AI possible, such as ChatGPT.
“That productivity creates more value for our network of engineers,” he says. “It’s largely because of AI.”
Other industry- and company-specific AI are likely to be created and used in a variety of workplaces.
Whenever you’re building an online marketplace to connect two parties — whether it’s home buyers and sellers, or ride-hailing drivers and riders — AI can improve your company’s productivity without increasing the overall number of people, says Jackson. One way to increase it.
Many experienced developers I spoke to expressed skepticism about the ability of AI coding tools to handle the most essential tasks of programming, including designing solutions to complex problems and understanding existing libraries of code in those companies. Those who have been building their systems for years. or even decades.
That said, people who are already using such systems think that they can eliminate the need for some of the tasks that are currently typically delegated to inexperienced and early career programmers.
“Now simple issues can be tackled relatively easily,” says Jerome Chu, head of development at Diffbot, an AI-enabled, business-focused search startup. “I might consider hiring more senior people because of this.”
This has happened before also. Yossi Sheffey, an engineering professor at the Massachusetts Institute of Technology whose latest book is on the future of work, warns that the job- and salary-polarizing effects of automation on an industry could disrupt the normal ladder of hiring and development. In manufacturing, this means that some countries, such as Germany, have created a dual education system that leads people to undertake both apprenticeships and university studies within the usual four years of college.
He says automation is finally coming to knowledge-related tasks – in the form of AI – may necessitate a similar rethink on employment and education for early-career white-collar workers.
“One of the main challenges of the future is how to hire junior people who don’t yet have experience working on machines that don’t work,” says Dr. Sheffey.
In Stack Overflow’s survey of developers, experienced and junior coders said they use AI code-completion tools, such as GitHub CoPilot, ChatGPT, and the many startups based on it, such as Bito, that differ. While experienced programmers said their number one reason for using such tools was productivity, junior programmers said their primary use was for education—that is, absorbing what the AI already knew.
While there are early signs that AI is already disrupting the market for some developers while increasing the value of others, overall, technologically driven changes in the demand for workers are occurring slowly.
When discussing the speed at which automation can put people out of work, a historical example that Dr. Sheffey likes to cite is the telephone network.
In 1892, says Dr. Sheffey, the first automatic telephone exchange was invented. By 1930, there were still 235,000 telephone exchange operators in America. Even today, a small number of human telephone exchange operators remain.
One reason we tend to overestimate the speed of technological change in the short term is that we fail to understand the forces that stand in the way of that technology becoming widespread. Those forces may include unions, government regulation, and social acceptance. With today’s AI, Dr. Sheffey sees history repeating itself. For example, there are already proposals for a new federal agency to oversee the rollout of new technology platforms, including AI.
No matter how fast the pace of AI development, the adoption of the technology is still up to humans – and over the course of individual human lives, we change slowly, if at all.
That said, the double bind many pre-career developers currently find themselves in is a cautionary tale for all of us. If AI disrupts a sector at the same time that workers in it face other challenges, regardless of what historians say, the impact of automation on jobs and those who occupy them could be exponential.
Write to Christopher Mims at firstname.lastname@example.org
For more WSJ Technology analysis, reviews, advice and headlines, sign up for our weekly newsletter.