As new tools emerge based ChatGPT and other multi-modal language models, we are seeing a strong focus from startups and companies in trying to address software development tasks. One even claimed to have created an «AI Software Engineer» called Devin. And now Elon Musk now predicts that AI will be smarter than any human by next year.
Reasonably so, a lot of people in the industry is quite nervous about the future prospects of these tools and their jobs opportunities. And while it is futile to predict the future, I think it is important to draw from history as possible to have additional information on what to face.
Picking a Rhyme
There’s a saying that history does not repeat itself, but it rhymes. I’d argue that in this case, specifically about knowledge services, the offshoring boom of the 1990s is a nice example to study and draw from.
With the advent of strong telecommunication services globally and driven by costs differences, a lot of companies in the US and Europe started to outsource some part of their operations in other countries to run more efficiently. For the US, India and Ireland were the major recipients of these offshoring initiatives.
This put a lot of people nervous, as predictions made it clear that many jobs will be eventually offshored. This 2008 paper explains it much better and has quotes that could very much be applied to AI tooling, like this one:
One of the reasons the offshoring of knowledge services has gained public attention it the seeming suddenness of it all. It appears that, virtually overnight, a sea change has occurred: telephone calls to U.S. companies that were answered in Atlanta are suddenly answered in New Delhi, and the best computer programming firms are suddenly in Bangalore, not California
What happened then?
I’m going to narrow this down to software engineering, but the number of US jobs in technology nearly doubled in the US from the late 90s to the early 10s. Worker compensation and perks went up – think kombucha availability.
This was not the picture everywhere. Call centers and accounting declined for instance – but it was not the end of the US based call center either. Apple runs that in the US for instance.
Overall, both outsourced and local jobs grew over time.
Also, some offshoring failed. Cultural differences matter most than many expected, per the same paper.
What are the differences?
The one thing that is indeed different and we need to recognize as a bias in our thinking, is the fact that humans are terrible when dealing with exponential phenomena like technological progress. It does seem that Devin is almost a hoax, but we would be negligent if we were not open to the possibility that it might be the real deal sooner than we expect.
Localization, availability, flexibility – these are all different this time. Will those differences manifest themselves in fundamentally different tracks? Power and capability are also key, Elon is brilliant and might be right – but you also have to take into account that has a doubtful track record regarding promises.
This time, it is also very accessible for everyone to have a personal AI (outsourced team in this context) to enable them to do more with less.
What is similar?
Many things might be similar to the wave of offshoring, but I want to highlight that is always efficiency that drives this changes. All consumers want shorter wait times, better prices and overall better experiences – businesses just try to fulfill that within given boundaries.
Sometimes this comes down to cutting costs. Other times, it’s about improving value out of your cost structure. It’s in the latter that I see great opportunities to do new things, and have a lot of work to do.
In both cases, most businesses will start at their biggest costs to understand what can be done. And for most tech business, that means software engineering.
Juan Saavedra 