Indian economy is growing at a surprising speed with 9.7% in 2022 and 7.2% in 2023. Partly because of capital influx to the Indian economy from Silicon Valley's large tech firms for cheaper labor. In fact, it's not US firms alone that are outsourcing software engineers from India. Most western European companies have already relied on eastern European software engineers, and now they go to India. Japan and Korea follow the international trend. Meanwhile, souther and eastern European countries along with Russia provide world-class softwares that startles the global markets. Recently, Vietnam tries to get into this market. Although the language barrier is steep, which blocked Korean IT sector to be globalized, it is no longer a big surprise to see an IT company with Vietnamese IT backup.

Experts predict that within a few years, most low level IT jobs will be outsourced to above mentioned countries. And this is the very reason that I do not want to be deep in software engineering anymore. For most debugging, now ChatGPT is faster than me. For key functions and basic back-/front-end tasks, our Indian outsource team is way more cost-effective. I still have to do a lot by myself, but as David and Mc claimed that I do not need a pack of full-time software team for GIAI. Websites are managed by Keith with some dev support from India. For heavier tasks, we place an order with full spec sheet to the same Indian team. Initially, it cost me too much time for drawing the spec sheet and tell them the details again and again, (they really don't seem to read the details...) but as they get to see us a long-term contract, we feel that they pay more attention that noticeably help us to keep everything tight without much intervention.

Our internal discussion topic these days is whether data science jobs will also be outsourced to India or any other countries with lower wage in comparison to the U.S., western Europe, or developed parts of Asia. We all agree that, in the end, all expensive jobs will be replaced by cheaper wage, unless it is needed to be physically present. But let's think about the dynamics during this transitional period and how long will that be.

How many data scientists / research scientists do we have?

Many have claimed that data science tasks are 'democratized'. It is largely because of coding libraries from a variety of different sources that help us doing college level problem sets without too much background understanding in math/stat. But that's about the most they can get. As Keith often says, in his time (he's not that old, probably sometime in early 2010s) he had to code up each layer and node of neural network, but we hardly do that anymore, thanks to the libraries like PyTorch, Keras, TensorFlow....

We certainly have easier access, but that does not mean scientific tasks become easier and we become smarter. Most people running PyTorch algorithm are likely still at 'primitive' stage in terms of understanding what the libraries really do. At GIAI (or through SIAI), we often are emailed by a number of AI institutions from China and India, and respectively, some of them are suspicious. We don't go deep in talk, but it is not difficult for us to rule out them once we test them math/stat training levels.

In Europe, it is far less the case, but Keith tells us that such groups of software engineers who think they are experts in AI but with little to no understanding of math is widely visible in Asia. What we see here in UK/Eurozone (they are out!) is that many 'dreamers' go to math/stat undergrad and use the 3-year training to step up to respected AI/Data science programs like ICL's. The market is fairly well constructed when it comes to required training level, but as is for all STEM majors, students' survival rate is (not-so-)surprisingly very low.

Math alone is a tough major, and using math as a language of science in another discipline can be heavier challenge (although pure mathematicians will highly likely disagree with us).

In other words, we will unlikely be able to see huge influx of data scientists / research scientists in next 5 years. After all, this is why the UK government lowered its bar for immigration for most of AI-related tech researchers.

How hard will it be replaced?

No doubt that given the well-known challenges in STEM, supply of talent will always be limited. What's more concerning for us is what are the jobs/tasks that can/cannot be outsourced.

For example, BI(Business Intelligence) jobs does not require tremendous scientific knowledge. What's needed is some SQL commands, understanding of DB structures, and business intuitons, which is coined as 'market insight'. This is probably the lowest tier data science job in terms of required skillsets, but still needs tender and big brain. Most likely, this job should be given to the person of extreme domain knowledge. For an e-commerce firm, the BI should be well aware of consumer behavior in that country. As an example, given that Japanese/Korean/Chinese (especially later two) are highly a homogeneous society, they have shown strong tendancy to flock to identical products. In western hemisphere, such a trend event may mean something to analysts, but in Asia, it is likely the item's popularity will short-lived.

National level difference in consumer behavior still is a superficial example. There can be lots of local issues that remote analyst may fall into wrong conclusion due to limited information.

With above two examples in difficulties of STEM major and regional experties in BI jobs, we believe data science jobs, be it math heavy or not, will unlikely be replaced by cheaper outsourced labor in most developed countries.

How long the edge will stand?

Where there is no data, it is our rule that we do not make any prediction. The best I can do for this question is just to give you an educated guess.

For research scientists jobs, people from all sorts of STEM disciplines have filled vacancy. Here at GIAI, none of us are from the same college major. Our grad school topics are more distant. But we do share the common tongue, which is math/stat for science. With the continuous influx of talents from less financial rewarding disciplines (mostly in natural science), this indus