Heavy cloud, but no rain runs the refrain from a song by Sting, the British rock artist. Some years ago, when every information technology (IT) services company was waxing eloquent about its “cloud" prowess, I would begin my talks at conferences by quoting that line. As an outsourcing adviser, it was already evident to me that the real cloud deals were going to the likes of Amazon Web Services, Google and Microsoft Azure, none of which were traditional IT service providers.

Traditional providers did not really have competitive offerings in the marketplace, though there was much talk from them of “public" clouds, “private clouds, and “hybrid" clouds in the “Smac" environment. Smac—or social, mobile, analytics and cloud—was a catch-all buzzword in use by an industry that didn’t quite realize that it was social networks and mobile providers whom it wanted to serve who were instead going to own the cloud environment.

It also turned out that social networks could teach the IT services crowd a thing or two about Artificial Intelligence (AI). Many of the engagements that service providers now have with such firms are on the business process side. The horror that business process outsourcing (BPO) employees have to endure while vetting violent videos on social media platforms, for example, has been the subject of much negative news coverage.

The current buzzwords doing the rounds are “digital" and “big data". My opinion is that no one quite knows what digital is, and the fungibility of internal reporting allows companies to bundle whatever they want under this grouping. Big data or “data science" is now one of the hottest fields in the market. It consists, supposedly, of any discipline that can combine the fields of statistics and computer science.

Statistics provide data for behavioural (and other) prediction models or algorithms. Algorithms form the underpinning of AI and are linked with statistical concepts that have been around for aeons. Old-style statistical regression modelling and Box-Jenkins time series analyses look for causality and correlations in data and can predict outcomes over periods of time. The computer science side of big data allows for programmers to convert paper algorithms into computer code that can compute the result.

The huge availability of computing power today means that these predictions and behavioural associations can be made in fractions of a second. This allows your web search provider or social networking site to serve advertisements just nanoseconds after you have shown interest in a product. And since everything you do online is stored on a server somewhere forever, all your past online forays can be instantly called up by an algorithm to predict the next thing you are going to want to do.

One would expect that a hot new field like this would require long years of training and study. Interestingly however, an article in the Financial Times (on.ft.com/2zEjKQc) says that many people who move into the field of data science are able to do so without formal training at traditional institutes of higher learning. This is because demand for employees who possess these skills has grown much more rapidly than what traditional academic institutions have been able to meet. It seems that anyone with a passing familiarity with mathematics or statistics as well as the ability to write programs can easily pole-vault him- or herself into a high-paying career.

However, it turns out that newly minted “data scientists", the hottest commodities in the job market today, are quite disillusioned by the realities surrounding their chosen field. According to the article in the Financial Times, which cites a Stack Overflow survey of more than 64,000 respondents, data scientists and their close brethren “machine learning specialists" are right on top of the list of employees who are looking for new jobs.

Big data seems to fall in the same realm as the other buzzwords that Smac spawned. No one quite knows what it is. Dan Ariely, a professor of psychology and behavioural economics at Duke University, in a cutting assessment of the field, says, “Big data is like teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it." He should know a thing or two about data science. After all, data science is supposedly one method that firms can use to predict consumer behaviour, and Ariely is the author of several bestsellers in the field of behavioural economics such as Predictably Irrational and The Upside Of Irrationality.

In a world where the very definition of a field is fuzzy and many employees as well as employers are upstarts, there is a vast mismatch between expectations and reality. The data scientist walks into a job with high expectations of sitting at a desk and working out complex algorithms, while the reality is that most employers are at ground zero— where they need to find data which is stored in silos within their own organizations and convert this into a centralized repository which can be accessed by various computer systems. This means that most “data scientists" find themselves as little more than resources in an effort to categorize and pool data, or to run queries on data in existing silos, which are all boring tasks.

It looks like the field of data science is back to the days of the IT outsourcing boom, when employee attrition and job hopping were the operating realities that every firm in the sector had to live with.

Siddharth Pai is founder of Siana Capital, a venture fund management company focused on tech