Data science has been a stardust activity since around 2014 when interest in the term started to climb from a ten-year flatline (scoring six or seven on Google Trend’s interest over time index) to reach a maximum 100 rating in September of 2019. For those involved and able to appropriate the term to their skills set, that stardust has been golden with salaries reflecting rising demand and limited supply.
The last five years have also given data scientists a golden period during which to hone those skills and demonstrate their value. Optimisation has been the watchword for this function, helping brands to do more, faster and better by leveraging big data to find predictive patterns.
All of this is about to change as we head into a down economy - in the UK due to Brexit, in the US because of its trade wars and in China because of the slowdown in demand from the UK and US, for example. What works in a growth economy seldom has traction when consumer behaviours switch from discretionary spending to defensive saving.
In the business realm, the next five years will be all about rationalisation - working out how to achieve the same with less. Some of this can look the same as optimisation, such as when stretching a marketing budget through constant incremental gains, and there are areas of activity that have already been under the data science lens for some time. Supply chain management and logistics are a good example, such as where marginal savings from optimised delivery routes can add up to million pound savings.
But rationalisation also means cutting resources - in particular the human kind. The core data science areas of automation and artificial intelligence are essentially about replacing tasks carried out by people with ones carried out by machines. With the sparkle of growth and progress on them, these have looked to date like building the future.
Once cuts in headcount really start to hit home, especially as alternative employment gets harder to find, however, the mood is likely to change. Data scientists may no longer be heralded as the architects of a new world in which we all work less and profit more, but as the servants of a ruthless drive for efficiency which has little concern about workers and their personal futures. This is when the US view of a tech-driven society will collide with the European concern for social cohesion and a degree of control.
Two things will hit data scientists hard. The first is that their last five years of experience spent optimising processes may not be relevant when confronted with a new economic reality and set of behaviours. Existing data sets reflecting growth (however moderate) are not predictive of slowing or shrinking demand. New models and ways of thinking about problems will need to be learned and applied at speed - not all current practitioners will make the grade.
The second is that data scientists will stop being viewed in a positive light within their organisations, but instead will be identified with cuts, job losses, disrupted careers and personal trauma. It is their models applied to finance, HR and resourcing that will decide who stays and who goes, after all. How will that sit with individuals who from Maths GCSE through to their PhD and beyond have only known praise and positivity?
My guess is that we will see something of a rush for the doors as data scientists look to recover their stardust, many of them literally so by returning to academia and astrophyics. Public sector opportunities will look more appealing since, despite being even more focused on cuts and savings, it keeps practitioners further away from the people their decisions affect.
They will also encounter the paradox of data science - that the very automation and AI it is tasked with building is also eating the jobs of data scientists. From ETL to model building, self-learning models to automated decision-making, machines are taking the place of experts. In manufacturing, it’s called built-in obsolescence. Chances are that nobody mentions this during the data science Masters and PhD course which have sprung up in the last five years.
Perhaps the first indicator of weakening demand for their skills is the nine-point drop in Google searches between September and October. As businesses start to batten down the hatches and ditch expensive overheads, they could soon be telling their data scientists, “so long and thanks for all the differential calculus”.