Job requisitions for the “sexiest job of the century” have fallen dramatically. In this week’s Tech Tuesday, I’ll look at why and what it means for companies either currently involved with data science initiatives or looking to jump in soon.
As of July 2019, job postings for data scientists were down 43% compared to a year ago, whereas tech jobs were down 36% over the same period.
Why? Isn’t data science the one area that can help most right now by automating processes that generally require humans-in-the-loop?
This is a really good question but it turns out there are a couple of good reasons why job requisitions for data scientists are down.
Let’s have a look.
One reason why data scientist job requisitions are down is that hiring managers are (finally!) starting to use more appropriate titles for the particular type of data science they’re looking to hire for (read more on titles here).
For example, a person who mostly builds machine learning models could have the title Machine Learning Engineer (duh!). It’s about time, too. Data science has become far too broad a discipline for us to keep using “data scientist” when we mean any number of more-descriptive titles, such as: data engineering, or statistician.
With the broadening of titles, it’s no wonder that jobs looking for data scientists are decreasing; it’s because requisitions are calling for different titles.
Sudden Customer Preference Shifts
Another reason is that during a pandemic (or any huge shift in customer preferences) predictive models are nowhere near as effective. Shifts like this break predictive models that have never seen data that, for example, show everyone buying all the toilet paper on the planet in a one-week period!
Adapting to this kind of shift in customer preferences and behaviors therefore requires less machine learning and more real-time dashboarding to monitor what’s going on and make human-driven decisions.
Slower and less efficient? Yes. But with such radical never-before-seen shifts, it’s all we’ve got. I wrote about this just after the pandemic took hold.
Those two reasons – new titles and ineffective models – significantly contributed to the drop in available data science positions.
But all is not lost.
The pandemic Won’t Destroy Data Science
I believe that, now more than ever, it’s essential to increase automation with companies looking for ways to keep costs down as we adjust to the “new normal” (read: as new A.I. models learn from the recent past and can again predict/forecast future events).
For companies looking to leverage their data for the first time, I suggest starting with a short look-back period where dashboards or predictive models use more-recent data and in some cases avoid data from March through July, for example.
You may read more about this topic here.
I hope you’re well and getting through ok!
Google’s new Deep-Learning Model “BigBird” Improves Natural Language and Genomics Processing
Researchers at Google have developed a new deep-learning model called BigBird that allows Transformer neural networks to process sequences up to 8x longer than previously possible. Networks based on this model achieved new state-of-the-art performance levels on natural-language processing (NLP) and genomics tasks.
When Politicians Cite COVID-19 Statistics, They may be Wrong but This Doesn’t Mean the Numbers Are
During the COVID-19 crisis, a lot of meaning has been imbued by people who want the numbers to support their argument. Politicians, for example, love to throw around big numbers and pretty graphs. However, as pointed out by statistician Nate Silver; “Numbers have no way of speaking for themselves… we imbue them with meaning.” Hence, when politicians cite COVID-19 statistics, the numbers could be right but the ways in which they use them may be wrong.
Companies That Fully Absorb A.I. in Their Value-Producing Workflows by 2025 Are Predicted to Dominate the 2030 World Economy With +120% Cash Flow Growth
A.I.-driven organizations are using data and machine learning to solve their hardest problems and are reaping the rewards. Companies that fully absorb A.I. in their value-producing workflows by 2025 are predicted to dominate the 2030 world economy with +120% cash flow growth. However, this is easier said than done. Here’s a set of services that will simplify Machine Learning Operations (MLOps) for data scientists and ML engineers, so that your business can realize the value of A.I.