Want to get paid like an NBA star, but don’t have Harden’s step back three? This week we look into how people are learning about machine learning and postulate the future of jobs.
When I was applying to college, I wanted to go to school for computational biology, otherwise known as bioinformatics. Deciding where to apply to school was easy because at the time there were only about 6 schools that offered it. After talking with the director of the program at RIT, I was especially excited about his interest in supercomputing and its application to simulating protein folding and thought RIT was a good fit.
To my surprise, it turns out high school seniors pick colleges for reasons other than their schools commitment to supercomputing and machine learning. When you look at the top college preference for high school seniors by state, it’s a strange coincidence that a lot of SEC schools show up…
Revenge of the Nerds
But all of a sudden, people love learning about machine learning. In fact, according to the Coursera blog, the most popular course of 2017 was Andrew Ng’s Machine Learning course, followed by its Deep Learning course at number two. People can’t get enough Deep Learning because the previous course on Deep Learning/Neural Networks ranks #9 on their most popular list.
Just this week, Google just released over 15 hours of videos, full with tutorials and questions about getting up to speed with AI/ML. While it is aimed at developers, Google also mentions people with no machine learning experience can learn from the course. Google also offers a Deep Learning program on Udacity, aimed at getting people familiar with their Tensorflow software.
Finally, last September, over 1,000 Stanford students registered for Andrew Ng’s Machine Learning course, and considering only 7,000 people attend Stanford as an undergrad, it’s insane. And just so we come full circle, according to StatClass, there are now almost 100 different universities that offer degrees in Bioinformatics.
Why do people love it?
So, why has there been an explosion of people flocking to this area? I like to think it’s because programming is fun, doing matrix multiplication is fun, and running algorithms on super cloud computing hardware is fun. It’s really some of the most fun you can have.
Money, so they say
A shortage of available AI/ML talent in the field is leading to serious salaries payouts from tech giants. According to the NYTimes, “both Ph.D.s fresh out of school and people with less education and just a few years of experience, can be paid from $300,000 to $500,000 a year or more in salary and company stock.” The article continues to say that those at the top of their field tend to get paid like athletes…
Everyone is trying to cash on this trend. According to Indeed, while the amount of job postings with the term “machine learning” have doubled since 2014, people seeking for a job with the term machine learning has tripled. Colleges have also started realizing this and programs offering Masters in Data Science or Analytics have increased 10 fold since 2012, with most of these programs offering at least an intro to machine learning and programming.
But a bigger issue looms
I have a theory that maybe people are learning about machine learning because of its impact on destroying jobs. Recent estimates about job automation number into the hundreds of millions by 2030, with 30% of Americans at risk.
While there is certainly a shortage of AI/ML experts, it’s hard to say what the actual demand for this expertise is. A recent Tencent study argues that the number of AI/ML practitioners needed is in the millions, but certainly not the hundreds of millions that AI/ML will replace. Not everyone will be paid half a million dollars from learning about machine learning.
While trying to figure out what the future of work will be is beyond the scope of a weekly Cloudy, it’s certainly something to start thinking about. We don’t know what jobs AI/ML automation will bring and trying to AI/ML proof your career is challenging.
Back to the classroom
We can’t expect everyone to go back to school, given the large amount of jobs that may be be automated. The Coursera/Udemy’s of the world will most likely play a large role in helping people retool, given their flexible and scalable nature. Given the size of the retraining pool, education could see one of the largest changes in history. I wouldn’t be surprised if in several years these offerings took a hard look at what skills people will need over the next ten years, and heavily pivot towards those.