Given the current economy, with large companies laying off machine learning employees in droves, one may wonder if spending 4 years and over $80k in education is worth it. How long will it take to get a job when competing with hundreds of candidates for the few listed positions? What salary can I expect?
These days, many machine learning positions in US offer salaries well under $100k per year especially for beginners. Many still offer over $200k, but usually require specialized experience, and typically not what you would learn in school. Engineers are in higher demand than scientists, further casting doubts on the value of a PhD. There are multiple aspects and ways to look at this.
Type of Degree
People sometimes get a degree in a field that they love, not to get a job. If you want to climb Mount Everest, you need a significant amount of training. You may spend a lot of money to achieve your goal – your passion – and may not earn any money in return. The same is true with PhDs. Everyone knows that you are extremely unlikely to get a decent salary and tenure in Academia with a PhD, no matter how great you are. Yet the degree is designed exactly for that purpose, and it is still offered because of the demand. Indeed, it may hurt you in the job market to have a PhD and some applicants don’t include it in their resume. People do it for other reasons: perceived prestige, potential credibility and recognition (if you write books), and passion for research being three of them.
But what about a practical master’s degree, one involving internship, real programming, the creation of a portfolio, and everything to impress a potential employer? The timing is not great for those who just completed the degree, but excellent for those just enrolling. First, given the scarcity of jobs, spending time studying is more productive than looking for elusive jobs these days. But the economy for machine learning professionals will bounce back. I see more and more hiring managers looking to hire, even though you only hear about the layoffs. Some of the laid-off people do not have a real degree, but certificates or a data camp. While there are good providers, there are also many bad ones (those usually don’t even tell you who the instructors are). Hiring managers are more likely to require a real degree these days.
What Companies Usually Do
Some employees were overpaid or unable to sell the value that they produced to their boss or the decision makers. Companies never lower your pay to adjust to the new market: they eliminate jobs and at the same time recruit new (less expensive) hires and increase the pay of their best employees. It happens in cycles, now and then, each time causing or because of a recession, and they do it all at once, as in a game of musical chairs. Those with a good education and reasonable salary expectations, trying to get into this market, will eventually win.
Impact of ChatGPT
Will AI replace workers in the future, including those who develop AI? While there is no doubt that AI is capable of many things, there is also a lot of hype surrounding it. I can imagine many time-consuming and boring tasks like debugging and data cleaning being more and more automated, but we are still far from there. Even a basic problem such as scoring news from fake to real still needs to be solved. It can be solved without learning algorithms (more on this in a future article) but for now, many of the people working on this – including machine learning engineers – are not great at training these models with their own brain, let alone with an algorithm. Some companies like fake news because it’s click-bait and thus a source of revenue, but they will face competition from companies addressing the issue.
Hype, and What is Here to Stay
Talking about hype, vendors try to sell expensive products such as GAN or deep networks, and they find buyers because of the hype. In the classes that I offer, I had to include it as well, because that’s what many participants – themselves machine learning professionals working for insurance, health, or finance companies – want to hear about. In the end, the buyers are going to recognize that for their relatively simple needs, maybe cheaper solutions do just as well. The market is adjusting accordingly.
However, a bigger long-term issue impacting all jobs, not just data science, is the decline in population growth, eventually turning negative. This will open more and more positions in health care and related sectors serving an older population. Addressing climate issues is unlikely to lose steam any time soon.
About the Author
Vincent Granville is a pioneering data scientist and machine learning expert, founder of MLTechniques.com and co-founder of Data Science Central (acquired by TechTarget in 2020), former VC-funded executive, author and patent owner. Vincent’s past corporate experience includes Visa, Wells Fargo, eBay, NBC, Microsoft, CNET, InfoSpace. Vincent is also a former post-doc at Cambridge University, and the National Institute of Statistical Sciences (NISS).
Vincent published in Journal of Number Theory, Journal of the Royal Statistical Society (Series B), and IEEE Transactions on Pattern Analysis and Machine Intelligence. He is also the author of “Intuitive Machine Learning and Explainable AI”, available here. He lives in Washington state, and enjoys doing research on stochastic processes, dynamical systems, experimental math and probabilistic number theory.
