Without question, one of the most popular topics in our business right now is how to get a job working with artificial intelligence / machine learning.
At one level, we have clients that want cutting-edge, top-of-the-line researchers. Fair enough. However, there is one issue we’ve seen arise in lot of people who received their advanced degrees in artificial intelligence between 2010 and 2016: They might have problems in applying their knowledge in a world that moves very fast, where they are asked to apply well-known ML approaches rather than inventing new ones.
Believe it or not, a lot of the fundamental ideas behind artificial intelligence, particularly the mathematical models, were developed in the 1960s. Unfortunately, there were two critical challenges facing us at the time: 1) Computing power was very expensive and 2) partly because of number one, there was not enough data.
Fast forward to 2011 where we had an explosion of data availability. Thanks to video games and other popular applications creating demand, a whole industry of GPUs came about to produce very elaborate, very complex graphics. The real breakthrough in 2011 came when two graduate assistants working with computer scientist Geoffrey Hinton figured out how to code GPUs to look at pictures. That was the beginning of a trend that would basically break the dam wide open where suddenly, you could look at millions of available pictures and have a way of processing them, even identifying each picture by image.
Since then, we’ve seen the amount of computing dedicated to machine learning doubling every quarter (per Ray Kurzweil). Every vendor has an offering in machine learning and is clearly working on a chip to do it better.
So while people realized how to code GPUs in 2011, that was merely the beginning. Today, you have people such as Elon Musk saying he doesn’t need to use GPUs from years ago because he’s just produced a chip that is much better for machine learning in an autonomous driving setting. He has designed a chip with the target of it being the most customized, energy-efficient chip in the history of machine learning. Musk describes his chips as 10 times better than the previous generation of chips that we were using until now. Still, Tesla does not slow down. They are continuously evolving these chips.
So we’re getting to the point where more machine learning projects are becoming feasible. It’s cost-efficient to get data and cheap enough to build the intelligence into machine learning applications so that you can start using it – typically not instead of, but in addition to people. Expect that we will have some kind of machine learning in practically everything we touch. And that probably will mean that we will need many more people working on it.
Talman Advantage #6: The Technical Expertise Clients Highly Respect
How many account managers within a recruiting firm have technical PhDs and MSs? Not many. Yet, you’ll find several of them at Roy Talman & Associates, which our clients in the technological space have come to highly respect over the course of 30+ years. No wonder they respond quickly in real time. And when we suggest the creation of a new position just for you, they seriously consider our suggestion at a minimum and frequently call us to discuss further.
See yourself represented from a higher place right from the very beginning. Talk to Talman first.
How PhDs Can Shift
Gears And Bridge The Gap
Here’s the challenge: At the moment, a variety of people who work in machine learning believe that they are going to be developing new, advanced machine-learning algorithms. A good idea in theory. In reality, that is often not the case. Most challenges lie in figuring out how to apply the vast number of very sophisticated, freely available algorithms that already exist. For example, you really don’t need to reinvent an open-source library like TensorFlow – you need to learn how to use it.
Then you have the other part of the equation, which is the data. Beyond collecting data, which takes significant work, you need to be able to clean the data to make it usable. This calls for a lot of software engineers, data engineers and project engineers to figure that aspect out.
The good news? For the PhD who understands how to best apply existing complex algorithms in machine learning and arrive at usable, clean data, the career opportunities should be plentiful.
The need for talent and capacity for growth in the area of AI is enormous, even though the capability to fill these positions has not been fully realized.
In fact, the opportunities are so vast that we’re even seeing the potential to apply machine learning to animal populations, not just of the human variety. Here’s a story to illustrate: Not long ago, there was an amusing story in the Wall Street Journal on how hard it is to build facial recognition system for pigs and cows. Pigs and cows! Why would we even want to bother with such a thing?
Well, consider this: If a farmer can have better facial recognition per animal, they can keep track of them even better, they can determine which animal is sick, they can see whether their particular activity is ample enough and more. It turns out that if you can do facial recognition on a herd of cows, then you can start monitoring the herd at the level of each individual animal. So as funny as it sounds to worry about a new kind of facial recognition for pigs and cows, it offers a new level of productivity and efficiency. But when the person building the application was interviewed, he complained how hard it is to get those animals to stand still or to face him in the right profile.
Whereas microchipping can be expensive, if all you need to do is take a picture, that in and of itself is extremely cost-efficient. Facial recognition represents the beginning of being able to collect data and once you do that, you might be able to recognize all kinds of other activities and traits. Even if your cows are not very cooperative.
If we stand to have many more people working in AI, how can you increase the opportunity that you’re one of them? Talk to Talman first. Roy Talman & Associates can help prepare you for the types of projects that might be suited to your background without forcing you into a less-than-ideal fit. Instead, we’ll tell you what you need to work on to elevate your skill set and position yourself to adapt comfortably into the growing areas of machine learning you can thrive within.