There’s plenty of talk about how “smart” machine learning systems are. The problem with these systems is that they’re really narrow – and let me explain what I mean by that term. Let’s take a healthcare environment. With today’s machine learning systems applied in healthcare, if you have just the right pictures of x-rays of a patient who potentially has cancer, the systems are as good as any in analyzing if there is evidence of cancer. The issue is that the systems are still fairly brittle in the sense that if the picture is just the wrong way or from the wrong angle, the system doesn’t work that well. Now, are we getting better all the time with this technology? Certainly. The pace of improvement is likely to be exponential.
According to Google’s Director of Engineering, Ray Kurzweil, the amount of computing dedicated to machine learning is doubling every quarter. We now have a number of effective machine learning algorithms that could potentially be used to solve real life problems but the vast majority of these problems haven’t even been touched yet. There is a lot of room for new machine learning projects to come about that are likely to be successful.
In terms of utilizing systems, there are quite a few schools that are now teaching a Masters in Data Sciences, Machine Learning and so on. Quite often, people who get those degrees find that most of the work associated with machine learning systems is not really the machine learning aspect but rather the data part. The question is: Where do you get the data?
What many companies are realizing is that they need to create an infrastructure that collects and holds greater amounts of data. Imagine that your business situation involves you having so much data that you can’t effectively transmit it. Obviously, you would need to figure out how you’re going to deal with that data. When we talk about the data associated with machine learning, it could require an environment that accepts one petabyte of data (a petabyte is equivalent to a million gigabytes).
Talman Advantage #1: Our Connections Run Deep
Why is it so critical that Roy Talman be the first recruiter you talk to? Before your resume is casually distributed to others, it’s important to understand how valuable it is to work with someone who brings a credible and highly reputable network of hiring managers. We’ve cultivated relationships with these managers for over 30 years – in fact, many of them were our candidates at one point.
But if you distribute a resume before we can leverage such connections to identify the best firm and role for you, it may be very difficult for us to help you further. So before you send out a resume, talk to Talman first!
As an illustration of this complexity, think about how much data your doctor has. In the past, if a doctor didn’t have any particular complex x-rays or other imaging for you, the data that they’d have on you would be very simple. Since the data would be relatively basic (i.e. your blood pressure and other fundamentals), the information would be little more than a couple megabytes. There’s not a lot of data involved.
However, that changes in a hurry when we strive to make sense out of the data that our wearables can collect. Wearables can collect 300 data points per second, 24 hours a day. That’s one heck of a lot more data, wouldn’t you say?
So What Do You Actually Do With That Data?
There’s data that you want to collect as well as the kind of data you need to consume as it happens – a perfect example of the latter being music. With music, typically you’ve got to hear it right now and to consume it, you need a Pandora, Spotify, etc. to help you decide if you like what you’re hearing. There’s no collection involved.
Conversely, Tesla is all about data collection. Their systems are trained to do autonomous driving, which they may not be fully doing yet, but nonetheless, the company is collecting a lot of data. Let’s say you’re on a highway and there’s an accident two cars ahead of you. The car recognizes this instance and slows you down in a hurry – all the while continually collecting that data and then deciding what data to upload to Tesla’s servers.
Now, if you were in the middle of a highway going 70 miles per hour with nobody around you at all, a company doesn’t really need this level of data because it’s not that usable. On the other hand, what if you’re going through a complex intersection with construction and obstructions? That’s a much different story. The company wants to collect all the data that it can so it can throw it into a simulation and learn from how you drove. The data also says a lot about how others would drive through the same environment so that the system can handle it.
The point: Once you realize that complex problems will require so much more data, you need to be in a prime position to absorb that data. Many systems don’t have that capacity right now. So do you wait until you have the complex problems that present data challenges and then enhance your machine learning system? Or do you enhance your system now in anticipation of the massive data that’s coming that it will need to handle?
No matter what you decide, what’s needed above all is the talent to work with a machine learning system and scale its technology. The talent you hire is the conduit that brings all the elements together for key stakeholders. How? By envisioning the amount of increased data coming down the pipe and what your machine learning system will need to handle that higher level of data in a superior way.
Talent of this nature isn’t found just anywhere. It’s identified through the kind of people who have specialized in technical recruiting for years and have a unique process all their own. Roy Talman & Associates is just that firm and we call our process The Talman Way. Discover how this time-tested, multi-faceted approach works to your advantage for sourcing the ideal candidate when you Talk To Talman First.