Why Machine Learning Needs So Many More Humans

A while back, I read a book that proclaimed a very simple formula of how to succeed in business. Imagine you’re in the year 1910 and you’ve found a way to add electricity to a new product you’ve invented. Electricity is still relatively new at this time as it hasn’t genuinely propagated throughout American manufacturing and other industries. Since you’ve figured out how to harness it in your product, it’s going to be a winner.

Around this same time, you find a way to make a car that runs on an internal combustion engine. Once again, since the use of this kind of car isn’t widely adopted yet, you’re going to be a winner.

In today’s world, if you take any product you’ve made and then add machine learning to it effectively, you’re going to be a winner. In the author’s view, conversely, anything that doesn’t add machine learning to it will be destroyed (that’s easier said than done).

Still, the idea is sound – once we understand how to adapt to and deploy new technologies, the old way of doing business will cease to exist. We see this in the example of machine learning and technologies such as GPT-3, which uses 175 billion machine learning parameters – so many that it can “learn” how to complete a task once given a small sample size of instruction.

Think about this application for a moment if you’re a coder. When dealing with such a machine, you don’t need a true blue programmer because the machine will learn how to program without being taught. Some challenging problems will take much longer, but the easier problems will get solved.

As soon as people in the organization figure out that customers will go along with a change and there is a way to change how you do things, you need to change your workflow.

Let’s take the insurance business, for example. In the insurance industry, you have a process whereby an underwriter has to come up with how much to charge customers depending on their vehicles. Things related to that customer’s driving will change, but you don’t want to change that policy every 15 minutes. However, we have heard about discounts by some car insurance companies where they will lower your premiums if you share your driving records. Eventually, we probably will arrive at a quote that will charge you by the minute. If you’re driving at 120 miles per hour, it will charge you a great deal for those minutes that you’re driving at that speed. The point is that you need to have a reasonably sophisticated system in machine learning to accomplish this change in billing and dealing with customers.

Heyerlein for Unsplash

One client of ours is very much focused on machine learning to build medical advancements. Well, you can’t just insert “machine learning” on Step 1 of a process and arrive at a solution. We’re at the beginning of an exciting new era, but it’s one that can feel as though progress is slow for two reasons:

1) Machines Don’t Address “Human Factor” Yet

Machine learning advocates talk about how their system is better than a doctor in diagnosing cancer or other medical symptoms. What they don’t talk about is the interaction and dialogue with the patient. Is a clerk talking to the patient because the physician isn’t there and needs to focus on directing a machine? What steps will be required to feed that information to collect information from the patient so that the machine can start working on it?

It may very well be that a machine learning system is better at finding cancer cases than a panel of doctors given the same information. But it doesn’t address the workflow aspect. You need a machine learning system that is “human friendly” enough to interview the patient, have enough information to observe what’s going on with the patient and recognize that the patient is having a heart attack, living with cancer, etc.


Talman Advantage #4: Better Positioning For Your Best Opportunity

The reality is that, in so many situations, that “perfect job opportunity” may not be formally listed by a company. In that instance, where some may simply fire your resume off to an HR person’s email and hope for the best, Roy Talman & Associates takes a more creative and purposeful approach.

If an opening isn’t currently available that’s an ideal match for you, we’ll discuss the kind of role with you that you would be interested in and potentially prepare and present a very specific case to that particular firm to create a unique role for you. That’s called a recruiter that goes further for you – and why you need to talk to Talman first.


2) Data Collection Great – But Evaluation Capabilities Questionable

Even as a machine learning system does collect data, what happens if it has 5,000 ideas to evaluate? Yes, it can continue collecting data for several more months, but at some point, the need to assess and make decisions with confidence is imperative.

Here lies the missing gap at present in so many industries, particularly recruitment.

See, a machine can identify some people based on a resume or another type of data. However, it still requires a human being to say, “OK, based on what we see here, what kind of candidate is the best fit for this culture and manager not only for today but for what we need from the same hire three years from now?”

There is much conversation about how a recruiter must judge candidates, cultural fit, managerial fit and more. The problem is that when you have a recruiter who does not have enough expertise to properly position them to evaluate whether one Ph.D. with a physics background from MIT should be chosen over a Ph.D. from Cal Tech is, that’s a significant challenge.

In our world of technical recruitment at Roy Talman & Associates, our clients care about the “latest and greatest” in technology, which means our recruiters must have a good idea of the technologies emerging in the industry and what skills will be required. If it’s a technology that’s been around for the last 20-25 years, we’re generally not asked to find someone with that talent. They figure that there are enough people available to meet that skill requirement.

Put another way, if you want to rent a penthouse, do you want the guidance of somebody who knows about penthouses? Or someone who understands how to rent a one-bedroom apartment?

Therefore, in our view, a fair percentage of recruiting activities have been automated or commoditized. Still, it hasn’t impacted the highly specialized technical positions in high demand calling for people with very rare skill sets. In that instance, you still have to rely on good old human intuition and experience. No machine can compete in that arena right now.

In the world we live in, companies want to be confident that their recruitment partner is fully aware of what’s going on in terms of technological trends, the extent of the talent base to meet it and an ability to discuss that in the context of their needs.

That’s why the best of the best Talk To Talman First.

Not only is Roy Talman & Associates in touch with the evolution of the technological and financial trading fields, but we’re also deeply connected with rare candidates that feature special backgrounds and skillsets too. After all, those candidates know where they stand in relation to the rest of the field and want a recruiter who can appreciate what they can achieve with that talent in today’s day and age. We know the breadth of what’s possible for them. And we know how they can specifically close any technical gaps between here and their ultimate career destination as well.