AI Graduating Beyond “Side Project’ And Moving Up The Priority List

While there has been an issue with how expensive it is to build machine learning systems and “feed” them, there’s good news to be had in that we foresee the cost of building these systems and feeding them continually going down. And as that happens, their usefulness and applicability is only going to grow. We’re entering a new phase where it’s not about more hardware or lines on a chip of silicone, but instead more about the cost per recognizing a photo, cost per recognizing a moving image or cost of recognizing speech.

As machine learning becomes more and more like electricity, it will represent a growing percentage of the vast majority of available jobs. To what degree these are new jobs, old jobs, reformulated jobs or new tasks for people who are doing certain jobs isn’t abundantly clear.  

That said, what we are seeing is that, at the top of the hiring pyramid, perhaps five years ago, there was room for 1,000 people using machine learning algorithms, developing new ones, developing new hardware, etc. Because it was expensive and the knowledge was so expensive, it would take substantial time and effort to train people to do certain things. So, the bright, young sophomore was sitting there trying to decide: Do I go into machine learning or do I go into aerospace? And if going into aerospace was perceived as a better prospect career-wise, that’s what this person was going to do. However, if they heard how much money a classmate made or how much fun a friend had working in machine learning, perhaps he or she would spend the next two years learning about that.

There’s a real cost of getting into certain fields, which is a challenging decision people have to make. Some of this is influenced by how much people make in that field. The more they make, the more it will attract new people to enter that field, which might result in more and more new people not making as much money. Nonetheless, there will be more of them. This cumulative snowball effect is driven by the fact that the cost of deploying useful machine learning is continually dropping.

This means many more projects in machine learning make sense and many more people are required. There’s never a shortage of people, just that sometimes there are a whole bunch of projects that don’t get done because they are too expensive vis-a-vis their benefit. Still, if the price of implementation of these projects goes down, many more of them will get implemented and our society will receive a great benefit.

Talman Advantage #7: We Already Know Many People At The Top

The built-in advantage of being a more specialized recruiter for over three decades is that Roy Talman & Associates established many strong relationships with senior leaders in the C-suite and Director level. How do we truly leverage that? Prior to your interview, we can provide you terrific insight on the person’s background, the questions they’re likely to ask you and even a few clues into why prior candidates were likely rejected.

A recruiter that equips you with more information in advance of the interview? That just might make all the difference – if you talk to Talman first.


As the cost of AI comes down and becomes more accessible, these types of projects will move up on a company’s priority list. It will no longer be a “side gig” to work on machine learning for 20% of your time. It might become closer to 80% of your time.

As we’ve mentioned in previous posts, Masayoshi Son, the CEO of conglomerate SoftBank, essentially says that the entire business is a bet on machine learning. Google is a company that thinks of itself as a machine learning company too. These are some of the most successful and far-looking organizations out there, which tells us many more companies need to get on board.

In fact, we should be building out our machine learning functionality much faster than it took electricity to propagate through the business world. It took electricity about 40 years to accomplish this feat and the key problem with electricity was that the existing factories built for steam engines were built vertically. It took enough of a benefit for electricity to get to the forefront where people started knocking down those tall factories. I suspect that our pace of innovation of machine learning will be dramatically faster.

What About The Pace Of Incoming Talent?

The pace of the talent coming in is such that there may be a top position in an organization that pays them over $500,000 a year while an average salary comes in between $125,000 and $175,000, which is nothing to sneeze at.

As a result, that’s the level of compensation you’re looking at to induce people into an AI learning curve. The more people going through the learning curve, the more productive they are. What really needs to happen is that we shorten the learning curve.

Step one is moving from the present-day state of learning with a complicated, 200-line Python code to get any kind of functionality out of ML to a state where writing five lines of Python code gets useful results.

Step two is building the infrastructure to collect the data. This does take quite a while. Take the healthcare business, where getting medical data is very tricky. Due to HIPAA and other factors, you don’t have an ideal environment where there are systems to easily collect data, so you have to train people to use various technologies to capture data. At the moment, the typical doctor is not equipped to gain much use from large data, but that’s the point we’ll get to eventually. It’s the pace at which we learn that can be accelerated. Room for growth? Absolutely.

The bottom line here is that the cost of computing is still collapsing and that means that the quantity of work in computing is growing exponentially. It also means there will be a tremendous amount of projects moving forward that we couldn’t even compute before.

A career working in AI interests you. You’re committed to reaching a point where you’re called upon to get stronger results out of your code more efficiently as well as understand how to best apply collected data. What needs to happen from here is a straightforward conversation with one of the most highly specialized tech recruiters in the industry at Roy Talman & Associates. Together, we can make a plan that leverages the strengths you have and enhances those skills that demand improvement. Don’t leap into your next career move hoping for the best. Talk to Talman first. And give yourself the best foundation for ongoing success there is.