We have already seen that many professions and enterprise departments are looking to deploy AI or have already done so. However, a number of reports show that skills shortages are holding up AI development. What skills are enterprises looking for to deploy AI and what skills are lacking.

Earlier this year, tech talent marketplace Hired published its annual State of Software Engineers report, which analyzes in-demand coding languages and engineers. The research shows that at least for artificial intelligence, demand for computer vision engineers went up by 146% over the year while demand for machine learning engineers increased by 89%.

Earlier this month, we looked at the use of artificial intelligence in the workplace and found that despite the problems and expense of deploying this kind of technology, many professions and organizations were already using it in their digital workplace. The result, as the Hired report showed, is that salaries for these roles also increased dramatically last year, with machine learning engineers now representing the highest paid software engineering role in many cities.

Different AI Needs

The problem, as many companies are finding, is that every situation and every company has a different ask and need when it comes to AI/ML talent, according to Abhijeet Narvekar of the Houston-based recruitment and consulting FerVID group. He shared that their recent work was more focused on AI strategy and someone who is very forward thinking, someone who understood different models, particularly for those who have strategic roles and for someone who is setting direction for the company. “Several AI experts have had the opportunity to use NLP based models, computer vision is also getting used more in real use cases,” he said. “But when we talk about quantum computing or reinforcement learning or IBM Watson related use cases, not many have had the opportunity to work on it. And some of these ideas are still away from being used in real applications."

From an industry perspective working on recruiting people in oil and gas industry, for example, AI is used for more trend-based modeling while engineers use physics based modeling. There are only a few who have been able to use the models on both sides and make sense of the problem with a solution.

Finally, some AI leaders and data analytics leaders have focused their career more on technologies such as Hadoop, Cloud Based applications, Data Lake technologies. This is where we also must evaluate if the person is going to help more on the data science side than the real AI modeling side. Different skills are needed to come up with a solution that can be commercialized.

Related Article: How AI Enables Enterprises to Get the Most Out of Their Data

AI Research, AI Engineers

Peter Song is a machine learning (ML) engineer, who works with the Haki Review Mashup. He argues that to discuss AI skills in shortage you need to break it into two groups — AI researchers and engineers. The enterprises that experience the scarcity the most would have specific demands for the researchers and engineers. When enterprises plan to hire AI researchers, they aim at solving their specific problems by researching and developing new or customized algorithms.

Researchers can perform these tasks thanks to their deep understanding and years of research. Besides finding talented AI researchers, there are not enough number of AI researchers in the world. If we narrow our focus on AI experts who are looking for a job and who have a domain knowledge, the number will become far smaller.

The same story goes to AI engineers. These days, one of the sought-after skills is utilizing cloud-powered AI services such as AWS, Google Cloud, and IBM Watson. These companies already hired talented AI researchers and engineers.

With their strong human resources, they developed those AI services. Enterprises these days want to hire AI engineers who can use the right tool, which will allow faster development and deployment. In summary, enterprises are finding it difficult to hire AI researchers who can understand a domain problem and develop or customize algorithms. Also, there are not enough AI engineers who have experiences with the AI cloud platform.

Learning Opportunities

Related Article: AI vs. Algorithms: What's the Difference?

Signal Processing Needs

Sophie Summers is HR manager at Proprivacy, a UK-based web company that specializes in privacy and knowledge sharing. She says that one of the AI skills that enterprises are looking for right now is called Signal Processing, an electrical engineering subfield that deals with analyzing, modifying, and synthesizing any signal.

Almost every field contains the signal to transmit information, so it has a broader scope. Whatever industry fascinates you whether it is finance — interpreting financial data, medical — interpreting medical images, entertainment — signal processing in cameras or other electronics or you name it. Every industry includes information transmission in one way or another. “Signal process is the growth skill set that is needed in the modern economy industries,” she said. “Enterprises are striving to give the best possible result to their clients. For example, the medical industry is endeavoring to get the best possible results from the tests, and the film industry is flourishing by coming up with the most memorable films. They need the talent to produce better results.”

It will also revolutionize the research method. Enterprises could have the paper robots in the future that will summarize the previously done research helping researchers to proceed further instead of beating around the bush. Every industry is looking to speed-up the development process in their respective fields. Signal process is a key part of this. It enables organizations to interpret the result of millions of researches that have been carried out in the same field in the past.

The Problem With Specialization

The growing appetite among businesses today for the use the of AI is apparent and vital, but what's getting in the way of its development and reaching its true potential is the push in recent years towards hyper-specialization in business which has left skill gaps at every level, according to Tendü Yoğurtçu, CTO of Pearl River, New York-based Syncsort.

Most of the universities started offering studies more tuned for interdisciplinary skills required for AI and data science jobs only very recently. This is contributing to the skills gap because AI jobs require interdisciplinary mix of statistics, computer science and data modeling combined with domain expertise. “The skills gap is multiplied by the siloed practices in the organizations. Habits formed at the University level where students have been encouraged to only focus on honing their expertise based on their specific major, translate to the industry-wide issue of business siloing,” she said.

While unintentional in most cases, the siloing of information and skills across a company can be damaging to the successful deployment of new technology and growth of an organization. Exposing employees to the breadth of skills, diversifying the skill set with cross team pollination — whether in data science, statistics, data engineering, analytics or machine learning — will help companies close the skill gap in the rapidly developing field of AI.