Custom Upskilling Seen as Path for Closing AI Skills Gap
Discover more from AI in Business
Custom Upskilling Seen as Path for Closing AI Skills Gap
Tailoring to the individual a key to success; as is knowledge of AI frameworks and data literacy; and pointing to a new AI job title at the end of the reskilling path makes sense
By John P. Desmond, Editor, AI in Business
With the AI skills gap putting AI implementations at risk, experts are suggesting that upskilling could play an important role and that an upskilling path that can be tailored to the individual may present a strategic advantage.
Over 93 percent of companies in the US see AI as a priority, with projects planned or underway, but 51 percent of them report they lack the AI skills needed to put their plans into action, according to research done by Vanson Bourne market researchers for SnapLogic, a company offering a platform-as-a-service.
Some 300 IT leaders from organizations with more than 1,000 employees in the US and UK were surveyed about where they were in their AI journey and what was holding them back.
Companies reported being optimistic they could upskill their existing workforce to help fill the gap, by helping them learn the needed AI techniques. The top skills sought for the AI/ML team included: coding, program and software development skills (35 percent); understanding of data governance, security and ethics (34 percent); data visualization and analytical skills (33 percent); and education, with an advanced degree in a field related to AI/ML (27 percent)
To build the right internal team, 68 percent of respondents were interested in retraining and reskilling existing employees; 58 percent would try to recruit needed employees from other companies; .and 49 percent would recruit new university graduates.
Technology innovation with AI is in plentiful supply, but the ability to carry out the projects is lacking. “Companies do not have leaders and managers that can understand, evaluate, implement and oversee AI solutions,” stated Kjell Carlsson, head of data science strategy at Domino Data Lab, in an account in Toolbox. “We already have enough technology innovations to alter and create new industries,” he added. “Platforms, procedures, and people are the sources of the issues.”
For someone seeking to learn about AI programming, the authors recommended starting by understanding AI frameworks. These are the most popular:
Scikit-Learn
Theano
Apache Mahout
TensorFlow
PyTorch
Microsoft CNTK
Caffe
Next, an understanding of the role of data is critical to working with AI systems.
“Data literacy is required to understand the transformational impact of data in the organization,” stated Debashis Mohanty, VP of product management and product for Selector, a company offering an AIOps platform. The platform supports the search, formatting and analysis of data, freeing up time to allow developers to “focus their time on critical thinking and decision-making,” Mohanty stated.
“AI/ML is just the set of technologies to deal with data,” Mohanty stated, advising. “With that in mind, organizations should focus on data preparation, including cleaning and preprocessing the data with correct meta tags and identifying appropriate machine learning models to solve business problems.”
Customizable Approach to Upskilling Supported by Digital Platforms
Once decided on an upskilling path, the company would be well-advised to offer each employee a customizable path to learning AI skills, suggests a recent account in VentureBeat.
“For AI tools to be deployed at scale, those employees whose jobs involve interactions with AI systems need to understand how those systems work and what the constraints and limitations might be,” stated Junta Nakai, financial services and sustainability leader at Databricks, author of the account.
Reskilling these individuals may include how to interpret the results of the AI/ML models, or how to intervene with AI/ML experts when the results seem off.
Digital learning platforms have evolved to meet the market need. “The notion of a one-size-fits-all training program or that employees need to take significant time away from the office to attend courses is no longer relevant,” Nakai stated. He mentioned the following digital learning platforms: Skillsoft, Udacity and Udemy. Also, platforms including WalkMe can help employees learn complex software systems quickly, and Axonify can deliver 5- to 10-minute microlearning sessions to employees within their daily workflow, he suggested.
A survey from Deloitte found that 94 percent of employees would stay at a company if it would help them develop new skills, but only 15 percent were able to access learning opportunities directly related to their jobs.
Resources to help accelerate AI/ML training the author suggested:
Industry consortiums that support your industry; as an example FINOS (fintech open source consortium under the Linux Foundation) helps with the processing of financial data throughout the banking ecosystem;
Cloud service provider training and certification programs; AWS, Google Cloud and Microsoft offer ML training and certification programs for free or at subsidized prices.
Solution accelerators from major technology companies, that offer easy to deploy tools and support for a range of machine learning use cases that saves development time. Companies offering these include IBM, AWS, PwC and Databricks.
“At Databricks, our financial services solutions accelerators help companies capitalize on the open banking paradigm, providing free code and training that helps with front-to-back-end automation,” Nakai stated.
Investing in AI reskilling has cultural benefits for a company as well. “Investing in employees’ skills and knowledge can build a positive company culture and reduce turnover by boosting employees’ confidence and productivity,” Nakai stated. AI deskilling efforts can also help companies make progress on diversity, equity and inclusion goals by making learning more accessible to those who have faced barriers to higher education.
New AI-Related Job Titles Have Emerged
The end points on the reskilling path are evolving, with new job titles emerging to fit a range of professionals who understand the business and the problems it’s trying to solve with AI.
These include: data scientist, data engineer, ML engineer, data steward, domain expert, AI designer, product manager, AI strategist, chief AI officer and executive sponsor.
A combination of knowledge of the business domain and the technical underpinnings of the applicable AI constitutes a strength.
“The technologies and the tooling that we have available is skewing more and more toward enabling and empowering domain professionals, the business users, or the analytics professionals to take direct ownership of AI within companies,” stated Bradley Shimmin, chief analyst for AI platforms, analytics, and data management at consulting firm Omdia, in a recent account in CIO.
Data scientists process and analyze data, build machine learning (ML) models, and draw conclusions to improve ML models already in production. Ideally the data scientist is a mix of product analyst and business analyst, with machine learning knowledge, suggested Mark Eltsefon, a data scientist at TikTok.
“The main objective is to understand key metrics that have a major impact on business, gather data to analyze the possible bottlenecks, visualize different cohorts of users and metrics, and propose various solutions on how to increase these metrics, including making a prototype of the solution,” stated Eltsefon. When working on a new feature for TikTok users, he needs the data scientists to determine whether the feature benefits or alienates users.
Data engineers build and maintain systems that make up a data infrastructure, a crucial function in AI initiatives since data needs to be suitable for consumption for the results to be trustworthy. “Data engineers build data pipelines to collect and assemble data for downstream usage, and in a DevOps setting, they build pipelines to implement the infrastructure on which these data pipelines run,” stated Erik Gfesser, director and chief architect at Deloitte.
For ML and non-ML projects, this puts the data engineer’s work at the foundation. “For example, when implementing data pipelines in one of the public clouds, a data engineer needs first to write the scripts to spin up the necessary cloud services which provide the compute necessary to process ingested data,” stated Gfesser.
Read the source articles and information from SnapLogic, in Toolbox, in VentureBeat and in CIO.
(Write to the editor here.)