By using website you agree to our use of cookies as described in our cookie policy. Learn More

TDWI Upside - Where Data Means Business

How Democratization of AI Could Prevent the Great Resignation

To guard against the effect of the Great Resignation, enterprises need to reassess their training programs.

It’s no secret that the technology industry has been hit hard by the Great Resignation. In November of 2021, the U.S. hit a record of 4.5 million people in the U.S. who voluntarily left their jobs and, according to Gartner, only 29 percent of global IT workers have a “high intent” to remain in their current roles. Even with a looming recession, tech skills remain in high demand and competition for talent between big players like Google and Amazon means companies need to find ways to retain these workers.

For Further Reading:

Why Training Must Be Part of Your Comprehensive Security Compliance Plan 

How Symbolic AI Yields Cost Savings, Business Results

The Great Resignation: 10 Lessons for Data Teams

Instead of putting all their efforts into compensation and signing bonuses, companies can consider a different approach to IT employee retention -- training employees in new skills, particularly artificial intelligence. Especially for younger companies and startups that don’t have enough capital to fund lofty salaries, AI's democratization in the workplace provides an opportunity for employers to avoid the sting of resignation by improving or expanding the skills of existing employees. Such training can also help bolster job security for employees who may be fearful of company layoffs given the current economic climate.

This approach can create opportunities for career advancement, increased compensation, and job satisfaction, all within their current work environment. AI is penetrating all aspects of business. In addition to the products themselves, AI has become a key part of customer service, financial planning, human resources, and much more. If you have a recommendation engine, that’s AI. If you have a product search feature, that’s AI. Furthermore, low-code tools allow people throughout the organization to build their own AI systems: sales managers who want to find better prospects, warehouse managers who want to manage stock levels, and even IT managers who want to detect hostile attacks.

How does this affect IT staff? “The buck stops here,” as they say. All of those tools, whether developed by IT, purchased from a third party, or built by an employee using a low-code tool, will need to be maintained by AI, especially the “low code” applications developed in-house, which are likely to be “proofs of concepts” rather than finished, quality software. Every IT organization will need employees who can develop, maintain, and deploy these applications, and despite fears of a recession, AI talent is still in short supply.

The good news is you may not need to hire talent from outside your enterprise. Training the employees you already have will give them new opportunities and give you AI capabilities that already understand your business, your products, and your goals. Will your business suffer from the Great Resignation? Not if you give your current staff a good reason to stay with you.

Getting Started with AI Skills Training

To improve workers’ AI knowledge and skills, employers need to provide significant resources. The first step is training on subjects such as data collection and preparation. AutoML tools, such as Amazon Sagemaker, can help people throughout the organization build models; but AutoML doesn’t yet help users with data collection and preparation. The old phrase “Garbage in, garbage out” was never so true as it is with AI, and IT employees will need to ensure that quality data, not garbage, is going in.

Integrating AI products into existing deployment pipelines is another unsolved problem. AI applications don’t fit neatly into the tools and techniques used in most modern IT shops (variously known as DevOps, continuous integration, continuous deployment, and several other terms). We’ve had good tools for managing source code for a long time, but we don’t yet have good tools for managing and versioning training data and AI models. It’s not clear how an AI application, which may take hours or even days to train, fits into our current deployment pipelines. It’s not clear what testing means for software whose output isn’t necessarily deterministic. These are core IT issues; training IT staff in AI, so they can understand these problems and work toward solutions, will make them more valuable.

Training in AI doesn’t stop with learning a few algorithms and libraries. It requires understanding basic statistics, cloud computing, and other applied disciplines. Understanding issues about data quality and bias without a background in statistics is next to impossible. Without statistics, you can’t think about when a model has become outdated and is delivering sub-par results (a problem that doesn’t really exist for traditional software). We expect that most AI applications will be deployed in the cloud rather than in an on-premises data center. Our 2021 Data and AI Salary Survey showed that data professionals who obtained cloud certifications earned the highest salaries. Whether or not you’re working with AI, cloud skills are increasingly “table stakes” for IT professionals.

Making Training Effective

With remote and hybrid positions in strong demand, employers will need to take special note of how they deliver training. For enterprises that have fully embraced remote work or have distributed workforces, gathering even just a handful of employees in one physical space can be a challenge. In O’Reilly Media’s 2022 Cloud Salary Survey, research showed that over 31 percent of the respondents worked remotely 1-4 days per week, and almost two-thirds of the respondents worked remote full-time. This clearly isn't conducive to traditional onsite training.

Not only must training be online, it must be ongoing, particularly for those learning new skills and keeping up-to-date on AI advances. Most learning initiatives are structured in a rigid, standalone manner. Training is most effective when conducted within the flow of work -- in other words, learning that occurs on the job and is embedded into the employee’s workflow. Fostering learning in this manner enables employees to immediately apply new knowledge and skills to their work.

Today, plenty of technology exists that enables embedding learning into workflows, including learning modules within applications, cloud-connected mobile devices, and augmented reality solutions. Some online learning tools also have NLP-powered capabilities that can quickly skim an entire library of content to answer technical questions quickly and efficiently. Instead of reading an entire chapter of a textbook searching for an answer, employees can find quick, contextually relevant answers to challenging technical questions within seconds. These technologies infuse learning in small doses as part of employees’ workflows, helping guide workers in their tasks and providing added knowledge in moments of need -- which arguably can affect retention in a positive way.

A Final Word

When all is said and done, there are two certainties about the state of the technology world today -- multitudes of people are leaving their jobs and AI adoption is on the rise. Today’s employers are at a critical decision point. They can choose to neglect AI adoption and potentially lose exceptional, hungry-to-learn talent while they attempt to fill gaps in hiring needs amongst a landscape of increased competition, or they can choose to invest in technical upskilling for their existing workforce. Choosing the latter has the potential to prevent companies from bleeding talent while fostering opportunities for employees to advance their own personal career development and goals -- a win-win for everyone.


About the Author

Mike Loukides is vice president of content strategy for O'Reilly Media, Inc. He's edited many highly regarded books on technical subjects that don't involve Windows programming. He's particularly interested in programming languages, Unix, and what passes for Unix these days, AI, and system and network administration. Mike is the author of System Performance Tuning and a coauthor of Unix Power Tools and Ethics and Data Science. Most recently he's been writing about data and artificial intelligence, ethics, and the future of programming. Mike can be reached on Twitter and LinkedIn.

TDWI Membership

Accelerate Your Projects,
and Your Career

TDWI Members have access to exclusive research reports, publications, communities and training.

Individual, Student, and Team memberships available.