Workers can begin implementing on-the-job generative AI training with these steps.

• It is essential for companies to have employees skilled in artificial intelligence as more companies plan to incorporate generative AI into their operations and workforce.
• To ensure that workers remain competitive and up-to-date, on-the-job AI training is becoming increasingly crucial.
• The key for employees is to comprehend how to prompt AI platforms to generate the required outputs, not to understand every technical detail of generative AI.

One way to ensure workers have strong generative AI skills is through on-the-job training.

If you don't adapt to AI, someone else will replace you in your job, according to John Blackmon, the chief AI officer of ELB Learning, whose products are used by companies like Google and Microsoft to train their workforces.

Understanding what's happening behind the scenes is the difference between proficiency and expertise, according to Blackmon.

The key for the general workforce is to know how to prompt platforms to generate the outputs needed, while programmers and engineers must be experts in generative AI.

Blackmon stated that prompting is the new way to communicate, and it's organizations that have a significant interest in employees being comfortable generating AI inputs.

Regardless of whether their leaders allow it, employees will use generative AI, and new hires are already equipped with ChatGPT," said Bryan Kirschner, vice president of strategy at DataStax. "What do they get when they start working?

The report from TalentLMS indicates that 58% of HR managers plan to implement upskilling and reskilling initiatives to address the skills gap caused by AI. Some of the key concepts in this effort include:

Crawl, walk, and then run with AI

"Think about crawl, walk, run in training, but prioritize crawl early on," advised Kirschner.

Generative AI requires a copilot, just as an employee needs a manager or coach, according to Kirschner.

While employees may initially be unaware of the potential of generative AI, they can develop their skills and become innovative through training. Kirschner emphasized the importance of employees developing a perspective and being able to think creatively about how to apply generative AI to meet customer needs.

DataStax's Kirschner team developed an AI maturity model that assesses a company's AI sophistication. The model consists of four key threads: context, culture, architecture, and trust. Additionally, the model highlights the most effective ways to implement generative AI from the outset until employees become proficient.

The maturity model for privacy by design progresses to continuous learning and real-time adaptation, while the culture thread starts with a values-driven organization and leads to variable compensation for ethical decisions and regulator engagement. Architecture involves automated code generation and testing, and trust is influenced by internal and external transparency.

Adopt AI responsibly and slowly

Upskilling for generative AI requires clear goals, knowing the target audience, and precise objectives, according to Blackmon.

To ensure responsible use of generative AI, many companies establish internal working groups comprising stakeholders from diverse departments. These groups help inform approved AI use cases and guide employees through the training and adoption process.

Brainchild Communications' HR director, Hasnain Malik, advised companies to select a limited group of users and focus on a specific set of features. He suggested gradually increasing the trial until the company feels confident in rolling it out to all users.

BigID, a data security software company, published a course titled "How to Accelerate AI Initiatives." The course provided guidance on how organizations can adopt AI responsibly, including measures to prevent data leaks and non-compliance, as well as reducing risks associated with using generative AI. The course emphasized the importance of classifying all data that goes into a large language model (LLM), as Blackmon stated, "If it's junk going in, it's junk going out."

The BigID course recommends implementing automatic flags for policy violations when sensitive or regulated data is misplaced. These flags should encompass both existing and new policies that address specific risks associated with generative AI, as long as they do not conflict with any existing policies.

Carrie Hoffman, a partner and labor and employment attorney with Foley & Lardner, emphasized that generative AI policies must be implemented in conjunction with ensuring compliance.

Foley & Lardner patent attorney Shabbi Khan stated that striking a balance between being permissive and monitoring AI-related activities is crucial. If employees are too restricted, they may use AI on their personal devices. To ensure employees understand and follow guidelines, a concise best practices guide should be provided, such as requiring employees to turn off training mode and avoid inputting confidential information.

Keeping a guard up around AI

It is essential to be cautious when using generative AI and verify the accuracy of outputs, as hallucinations or unconfirmed outputs can occur, although they are rare.

As the models improve, it becomes more difficult to detect hallucinations, leading people to assume that the outputs are accurate, despite the presence of mistakes.

The assumption that machines are always right is known as automation bias, and it's crucial to maintain critical oversight.

Kirschner, a seasoned tech strategist, stated, "In the early days of mobile, it was considered the front door to businesses. Similarly, I believe generative AI will serve as the front door to businesses in the future."

To remain competitive in today's rapidly changing world, it is crucial to provide upskilling opportunities that empower your employees to drive the organization forward.