Training
Transformative results start here
Although 72% of leaders have conducted employee training sessions on generative AI concepts, nearly a third view employee knowledge or training on generative AI as a barrier to implementation that is second only to security concerns.
Carm Taglienti, Chief Data Officer and Distinguished Engineer at Insight, believes the inherently iterative nature of generative AI partially explains the difficulties around employee training.
These AI models are trained, tested and then modified repeatedly based on outcomes and new data that lead to gradual improvement and the ability to generate more accurate or realistic outputs over time. Which means, as Taglienti observes, “you can’t just take a training session and off you go.”
Rather, competence requires consistency. “Training with generative AI is more like learning how to drive,” says Taglienti. “If you aren’t out on the road practicing, you don’t really know how to follow the rules of the road. You have to practice, get better at it, then adapt. It just takes time.” But, unlike practicing for a driver’s test, there are no standardized instructions or universal guidelines to steer employee training in the right direction.
Yes, we have comprehensive, ongoing training.
Somewhat, but it’s not fully developed.
No, we need to develop this.
No, because we have no plans to use generative AI.
The challenge that organizations are running into is that there is no manual.
“The challenge that organizations are running into is that there is no manual,” explains Juan Orlandini, CTO of Insight North America.
In the absence of definitive rules, and given the hands-on learning curve of generative AI, 66% of leaders have deployed private, secure generative AI tools — e.g., large language models and/or Generative Pre-trained Transformers (GPT) — that are trained on corporate data. The advantage of this approach is seen in Insight’s own private instance of ChatGPT called InsightGPT.
Instead of one-time learning sessions about the potential of generative AI, employees have the opportunity to interact with the technology on a daily basis. “I think it comes down to understanding what the capabilities are and setting the expectations properly,” says Taglienti. “When we train employees, we shouldn’t expect them to be prompt engineers or ‘AI whisperers.’ But we can teach them how to interact with the language model and to not assume that the responses will always be accurate.”
The investment in training is about creating a culture of continuous learning and adaptability — both hallmarks of an agile workforce primed to take advantage of advancements in generative Al. "As we shift gears into the next phase of adoption," says David McCurdy, Chief Enterprise Architect and CTO at Insight, "advanced training will be crucial to success."
Ultimately, the results of generative Al will reflect the values and commitments of the users deploying its capabilities. Training language models to deliver accurate, relevant information therefore requires an organizational climate that supports constant innovation, skill enhancement and knowledge sharing.
You can't just take a training session and off you go. You have to practice, get better at it, then adapt. It just takes time.
98% of leaders have access to generative Al tools at their organization. About 4 in 5 business leaders use generative Al tools for content generation (either written or audio/visual). This high rate of adoption suggests that those who invest in long-term employee training are likely to outperform workforces that aren't adept at integrating generative Al into their workflows.
