In April 2023, accountancy giant PwC announced it would invest $1 billion over the course of the next three years to expand and scale its artificial intelligence capabilities and upskill its 370,000-strong workforce.

Part of this wider upskilling initiative was a specific focus on developing the AI skills of PwC employees, deploying a mix of generative AI tools, training, and practical experience to build skills. 

The initiative, dubbed MyAI, centered on upskilling the consultancy’s 65,000 US employees, and as PwC’s Chief Learning Officer, Leah Houde, explains to UNLEASH in an exclusive interview, it also meets the needs of its client-facing operations.

“We needed all our people, not just the technologists, but all of our people to be savvy, responsible users of AI. So, we very quickly got started on a learning strategy for AI that included typical learning pathways,” Houde explains.

Those ‘typical’ learning pathways – which also included guest speakers and company-wide trivia games – served as a jumping off point to address the use of AI tools among employees, which was “not at the levels that we wanted it to be.”

Houde says she recalls the conversation “vividly” – the training was there, but the opportunity to experiment with the tools was the missing ingredient, somewhere for them to “mess around in.”

“It was literally this notion of a practice field, a playground, that we started talking about and got to the idea of prompting parties,” she explains.

While the concept of creating safe spaces for employees to test out the tools available without risk isn’t unique to PwC, Houde also states they wanted this to also be a fun experience for employees at the same time.

This line of thought led to the question: “What if we threw a prompting party to get people in there together and see some good use cases through our super users, but also to engage themselves and see how they can improve each other’s prompts?”

Opening the aperture of possibilities

PwC launched prompting parties in March 2024 and within the first couple of months, received more than 400 requests for more.

Since its inception, PwC has completed more than 500 prompting parties – roughly ten each week – with a further 800 in the pipeline. 

“It was amazing how quickly it took off,” Houde recalls. Indeed, one of the early parties in April last year which featured a guest speaker drew more than 22,000 employees to attend.

“What we heard from people is that these prompting parties have helped them. It opened their aperture around what’s possible,” she says.

Getting users to think differently about what the AI tools could do for them was a key driver behind the initiative. 

Houde details that employees had been using tools like ChatPwC – it’s internal instance of ChatGPT – and Microsoft CoPilot for efficiency tasks like document summarization, rather than a creative companion or, as she puts it, a “brainstorming buddy.”

“People are saying it has helped make them 20-30% more efficient once they got in there. There’s a thing about seeing it in action, just really in the context of what you are doing, and I think that’s the thing that made the prompting parties compelling – we encourage people to do them in their intact work groups.”

“The work group could get together and say: ‘Okay, what’s the work? Where are we in the work right now? How might we use these tools in the work?’ Being able to do that means they really see what’s possible.”

How to make learning a social experience

Making the learning experience a social one was a key aspect in the design of the prompting parties. 

As a “proponent of collaborative learning”, Houde explains that while solo learning is still a valuable method, collaboration seen in the prompting parties is “compelling and engaging”, with participants able to build on the ideas and inputs of others.

“It’s like any collaboration, right? You’re getting ideas that you wouldn’t have thought about yourself and being able to see things a little bit differently,” she says. 

Prompting party participants, Houde adds, have said this helps them to both think differently and use the AI tools in new ways.

Just being able to be there, ask questions and get feedback, see what other people are doing, practicing as a group and then revising…it has been incredibly helpful, because it’s pushing their thinking around what’s possible with these tools.”

Alongside these benefits, PwC is also aware of the so-called ‘Great Detachment’ – an ongoing trend in the US whereby employees are disengaged from their work and seeking new opportunities elsewhere.

According to research from Gallup, 51% of American workers currently fell into this category at the end of 2024, with just one in four (25%) employees confident to recommend their employer to others.

With disengagement rife – indeed levels of employees looking at new opportunities was last at this level in January 2015 according to Gallup – having an initiative that helps workers feel part of the process and engaged with their colleagues is crucial, particularly when virtual work has skyrocketed since the Covid-19 pandemic.

“In the hybrid work world we are still collaborative, but it’s having those moments where we are together which creates a different kind of stickiness and a different kind of experience,” Houde explains.

“It’s been nice that we’ve been able to use this vehicle to also get people being creative and innovative collaboratively. It adds energy to the environment, the experience.”

Keeping learning on the bleeding edge

As you’d expect from a regulated business such as PwC, consideration of responsible and sustainable use of AI is high on the agenda. 

Houde explains PwC’s ‘Human-led, tech-powered’ ethos means having the “constant drumbeat” of ensuring its “people are savvy, responsible users of AI.” 

This extends to avoiding the ‘garbage in, garbage out’ scenario with its AI, but also consistently updating responsibility guidelines alongside new technology developments.

“We’re going out in January with an updated learning bundle around what responsible use look like today, because it’s different than it did a year and a half ago when we launched this,” Houde details 

“Getting our people to use the tools and then reflect on how they’re using them, how it’s impacting their ability to do their work, how it’s impacting their ability to add value to our clients means they can have better, smarter conversations with our clients about how they might be able to use AI in their transformations.”

Another significant consideration is the constant innovation taking place in AI tools. Houde explains that PwC has worked in “relatively short sprints” with AI because “what we know is that we don’t know.”

She adds that the consultancy doesn’t want to “put a stake in the ground” only for new innovations to surpass them, so the learning department is consistently working with senior leaders on exploring the bleeding edge of AI.

“One small example in the humble world of learning is we are pilot testing an AI-enabled Socratic tutor, an agent that can actually teach users rather than just give the answers,” Houde explains.

Our role within the AI upskilling team is to make sure we’re keeping our eyes on that so that we can help expose the right people to the right skills that they need to have.”