Skip navigation
Employees working in a dim office setting.

Why Empowering Employees Is The First Step To Successful AI


DATE: 19 February, 2024

In 2024, AI has put us on the precipice of a perfect storm, the likes of which only come around every 100 years. Much like the moment when computers evolved from taking up a whole room run by expert scientists, to personal computers arriving in our home offices (remember the Commodore 64?).

In 2023 Open AI's ChatGPT signalled a revolution—the year where AI technology, which had been brewing since 1957, was now accessible to all via its intuitive interface.

This year, as organisations start to consider how AI tools will modernise their functions and deliver newfound productivity—the key questions on the lips of leaders are, “What is this AI thing?” and “Where do we start with AI?”.

And it's here that you can trip up as there is a key consideration that many companies overlook.

Rather than focusing on how AI delivers services to the customer, the first step is to find out how you're going to empower the people who deliver those customer services—your employees.

For the past 12 months, I've been involved in an AI experiment. At KPMG in March 2023, we rolled out a secure internal version of ChatGPT with Microsoft, which we named ‘KymChat'. We didn't just roll it out to a few people in a pilot group but to all 10,000 employees. 

In fact, let me be totally honest. Because I am one of these lucky KPMG employees I have just asked KymChat “How does KymChat help KPMG employees in their everyday work?”.

The answer KymChat provides is this:

“KPMG's KymChat assists employees in their day-to-day work by providing access to the processing power of a supercomputer through a digital assistant on their desktop and phone. This allows employees to safely use ground-breaking AI technology in the workplace without having client data leave the KPMG environment.

KymChat supports the firm's culture of innovation, boosts efficiencies, and creates a better people experience. It enables cross-team collaboration and helps those new to the firm with a more seamless and efficient people-navigation experience. Over time, KymChat will allow KPMG to train its own AI model for specific purposes, with the initial use cases focused on innovation and efficiency within the firm. It is directly integrated with Microsoft Teams, with further new features and capabilities planned.”

You see I no longer have to think of everything myself. I didn't have to dig around for information or spend ages agonising over the right wording that best explains what I'm trying to say. What's more, KymChat gave me the links to all the source articles that it referenced, so I could validate the information.

I also know that this information has been fed into the bot over time, training it to know the answers to many questions that employees might ask. For example, we have an 800-page risk manual (few of us have read cover to cover) and most likely we'll need to check a policy in the course of our work, so we want an answer quickly. We can now ask questions to this 800-page data set and it acts as risk guidance for employees.

I'm not for one moment suggesting that you rely on a bot for these things, far from it, you still need to refer to a human and check the sources. Nor does a bot do all my work or make my decisions for me—far from it. 

AI does, however, complement my knowledge and helps me to get tasks done so much faster, and really helps me have another perspective or when I've got gaps in my knowledge.

The importance of focusing WITHIN your business first to empower employees and build your business is a crucial step for two key reasons:

1. Safety first

By far the question I'm asked most outside of work about AI is—why can't we just use ChatGPT at work?

While ChatGPT might be free and accessible, it comes with risks for businesses and employees, who can inadvertently share IP or private information. This can lead to breaking confidentiality and intellectual property rules. It can mislead employees as the model has not been trained on their specific industry or company data.

KPMG employees can access the fifth largest supercomputer in the world using a digital assistant on their desktops and phones. This allows the firm's teams to safely use ground-breaking AI technology in the workplace without having client data leave the KPMG environment.

The use of generative AI comes with risks. These include the potential for misinformation, bias, discrimination, copyright issues, and risks to confidentiality and accuracy. Therefore, it's crucial to ensure you have a human review process in place and expertise is built into the generative AI use process. 

“Great bots are not born, they are trained.”

This is one concept that does not appear to have landed yet in the mainstream (many people I speak to wonder why the bot doesn't know the answers to everything from day one). If you want quality answers, you need to provide quality data to the bot and validate that data by getting feedback into the knowledge loop.

And what better way to do this than by asking the experts in your business? While training them to use a large language model.

At KPMG we have a feedback button—a thumbs up or thumbs down to indicate if we got a correct or quality response, so the answers will improve vastly over time.

2. Empowered employees equals better productivity

Much like accountants before the calculator or spreadsheet, AI is a tool that helps us power through repetitive drudgery to get our job done. We can now automate and execute language-based tasks with unprecedented speed and efficiency.

Simply put, large language models like KymChat reduce annoying time-consuming tasks and empower employees with the right information, when and where they need it, in a few seconds.

Instead of spending hours wading through company information, imagine if you could simply download a bespoke template or report by simply asking a bot the right question. In fact, you could produce the same report for two different audiences in minutes, freeing up time for employees to focus on higher-value tasks.

And what if you needed to find a particular person in your company quickly? Instead of spending ages pinging or calling people you know in your network to ask them, what if you could ask your own company chatbot this: “Who is the best person in our company to answer this question?”

For many years now onboarding inside organisations has been… let's say ‘fairly lacklustre' and uninspiring.

For new graduates in 2024, they can now augment their learning and ask hundreds of questions of ChatGPT or their organisation's LLM. It never gets tired or makes them feel silly while they are learning the new role, all without annoying their new boss.

They can figure out the best templates for things like reports and upskill at their own pace. Sharing a good draft with your manager always feels better than asking where to start.

The list of gen AI applications is seemingly endless from summarising and classifying legal documents, responding to employee questions and assisting expert advisors to generate business reports.

Technical folk can see huge productivity gains too—imagine being able to spin up the code you need to create a new workplace tool in minutes. That's exactly what gen AI empowers developers to do and it's becoming more accessible to all employees via low code.

What's more, working alongside AI goes far beyond knowledge management roles like those in banks and professional services. In the UK and EU, radiologists now have a full-time second opinion that can work 24/7 and it never gets tired. AI radiology tools have proven they can identify findings on scans that otherwise might have been missed by the human eye. This is a game-changer for healthcare and with early intervention, will result in better patient outcomes.

While there is still fear around the disruptive forces that are in play around AI, balancing competing priorities and managing workload effectively is still a persistent challenge for a great number of employees.

There is a myriad of productivity benefits for company employees who are lucky enough to use LLMs today, along with improved well-being and work-life balance. Even simply using AI to overcome information overload, to make more well-informed decisions in a timely manner, reduces both health and business risks.

A recent study by Microsoft shows that (in Australia) by shifting worker tasks to higher-order thinking roles, AI and automation technologies could increase the incomes of low-skill workers by 10% by 2030.

The next 12 months will no doubt set apart the organisations that decide to invest in AI and empower their employees to thrive in an increasingly dynamic and demanding professional landscape.

Fiona Wilhelm is part of the AI enablement hub at KPMG, she has over 20 years of experience in digital disruption with a focus on Human Centred Design, through a varied career across professional services firms. Fiona has steered projects in collaboration with Microsoft and KPMG Global. She runs innovation and training sessions centred around the potential of AI to support employees.

You can book Fiona for a keynote via this link.


PREPARING FOR AI: The Implications of Artificial Intelligence for Jobs and Skills in Asian Economies August 2019