Full integration of generative AI elusive for many employers: report

Not surprisingly, generative AI continues to gain speed among many employers: 77% of 500 executives polled at midsize organizations in the United States and Canada say their organizations have adopted solutions such as ChatGPT and Microsoft Copilot.

And with good reason, as 85% of respondents using generative AI saying its impact has exceeded expectations, with benefits including increased efficiency, better customer service and lower costs

However, 41% of respondents in the survey also said their businesses have only partially implemented AI technology.

And just one-fifth have achieved full integration, which means incorporating generative AI solutions into existing processes and systems, like CRM (customer relationship management) or ERP (enterprise resource planning).

“What we're seeing is a lot of organizations experimenting in some way, shape or form… we're in very much an early adoption phase around AI and around generative AI,” says Sawan Dhaliwal, director at RSM Canada in Edmonton.

But when it comes to integrating the technology, first adopters are going to do better, he says.

“There's a group of innovators who have been looking at AI for a long time — they have teams of data scientists that have been a part of their team for some time now — and now, with the advent of generative AI, we're seeing a lot more folks really figuring out how to integrate generative AI into their operations.”

Full integration in this sense is about your dedication around looking at opportunities for AI, and integrating it where it makes sense, says Dhaliwal.

“It's about having concentrated, focused strategies and teams that are actively deploying AI of various types, whether that's machine learning or whether that's deploying large language models, etc. It's about having dedicated teams that are actually going after the opportunities versus, you know, saying that ‘We have AI across every single process that we have.’”

‘Low-hanging fruit’: How AI is being used

When it comes to the more common ways that generative AI is being used, RSM found that:

• 61 per cent of middle market Canadian companies are using it for automated tasks
• 51 per cent are using it to enhance customer service
• 46 per cent are using it to increase employee productivity and creativity.

“There's definitely a lot of low-hanging fruit that's out there right now, and we should be looking at those as the starting point for how to enable this technology,” he says.

For example, with customer service, that's an easy place to start where you have information and you can provide a routine answer to common questions, so “you can train these bots to be able to manage those types of queries,” says Dhaliwal.

“When we're looking at the low-risk scenarios, those are some of the areas… that any organization really should be looking at.”

A risk-based approach is key, he says.

“We're also looking at what's the risk of implementing and what's the feasibility of the technology at this point? So, you do have to consider all those pieces of the puzzle as you're figuring out the right place to start on this journey.”

It makes sense for employers to bring in these tools when they start looking at where they are gaining efficiency, says Dhaliwal, “as long as you're looking at the ROI and looking at the maturity of the tool that you're implementing as well.”

Understanding value of AI for full integration

Right now, many companies are using generative AI in a copy-and-paste manner, to generate a business letter or social media post, for example, says Fredrik Odegaard, associate professor at the Ivey Business School at Western University in London, Ont.

The next step is the whole integration process, where those steps happen automatically with the push of a button.

“That’s going to be the key aspect where you will see which tasks is it that will be generating value or ROI in the investment in these gen AI tools,” he says.

“And I don't think that value is going to come by people just having one more thing to do in terms of copy and pasting back and forth from ChatGPT or [Microsoft] Copilot and so on.”

It’s about figuring out which tasks are going to be automated, says Odegaard, and if they are, what value does that generate?

“Or, does it just eliminate these tasks? And, once we have done that, then of course people's roles will change. And so then what are those value-added activities that are going to come out of that?”

Challenges to AI integration

More than half (54%) of the respondents to the RSM survey said the technologies have been more difficult to roll out than expected, found the survey, conducted from Feb. 26 to March 4, 2024.

There tends to be a discrepancy between the person who is responsible for IT integration and their budget versus the ones who are supposed to be using the new tech, such as customer service, says Odegaard.

“The users may have a particular process by which they work and now there's some IT person who's coming in [saying] ‘Alright, we have this new tool, you shouldn't be doing it the way you have been doing it, you should be following this process.’”

With the new tech, employees’ processes should be simplified, not made more cumbersome, he says.

“But a lot of people work differently so that people… in finance and accounting, they work differently from the people who are in marketing; marketing works differently from people in ops.

“So, it's hard to find one perfect, singular solution that's supposed to work for all the people involved. And I think that is one of the benefits of gen AI is it's so adaptable to be able to understand what it is being queried for.”

Studies have shown that the way to have this new tech grow within organizations is to start off with AI “islands of exploration,” with pockets happening in different areas, says Odegaard.

“When you start to accelerate is when you start to talk across business functions and business units and come together and have a focused strategy on how you're going to actually deploy AI in your organization.”

Then, it’s about moving into “centers of excellence,” and finally “federations of expertise,” he says.

“But the quicker you can get to [have] cross-functional teams working together to figure out what the best-use cases are, and how to best deploy this across the organization, the quicker you're probably going to realize value out of it as well. You're going to find more enterprise-use cases that'll have broader implications for the organization.”

Outside help preferred by employers

Another interesting finding from the RSM survey is 67% of respondents acknowledged that outside help is needed to maximize the benefit of the generative AI solutions they have chosen.

This is an area that does require upskilling, and depending on the size of your organization, you may not have the IT resources or development resources on staff, for example, that are needed, says Dhaliwal.

But it’s also important to a step back to look at what AI means to your organization, and that should involve questions such as: “What's the overarching AI strategy? How should we be growing our teams around AI?” he says.

“Because there's a lot to invest in, whether it's technology or people or other resources, you have to make some of those strategic decisions.”

And that means also deciding whether you want to be developing bespoke tools for yourself, says Dhaliwal.

“Is that going to give you a competitive advantage? Or should you be looking to adopt tools that are already available in the marketplace? And then what type of skills do you need around that, to make sure that the folks on your team have the necessary skills to manage that?”