Not Getting Anywhere With Generative AI? Here’s Why — and What to Do About It.
Written by Joseph SladeContributing writer
Jillian VinerManaging editor
If you’re feeling the pressure (or the disillusionment) of unrealized potential with generative AI in your business, you’re not alone. Many leaders are navigating the generative AI landscape with big ambitions but few proof points.
It’s not for lack of trying.
According to a Harris Poll commissioned by Insight, of those asked, 66% of companies have deployed private, secure gen AI tools, and nearly 3-in-4 (72%) have conducted employee training on gen AI.
However, despite the buzz and the promising demos, turning generative AI’s theoretical power into tangible ROI remains a formidable challenge.
Don’t be discouraged. There is a path forward.
As an early adopter of gen AI, Insight experts have first-hand experience and client knowledge of the common mistakes and barriers standing in the way of meaningful gen AI returns. Here’s what you need to know to turn potential into progress.
Yes
No
Not sure
Over half (54%) of companies engaging with generative AI point to productivity improvements as one of the greatest outcomes they’re trying to achieve. But how do you measure productivity across the various roles in your organization?
According to the same survey report, 20% of professionals identified a lack of clarity around ROI and business value as a barrier to generative AI implementation. And when 24% say their greatest barrier is skepticism of the technology, the magic of generative AI seems less and less… well, magical.
Uncovering the value of generative AI starts with separating its theoretical potential and its real-world uses. After all, generative AI is a powerfully smart and fast assistant — not a magic wand. It’s a tool that operates under computational rules and needs well-organized data and specific instructions to be effective.
“It’s really easy to demo something that’s just going to wow you, but productizing it just takes time,” says Juan Orlandini, Chief Technology Officer, Insight North America and Distinguished Engineer. “You have to iterate through it. You have to secure it; you have to make sure you put boundaries around it, access controls, all the traditional stuff.”
His explanation isn’t intended to dissuade anyone from exploring or getting excited about what’s possible. Rather, how critical it is to have an AI partner — someone who can do the heavy lifting and get your proof of concept further, faster.
When venturing into generative AI, your data estate is the fuel that powers results. A well-prepared data estate, rich in quality and diversity, sets the stage for gen AI to thrive. It’s like preparing the soil before planting — the better the preparation, the more plentiful the harvest.
“If your data state is not in a condition that’s conducive to being consumed for learning, you’re going to get paltry results out of your AI initiative,” says Mike Gaumond, Senior Vice President of Strategy at Insight.
The first crucial step in using generative AI is to make sure your data estate is well-managed and prepared for extracting valuable insights. This involves creating a data architecture where information is clean, organized, easily accessible and covers a wide range of scenarios generative AI will face in practice.
However, a functional data estate is not enough. You’ll need to implement security safeguards that maintain privacy policies and prevent malicious activity.
“There’s a lot of things that impact what we call security these days, but it includes privacy as well as responsible AI,” says Carm Taglienti, Chief Data Officer and Distinguished Engineer, Insight.
By prioritizing both data quality and security, you lay a solid foundation for your generative AI initiatives.
You can’t do a two- or three-hour training session and presto! You understand how to employ AI, how to structure prompts or what the good use cases are.
Although a majority of companies (72%) say they’ve conducted employee training on generative AI, employee knowledge or training on gen AI tools emerges as a large barrier — second only to security concerns — to implementation for nearly a third (32%).
“You can’t do a two- or three-hour training session and presto! You understand how to employ AI, how to structure prompts or what the good use cases are,” says Gaumond. “I always use this stupid analogy that giving someone a hammer does not make them a carpenter. And that’s kind of what we’ve done with AI. We’ve given them this tool and we’ve given them two hours of training, and they’re not an expert on AI. They’re just a dude with a hammer.”
Generative AI stands to benefit all knowledge workers, but there are two major user groups: those who likely use gen AI in a chat function to find information and complete tasks, and developers who can use it to code and create frameworks. Bridging AI literacy gaps for both groups is essential.
For developers, this means continuous training on both the basic and complex gen AI functionalities, with an emphasis on building and maintaining effective AI systems.
For the remaining knowledge workers, the focus is on understanding the many ways it can streamline any given workflow. Training should demystify generative AI by developing a clear understanding of what it can do and the resulting benefits. And these trainings for non-developer knowledge workers should be straightforward and accessible for those with varying or no technical skills.
Daily
A few times a week
Rarely
Never
“When we train people, they need to understand how to craft their prompt to get the type of response they’re expecting,” says Taglienti. “We need to teach people to understand what the capabilities of the large language model are and how to represent the English language. Because it is a language model at the end of the day. Knowledge workers don’t need to understand the complex details of how generative AI works.”
Rather, their focus should be on effectively using the generative AI interface, applying its capabilities to enhance productivity and innovation.
“Make sure people understand how to craft the right kind of context to ask their question, and then how to understand the response so that they know whether or not it matched their expectations correctly — without assuming it’s always truthful,” Taglienti explains.
Ongoing AI education should happen at all levels of a company. As AI policies become integral to business operations, it’s important to have knowledgeable users. These guidelines not only promote safe and ethical generative AI use — they also seek to create a cooperative relationship between AI and its human users, ensuring that AI enhances human capabilities instead of replacing them.
The conversation around generative AI is rapidly expanding. Here are some insights from our AI experts to help center your understanding:
“2023 was the year of exploring what’s possible with gen AI… we expect 2024 to be the year of deploying real use cases into production.”
Juan Orlandini, Chief Technology Officer, Insight North America and Distinguished Engineer
The future of AI applications
“The applications should be intuitive... translating user input into the right actions for the AI... we’ll see different user experiences, applications, types of applications, engagements and different ways of doing things.”
Amol Ajgaonkar, Chief Technology Officer, Product Innovation, Insight
Understanding the big security and data concerns
“If you enhance or adapt a public model... you might expose sensitive data to unauthorized users. It’s about limiting what is exposed and tying that into roles and access controls.”
“From a security perspective... we have to worry about prompt injection, ensuring privacy and responsible AI.”
Carm Taglienti, Chief Data Officerand Distinguished Engineer, Insight
“We keep the C-suite engaged and grounded amidst all the AI hype... it’s about finding the right use cases for AI within the organization.”
Mike Gaumond, Senior Vice President, Strategy, Insight
Generative AI has been a series of sprints. “Businesses that take the lead now will create a gap so wide, there’s no chance of catching up,” says Joyce Mullen, Insight president and CEO and passionate generative AI promoter.
But here’s the challenge we’re all running into:
“Lately, I have been trying to focus not just the hype, but the nature of how to deploy generative AI. It’s more about how we take advantage of these capabilities without necessarily having to get caught up in all the hype. How can I look at it from a more practical perspective?” Taglienti says.
So, how do you get a more practical perspective?
Go with a guide. As the leading Solutions Integrator and as a generative AI early adopter, Insight has experience, lessons learned and quick first steps to help you make your generative AI ambitions a reality.
We’ll show you how.
Learn more
