The executive’s guide to generative AI

ChatGPT publicly launched in November 2022, and in the months since, generative AI has rarely been out of the news. Like the smartphone or the world wide web, this seemingly magical technology has gripped the imagination of journalists, the general public, investors and business leaders – and it’s not hard to see why.

Generative AI’s ability to quickly create readable text on almost any subject, as well as images and video, can seem like something from a sci-fi film. But as with the ‘traditional’ forms of AI used to detect fraud or recommend new products to customers, generative AI actually relies upon a combination of vast amounts of data and huge amounts of computing power to perform these seemingly amazing feats.

What makes generative AI special is the sheer range of things it can accomplish in response to simple prompts. “Unlike narrow AI, generative AI's broad capabilities make it a truly business-wide technology,” says Richard Fayers, senior director, data & analytics practice lead at business and technology consultancy Slalom. “It can assist in strategic planning and be used across every area of a business, instead of being siloed into specific teams or tasks.”

Foundation models (FMs) are effectively the engine under the hood of generative AI. They’re trained on massive amounts of structured and unstructured data and can be adapted to suit a wide range of tasks and applications, including text, image, video and audio generation. This has enabled generative AI to excel at numerous ‘human’ tests – including the bar exam and the math, reading, and writing portions of the SATs, a college entrance exam in the US. 

From a business perspective, the technology can be used for everything from creating new products to writing code. It can also research and summarise topics in a matter of minutes, highlight key findings from complex data, and forecast trends and demand. This means it could potentially transform countless industries around the world.

In manufacturing, for example, companies are using generative AI to design physical objects and optimise their material efficiency. Financial firms can simulate financial market scenarios, develop business strategies or analyse complex regulatory requirements. In healthcare, it is also helping to enhance medical image resolutions, identify abnormalities and even predict molecular structures in drug discovery.

“Imagine leveraging AI for personalised medicine – aggregating decades of research data to provide bespoke treatment plans,” says Aisha Mendez, associate partner for AI & Automation at Infosys Consulting UK. “These aren't just pie-in-the-sky ideas; they're in the pipeline, primed to improve patient outcomes significantly.”

Today, generative AI is used to rapidly generate reports, summarise a presentation or meeting, or refine code. But this is just the tip of the iceberg. “Streamlining processes, reducing manual workload, and allowing innovation in areas from consumer marketing to biopharmaceutical R&D are all within the offering...[of] well-integrated generative AI,” says Dr Clare Walsh, director of education at the Institute of Analytics.

Chanell Daniels is a responsible AI manager at Digital Catapult, the UK innovation agency for advanced digital technology. She thinks that one of the most interesting aspects of generative AI is its ability to help us see things in new ways. “It could be trends or insights from information that we’re so used to seeing that we aren’t necessarily able to connect the dots,” she says. “Or maybe from information that is so unstructured it would take a human a lot of time and attention to generate those insights.”

Most generative AI models produce content in one format. But multimodal models that can, for example, create a slide or web page with both text and graphics are also on the horizon; new use cases for generative AI are being discovered every day.

“As with any transformative technology, those businesses that are early adopters and find meaningful ways to integrate it into how they run their business or how they develop new applications will likely gain a competitive edge,” says Dr Peter van der Putten, director, AI Lab at Pegasystems and assistant professor in AI at Leiden University. “They can enhance operational efficiency, speed up digital innovation, and provide more personalised experiences for their clients or customers.”

All of which means that generative AI is far more than just the latest headline-grabbing technology. “It’s table stakes for future relevance,” says Mendez.

To unlock the game-changing benefits of generative AI, businesses need to develop a clear strategy for implementing it. Yet while 84% of businesses in the UK and Ireland have started using AI in some capacity, according to research from the business and technology consultancy Slalom, only 6% have a robust AI business strategy in place. So how can businesses shift from small experiments to truly transformative impact?

To begin with, says Nigel Vaz, CEO of Publicis Sapient, they need to know exactly what business outcome they’re pursuing and how generative AI could help to bring it about. “It’s vital to pose the questions in this order, with desired outcome first, not vice versa,” he explains. “Knowing this will help determine what changes – organisational, technological or both – need to be introduced.”

Conducting a comprehensive SWOT (strengths, weaknesses, opportunities, threats) analysis can also spotlight potential generative AI applications tailored to the company’s unique strengths and challenges. “Coupled with this, it’s imperative to assess high-value opportunities aligned with core business priorities, such as accelerating coding or personalising customer interactions,” says Dr Clare Walsh, director of education at the Institute of Analytics.

Having established where generative AI could add value, leadership teams next need to figure out how best to integrate it into existing business and technology systems. “All established organisations wrestle with the balance of running the day-to-day business while investing in and implementing new technologies,” says Vaz. “Generative AI models need to work with your existing systems to optimise value.”

It’s also vital to have buy-in from the leadership team for the rollout of generative AI models. “As generative AI will play a central role in future business success, key leaders must be aligned on strategy and approach,” says Vaz. “As well as your CIO and CTO, this means leaders from across the organisational functions, so that generative AI does not become a siloed pet project.”

Indeed, the sheer range of use cases for generative AI – both now and in future – means that a wide variety of business functions should be involved in generative AI projects from the get-go. These include finance, HR, legal and marketing, as well as more obvious teams such as IT and operations.

IT and finance need to be closely aligned, for example. “When deciding on generative AI investment based on desired business outcome, your CIO/CTO may advise on off-the-shelf, customisable, or proprietary generative AI solutions – each of which can come in at vastly different investment levels and require phased funding and oversight,” says Vaz. 

Aisha Mendez, associate partner for AI & automation at Infosys Consulting UK, says legal and compliance teams should also be heavily involved in generative AI implementations. “They help navigate the regulatory landscape and ethical considerations around AI use, ensuring that the organisation’s policies reflect the highest standards of responsible conduct,” she explains.

As job roles evolve and responsibilities shift, HR can help to ensure a smooth transition for staff. “Also, don’t underestimate the importance of marketing and communication teams,” says Mendez. “Their task is to articulate the benefits of these technologies both internally and externally, translating technical capabilities into strategic advantages that resonate with stakeholders.”

While the business benefits of generative AI are potentially significant, upfront investment in data and AI capabilities may be necessary to move beyond small-scale implementations by specific teams or individuals. For while it’s possible to experiment with out-of-the-box generative AI solutions at relatively little cost, an AI team comprised of data engineers, analysts, and ethics advisors may be needed to implement larger projects that could generate game-changing results.

Although this expertise might be out of reach for some businesses – particularly smaller ones – all leaders should keep in mind that generative AI is a tool rather than some “esoteric riddle wrapped in an enigma,” says Mendez. She also believes that front-line workers should have a seat at the table during any discussions about generative AI. “They know the processes better than anyone,” she says. “Add a few sceptics for good measure; you need people who’ll ask the hard questions.” 

Although much excitement is swirling around generative AI, it’s important to remember that it’s still in its infancy.

From startups to enterprises, organisations of all sizes are experimenting with the technology. They want to capitalise on it and translate the momentum from betas, prototypes and demos to real-world innovations and productivity gains. But, as Adam Selipsky, CEO at AWS, said in a 2023 interview with CNBC: "We're about 3 steps into a 10K race."

Rather than focusing on who will win the generative AI race, the priority should be getting as many runners on the track as possible. It is not a winner-takes-all situation; there is potential for many winners in this race. Every business should have the opportunity to unlock the potential of AI.

Mario Thomas is principal in the Executive Centre of Excellence at Amazon Web Services and says the situation today is reminiscent of the early days of the internet. “Ultimately, we know where that ended up,” he says. “This generative AI period we’re now in is creating a similar – if not more pronounced – frenzy.”

With investment pouring in and more companies testing the generative AI waters, employees and other stakeholders are increasingly asking how their firm plans to use the technology – and if they’re not, they should be. 

“AI can transform virtually everything that we know and do in an organisation,” says Thomas. “It’s going to go beyond chatbots into every aspect of business, and organisations that don’t use AI will start to fall behind.”

Even today, tools like Amazon CodeWhisperer, which provides code recommendations in real time, are changing how developers work. Advanced conversational search, rapid product design and highly personalised customer communications are also within the scope of current foundation models. It’s easy to imagine a future where generative AI-powered chatbots converse with borrowers and quickly go through their mortgage applications. Or where pharmaceutical companies use generative AI to accelerate gene therapies.

Philips and AWS are already pioneering AI in healthcare. Amazon Bedrock is a fully managed service that makes leading foundation models available through an application programming interface. It is being used to accelerate the development of cloud-based generative AI applications that will provide clinical decision support, more accurate diagnoses and automate administrative tasks.

Amazon Bedrock addresses the fact that one solution, or one model, is unlikely to solve every business problem facing the C-suite. “We’ve architected Bedrock so that the foundation models, including Amazon Titan, are available for the customer to bring into their own environment in AWS,” says Thomas. “The models are pre-trained… and the customer can then apply their data from their own AWS account. The beauty of that is that none of the customer’s data is used in training the model, which is a vital security point.”

With Amazon Bedrock, customers pass their data to Bedrock exclusively through the AWS network, ensuring it is never transferred via public internet. To ensure data security and avoid competitors benefitting from model customisations, privacy should be the foundation of any enterprise generative AI technology.

Finding the right solution

With all the hype around generative AI, it's easy to forget that AI comes in many forms – some of which may be better suited to the needs of the business. 

“A generative AI narrative might not work for every customer in the same way that a machine learning or data science narrative might not work for every customer,” says Thomas. “But the core point is that change is coming with this technology, and organisations need to start to identify the use cases that will help them transform, become more efficient, grow more quickly and drive an increase in market share.”

To find these use cases they should focus on what they want to achieve with AI rather than specific foundation models or apps. “Until we talk to the customer and understand what it is they’re trying to achieve, we can’t say if Amazon CodeWhisperer or the custom chip AWS Inferentia would be the right solution for them,” says Thomas. “The first thing for me is to understand what’s driving the customer.”

And that is often one of the following things: increasing revenues and growth; entering new markets or creating new products or services, delighting customers and attracting new ones; achieving operational excellence or becoming a better corporate citizen. Once the key drivers have been identified, it’s easier to figure out how generative AI could be used to achieve them.

Strong data foundations and a cloud-enabled business are crucial to achieving these goals with generative AI. And with digital transformation so high up the corporate agenda in the past few years, many organisations are starting from a good position. 

“Access to large-scale compute and large-scale inference has been enabled by the cloud,” says Thomas. “You don’t need to invest in hardware and those sorts of things… so the accelerated use of generative AI is being driven primarily by this easy access to the technology.” 

For those organisations that already have their digital ducks in a row, the bigger challenge is how imaginative you can be as a business.

Generative AI gives businesses an opportunity to reimagine their operations and energise employees by augmenting their capabilities. But the barrage of ‘robots are taking our jobs’ articles could make it a struggle to convince some staff members that AI is a valuable tool rather than a usurper.

“Generative AI is fueling heated public debate across all sectors, so it isn’t surprising that many employees feel threatened and are worried their jobs will be replaced by AI,” says Richard Fayers, senior director, data and analytics practice lead, at business and technology consultancy Slalom. 

But, as Aisha Mendez, associate partner for AI & Automation at Infosys Consulting UK, says: “…the notion that generative AI will eliminate jobs is rooted more in fear than fact. This technology is a catalyst for transformation.”

“We’ve witnessed this trajectory before with technologies like automation and robotics,” she continues. “Rather than eradicating jobs, they’ve reshaped them, elevating the work by automating routine tasks and opening doors to opportunities and innovation.”

According to a report by Capterra UK – a subsidiary of the research firm, Gartner – the majority (96%) of employees who use generative AI tools at work said it increased their productivity to some degree, with more than half (56%) saying it did so ‘significantly’. 

But as Chanell Daniels, responsible AI manager at Digital Catapult, points out, the generative AI pioneers will be working alongside people who are conservative about adopting new technologies. “They may need a bit more time and a bit more handholding to be able to integrate something like this into their job,” she says. “So you need to make sure that all of those needs are met.”

Mendez believes that the key is to understand and communicate widely that AI is an enabler, not a replacement. Employers need to clearly define and share the benefits for employees. “It’s also important to create a diverse pool of thought around how it is developed and used – in terms of different areas of the business and people from different backgrounds.”

According to Dr Clare Walsh, director of education at the Institute of Analytics, workshops that address employees’ ethical concerns and explain generative AI’s collaborative nature and need for human oversight can help to foster acceptance of the new technology. “Celebrating success stories where AI and humans effectively worked in partnership can change the narrative from fear to excitement,” she adds.

One obvious and effective teamwork involves AI taking on more of the repetitive and rote tasks. “If we use generative AI to automate the ‘robotic’ aspects of work, then more time can be spent on more ‘human’ work, which could lead to improving its quality,” says Professor Keiichi Nakata, head of business informatics, systems and accounting at Henley Business School and director of AI at Henley’s World of Work Institute. 

“For example, analysing a large amount of text to tease out key information can be an arduous and time-consuming task. Instead, generative AI can be used to digest and summarise the information, meaning that more time can be spent on formulating solutions to the problems identified in the text.”

Caroline Carruthers, chief executive of data consultancy Carruthers and Jackson, also points out that generative AI can help employees to overcome ‘blank-page syndrome’. “A lot of people who are trying to do something new are inhibited by that ‘blank piece of paper’,” she says. “So even if generative AI gives you something that’s not quite right, it’s at least a starting point.”

Indeed, paralegals and marketers are already discovering that generative AI tools can enable them to spend more time refining content, and less time struggling to get started on a new project. There are also tools that can help coders to work faster and more effectively by generating accurate lines of code.

Companies could also look for other distinct and time-consuming areas of work which could be automated with the help of a co-pilot. “For instance, studies show that knowledge workers spend over 2.5 hours daily searching across fragmented systems for information,” says Tadhg McCarthy, chief design officer at Elsewhen, a digital product consultancy.

“AI co-piloting provides a strategic path to recapture this lost time through focused integrations. So, for example, an AI assistant could be deployed to customer service teams to answer common support questions, which would drastically reduce the need for redundant searching across databases.”

Mendez sums up the situation well: “The future of work isn’t human versus machine; it’s human and machine co-creating value in ways we’ve just started to realise.”