Executive leadership is a strong advocate for generative AI in their organizations and the technology is now top of the boardroom agenda, according to an interesting new report by Capgemini Research Institute. But whilst these tools which often harness large language models (LLMs) to learn and reapply the properties and patterns of data are quickly gaining traction amongst business leaders, enterprises are also concerned about their environmental and sustainability impact.
Capgemini surveyed 1,000 industry leaders across 13 countries, each at varying stages of generative AI implementation, and the results provide some useful detail for buyers that are thinking through what the technology means to them.
Generative AI has quickly become a hot topic for companies, across all their functions (not just in IT), since the launch of OpenAI’s ChatGPT, which saw rapid user adoption in just a number of weeks. The technology is impressive in its ability to understand large datasets, often (but not exclusively) taken from the internet, and respond to user requests about what the data contains. Companies are augmenting these LLMs and applying them to their own datasets to not only adapt how they work internally, but also develop new offerings for their customers.
According to the Capgemini report, nearly all executives (96 per cent) cite generative AI as a hot topic of discussion in their respective boardrooms. Pat Geraghty, CEO of GuideWell, a US-based mutual insurance organization, told Capgemini:
Every single board meeting we’ve had this year has had a standing agenda item of AI and ChatGPT. We want to make sure we’ve got our board with us as we’re thinking about where we’re going.
Amongst the 96 per cent of organizations that discuss generative AI in their boardrooms, over half (59 per cent) of executives say their leadership are strong advocates for the technology, only six months after it hit the mainstream. Only 39 per cent of executives say their leadership is taking a ‘wait and see’ approach and only 2 per cent globally say their leaders are either not convinced or divided by the potential of the technology.
Now, as we know, executives sometimes love to dive head first into the latest new technology in an attempt to be seen to be doing something - but the Capgemini survey does a good job of highlighting use cases that are already underway. When you consider how recently generative AI became a mainstream topic, it’s interesting to see how quickly it is being applied in some industries.
For example, the report highlights US financial services organization Morgan Stanley, which owns a vast collection of investment strategies, market research and analyst insights. Working through this to get the information you need can be time consuming for wealth management advisors and Morgan Stanley is using GPT-4 to power an internal chatbot that gives access to any area of the archive.
Jeff McMillan, Head of Analytics, Data, and Innovation, said:
You essentially have [access to] the knowledge of the most knowledgeable person in wealth management – instantly. Think of it as having our chief investment strategist, chief global economist, global equities strategist, and every other analyst around the globe on call for every advisor, every day.
We believe that is a transformative capability for your company.
Looking at the way generative AI will change the nature of work, Capgemini found:
70 per cent said that generative AI will augment the roles of knowledge workers and reduce their workloads
69 per cent believe generative AI algorithms will provide concepts and initial designs, where employees may shift from traditional ideation and creation to review and refinement
60 per cent said that generative AI will completely revolutionize the way we work
69 per cent believe generative AI will lead to new job roles (e.g. prompt engineer)
Overall, the majority of executives - 74 per cent - believe the benefits that generative AI brings outweigh the associated risks.
The Capgemini report also provides some interesting detail on how organizations have started using generative AI (albeit in its infancy). At a top level, the majority of executives (74 per cent) believe that the benefits will outweigh the associated risks. For instance, they said:
78 per cent said generative AI will allow the design process to be more efficient and streamlined
76 per cent said it will all organizations to create products and services that are more accessible and inclusive, serving a wider range of customers with diverse needs and preferences
71 per cent said generative AI will create more interactive and engaging experiences for customers
67 per cent said it can be used to improve customer service by providing automated and personalized support
And 65 per cent said generative AI can be used to improve internal operations and enhance facility maintenance
However, executives are taking a modest view of the projected benefits, where they said over the next three years generative AI will:
Improve customer engagement and satisfaction (e.g. NPS) by 9 per cent
Increase operational efficiency by 9 per cent
Increase sales by 8 per cent
And decrease costs by 7 per cent
The use cases themselves vary greatly by function in the enterprise. For instance, IT will likely make use of generative AI to speed up testing and code development, whilst marketing will create personalized marketing campaigns, and finance will use it to process invoicing.
The below chart showcases some of the examples that are emerging:
It is heartening to see that Capgemini also dedicates a significant portion of its survey to generative AI’s impact on sustainability - a topic that’s often overlooked when talking about the technology. It’s true that AI more broadly may help us tackle some of the complex climate change challenges facing us in the coming years, with its rapid data analysis capabilities, but it can’t be denied that generative AI comes with a heavier carbon footprint compared to other tools.
The report states:
As the race to build high-performance, generative-AI- powered tools and platforms intensifies, a pressing concern is the significant rise in computing power and subsequent impact on energy consumption and carbon emissions. The training of GPT-3, which forms the basis for ChatGPT, is estimated to have consumed 1,287 MWh of energy and resulted in over 550 tonnes of carbon emissions. This is around ten times higher than the emissions generated by an average car over its lifetime.
However, the environmental impact extends beyond training. Serving millions of users with these models further amplifies the energy demands. Integrating generative AI into search engines such as Microsoft Bing, which handles about half a billion searches daily, requiring at least four times more computing power per search than in a standalone product. Executives in our survey are aware of this; 78 percent agree that generative AI can have a larger carbon footprint than traditional IT programs.
Interestingly, the respondents of the survey are aware of this, with nearly 80 per cent of organizations saying that they are conscious of the need to build generative AI in a sustainable way. For instance, Louis DiCesari, Global Head of Data, Analytics, and AI at Levi Strauss said:
As the technology matures, we should be able to figure out more efficient ways of addressing sustainability concerns of organizations related to generative AI.
However, more worryingly, few organizations have a strategy to mitigate the environmental impact of training generative AI models. Only a very small minority (8 per cent) of organizations globally (64 of the 1,000 sampled) plan to invest in training generative AI models from scratch and of these approximately half have taken steps to mitigate their environmental impact, through renewable energy or offsetting carbon emissions.
A thoughtful survey that actually dives into how organizations are thinking about using generative AI at this moment in time. It’s worth reading in full, to get a sense of the direction of travel over the coming months and years. However, the concerns around sustainability should be the real eye opener here - I have heard little conversation about how anyone plans to tackle this in any meaningful way. And whilst organizations like to talk a big game around sustainability, this often gets sidelined when shiny new toys emerge and the prospect of increased revenue is put in front of leadership. The real challenge is adoption of generative AI with sustainability front of mind.