Published 1 April 2024 in Technology • 13 min read • 
Audio available
No one needs to be convinced of the impact generative AI (GenAI) will have on our lives, organizations, and society. Yet uncertainty remains about how exactly to put it to work, see and seize the opportunities, and manage the risks. Amid the hype, executives can feel paralyzed by indecision or compelled to jump in headfirst without a plan.
GenAI is undoubtedly one of the most powerful and transformational technologies to emerge in decades. For most, it will bring more productivity and efficiency. For others, it will open new lines of innovation and offerings that weren’t achievable before. There will also, inevitably, be negative impacts that we must seek to mitigate.
But, despite its power and potential to become a teammate, a coach, and an assistant, let us not forget that GenAI is also just a tool. Whatever their industries or markets, businesses develop and achieve a set of value propositions for their customers, stakeholders, and employees, with or without GenAI. Just as the advent of the internet didn’t change the need for organizations to focus on growing revenues, cutting costs, and finding ways to attract and retain customers, the same is true of AI. Some companies will be or become “GenAI” companies, but Michelin will continue to produce tires, UBS will remain a bank, and Levis will keep selling jeans. The democratization of AI will be the enabler that helps the vast majority of companies do what they do more efficiently for less cost and perhaps more innovatively. As such, GenAI and its related technologies shouldn’t be the end goal unless you are in the AI industry. That said, given the transversal nature of this technology, you will need to have a structured plan for it and know how to make it work for you.
What should your in-house approach to AI look like? How should you go about weaving GenAI into your over-arching corporate strategy and the cogwheels of your organization? Where should you start this journey? What should you give up, what should you keep, and what should you invest in? To help organizations succeed in their inevitable adoption of GenAI, we have identified five strategic tensions that executives and organizations should reflect on and resolve to extract the most business value securely, responsibly, and sustainably.
It is worth pointing out that each organization is different and most likely at a different stage of this journey: there is no “silver bullet” answer or “correct” position for any of these five trade-offs. Your reflections and answers will likely change as you move through the different stages of your AI roadmap and as the technology evolves. We recommend regularly returning to these trade-offs to adapt to changing circumstances and needs.
AI is revolutionizing the world of business, at pace. How do executives guarantee this mass adoption benefits both their organizations and society? In Issue 13 of I by IMD, we explore how to lead effectively – and responsibly – in the age of AI.
Since OpenAI launched ChatGPT in November 2022, the interest in GenAI has skyrocketed. Amid all the talk, many organizations are already investing in resources. A Gartner survey of 1,400 participants conducted in the third quarter of 2023 found that 45% of respondents said their organization was piloting or experimenting with AI, while another 10% had already put AI into action.
So, how fast should you go? On the one hand, you don’t want to be left behind in the race for productivity and other advantages, but there is also a danger of over-investing and moving too quickly when it is still unclear what the risks and clear benefits might be.
There is much optimism about the bountiful gains in productivity, quality, and even creativity that AI will bring. However, the challenge facing firms is that – so far – tangible gains are still being explored and understood. As the initial enthusiasm starts to wane and concern over risks and governance grows, many firms may come down on the side of caution and pause or block the use of AI tools. Around 20% of IMD’s clients have banned the use of GenAI in their organizations – at least for the time being. When weighing up whether to use the technology, executives must consider a lengthy list of evolving risks, including security concerns, copyright infringement, accuracy, and bias.
Another challenge is that AI models are opaque, making it hard sometimes to understand and trust their output. Organizations must balance the need for accurate predictions with the ability to interpret and explain AI decisions, especially in high-stakes applications like healthcare or finance.
Organizations must balance the need for accurate predictions with the ability to interpret and explain AI decisions.
So, how should executives navigate this tension? One way is to assign AI selectively to specific tasks where the costs of failure are low and the benefits and scalability of automation are high. Automating mundane and routine tasks, such as drafting emails, summarizing documents, and writing basic code, is a case in point.
Organizations concerned about privacy can look at installing guardrails to restrict the use of these tools, for example, by limiting where the data goes. Companies including Microsoft, Google, and Adobe Firefly are marketing products guaranteeing that the proprietary data uploaded into their engines is not shared with others or used to train models. However, many organizations remain skeptical about whether they can trust these claims.
We recommend: Start with lower-risk, simple applications in areas that will have the biggest return on investment, and then monitor, learn, and adapt as you go. Starting simple has many benefits, including giving your organization a quick win and letting your people see first-hand what it takes to ramp up to more complex AI projects
Don’t get stuck on whether to leverage AI to improve your internal processes or to create new products or services. GenAI can do both. The decisive factor for your organization is where it will add the most value. Developing an AI strategy can help foster clarity and lay down principles that align with your organization’s values and overarching corporate strategy. Given the issues with accuracy, organizations may be tempted to focus on internal applications. For many, integrating AI into large business process platforms such as Salesforce is the first port of call. Tools like Microsoft CoPilot are being widely implemented to act as personal assistants for workers and free them from the drudgery of mundane and routine tasks. Given the eventual ubiquity of AI tools that drive higher productivity, organizations that do not develop customer-facing AI-enabled innovations risk becoming obsolete.
Gartner’s AI Opportunity Radar is a simple framework that helps organizations think about external versus internal use cases while mapping out productivity gains against innovation. In line with our first recommendation, understanding the complexity level of the various AI-enabled ideas is also critical. Don’t leap to the big ideas until you’ve built the organizational know-how first.
One Big Pharma company, which regularly sends out pamphlets and free samples to physicians, struggled to keep its contact list updated. It used ChatGPT to write specific Python code to track the exact locations of physicians. Within minutes, AI updated the addresses with 80-90% accuracy, saving the company the work of two full-time employees and hundreds of thousands of euros in misdirected postage costs.
Others are using AI to bring outsourced work back in-house at reduced cost and increased speed. For example, a Middle East-based telecoms company uses OpenAI’s DallE and Midjourney, which create images from natural language descriptions, to develop social media campaigns that it can iterate, adapt, and roll out quickly.
A further application includes back-end tasks that help customer-facing roles. An automotive company has developed an AI app to tell car salespeople whether a golf bag fits in the trunk, saving them the task of looking up measurements.
Finally, some organizations are using AI to create innovative products. An interdisciplinary team at the California Institute of Technology designed a 3D-printed catheter to combat urinary tract infections. The team used AI to run digital models of catheters through a series of simulations to find the best design to stop bacteria from moving upstream. The result? The design reduced catheter-related bacterial infections by two orders of magnitude.
We recommend: Internal versus external is a false dichotomy. You should do both. Identify a set of clear business challenges, both internal and external, and focus on where AI can provide the most benefit. Avoid the temptation to launch dozens of interesting but disconnected AI initiatives.
When deciding which tasks in your AI pipeline or entire AI-enabled projects would be best suited for automation and scale, the next question is whether to build or buy the technology. First and foremost, organizations need to recognize that new AI adapters, models, and other tools are becoming available daily from commercial providers. You may identify a gap where you must build something yourself, only to find a few months later that a cloud architecture or AI provider has created the same thing for a relatively small cost. Is it better to carry the technical debt of what you have developed or to switch out that capability for commercial tooling?
Given concerns around privacy, many firms are limiting the use of public offerings while they source and build their own AI tooling. To do this, companies will require massive training datasets, computing power, and large algorithms. Open-source platforms, such as Hugging Face and Kaggle – where the machine learning community collaborates on models, datasets, and applications – can provide some, if not all, of what’s required from a tech-stack standpoint. However, even in this case, companies must have a team with expertise in data science, machine learning, and software development to customize the AI for their needs. For this reason, the build option is a non-starter for most.
Another option is buying one of the many off-the-shelf options available, such as licenses for GitHub Copilot for coding, Jasper for content creation, Midjourney for image generation, or Zapier for process automation. The challenge here is that there can sometimes be fewer customization options. The sweet spot for most organizations is somewhere in the middle, depending on what is commercially available already.
Beyond the AI tool, a clear differentiator is the quality of the data – something that has been true for businesses throughout the digital era. But where we have built towering data silos in previous revolutions, the future will instead see a great data de-silofication so that large language models can swiftly consume and make sense of organizational know-how. How quickly organizations can ready their data to support their AI initiatives will be critical to the success of any such initiative, as well as defining how data privacy and security will be handled.
We recommend: The Build versus Buy question has several factors that will require the organization to be agile in how it architects and executes its AI strategy. It is generally good practice to utilize commercial tooling where possible. At the same time, make sure the technologists have time in the plan to adjust the AI solutions you adapt from external providers or develop in-house.
“A recent study from MIT found that it is still much cheaper to use humans than AI for most jobs in the US, suggesting that AI job displacement may be slower than expected.”
The democratization of GenAIhas caused many people to worry that the technology may soon put them out of work. The IMF has projected that AI could affect up to 40% of jobs worldwide, potentially rising to 60% in more developed economies. Added to this is the explosion in new job titles such as AI Ethicist and Chief AI Officers, causing many employees to question whether their existing skills might soon be obsolete.
For now, the future impact on the labor market remains unclear. While a powerful tool, GenAI is still prone to making errors and is costly to run due to the vast amounts of power-guzzling chips that these systems require to operate, even as companies develop more efficient applications. A recent study from MIT found that it is still much cheaper to use humans than AI for most jobs in the US, suggesting that AI job displacement may be slower than expected.
This suggests that the pendulum is swinging towards reskilling rather than replacement. Yet you can’t just drop a powerful AI model on top of your organization and expect it to bring you value. The model will require new processes, maybe new incentives, and, crucially, new ways of working. How you go about it depends on your organization’s existing levels of change management and tech aptitude. For legacy firms, where staff may be less tech affine, companies should consider mass training programs that demystify the technology and give staff a basic understanding of how to derive the biggest benefits.
As the productivity gains from GenAI become more apparent, organizations may find they can do the same work with four people rather than five. Here, they will face the question of what to do with this extra person. Do you deploy that worker’s time toward higher-value activities? Do you reduce your headcount? Or do you reallocate the workload so everyone can work a four-day week? The answers to these questions will depend on your organization’s and industry’s specific needs and goals.
We recommend: The most responsible employers will consider the social and ethical implications of using AI to replace human labor along with productivity and cost gains. Explore strategies for retraining and upskilling affected employees.
“By 2027, global AI demands may use more water than some countries do in a year. This is a serious problem, as freshwater is becoming a scarce resource due to population growth, water depletion, and outdated water infrastructure.”
The environmental impact is one often overlooked tension in the debate around GenAI adoption. These models have a huge appetite for computing power, both during training and for inference, when the trained systems are put into action, meaning they consume vast amounts of electricity and water.
Hugging Face reported that its BLOOM model consumed 433 megawatt hours of electricity during training – enough to power 40 average American homes for a year. Another scientific analysis found that Google would need as much power as Ireland if it used ChatGPT for its search engine. Muddying the waters is the fact that the exact CO2 emissions of GenAI are difficult to calculate – due to a lack of transparency among Big Tech about their actual footprint.
GenAI requires vast amounts of water to cool down the data centers that run models. According to one estimate, Gen AI needs the equivalent of one 16oz (475ml) bottle of water for every 5-50 interactions. Another study predicted that by 2027, global AI demands may use more water than some countries do in a year. This is a serious problem, as freshwater is becoming a scarce resource due to population growth, water depletion, and outdated water infrastructure. GenAI also consumes a lot of hardware, which uses rare earth elements that must be mined and transported. With pressure growing on companies to monitor and report their direct and indirect carbon emissions, the environmental impact of GenAI is something they cannot afford to ignore. Many firms have made bold public net-zero commitments in recent years. A massive scaling of GenAI applications across their organizations could jeopardize these targets.
On the flip side, if GenAI makes us more productive and efficient, it could lower the environmental impact whilefinding new ways for companies to become more sustainable – even finding solutions to mitigate its own footprint. For instance, Microsoft and the Pacific Northwest National Library have recently uncovered a new material – with the help of AI and supercomputing – that could replace lithium, the finite resource used in everything from electric vehicles to mobile phones. The technology also might help firms with their ESG reporting, for example, by calculating carbon footprints. Some firms are looking at how to creatively reuse the heat generated by data centers, such as to heat local homes or water systems. So, how can organizations offset this tension between damaging the environment and increasing organizational productivity?
Companies can adopt several strategies to use GenAI more responsibly and efficiently. For example, they can choose vendors that source their electricity from renewable sources and look to fine-tune existing models rather than building their own, which is more energy-intensive. Furthermore, they should adjust the amount of training needed depending on the accuracy required for the task. For use cases where accuracy is not so critical, less training may be required.
We recommend: Ensure that sustainability is one of your key AI principles. Educate your organization about AI’s footprint and prioritize the measurement of resource consumption. With this transparency in mind, strive to consistently reduce the harmful impact of AI on the environment.
There is little doubt that the pressure to adopt AI-enabled technologies to maintain a competitive advantage will continue. Companies will need to navigate this adeptly and deal with other strategic tensions, using skills they already have and new muscles they will need to develop.
As the technology matures, some existing problems, such as data bias, will likely diminish. Regardless of the challenges, by defining a set of AI principles and a clear strategic approach, organizations will be more equipped to make decisions about which trade-offs to make and which innovations to invest in to ensure that AI is used to generate real value for your business and society.
A flurry of venture capital activity presents investors with a challenge: how to distinguish genuine innovators from opportunistic players capitalizing on the AI hype. Enter Harshul Sanghi, a seasoned technology investor with...
26 April 2024 • by José Parra Moyano in Artificial Intelligence
AI holds incredible promise to improve virtually every aspect of our lives, but we can't ignore its risks, mishaps, and misuses, author Art Kleiner tells José Parra Moyano....
23 April 2024 • by Jerry Davis in Artificial Intelligence
GenAI can deliver cost and productivity benefits for colleges and students, but there is a dark side....
Audio available
22 April 2024 • by José Parra Moyano, Christine Legner, Konrad Schulte in Artificial Intelligence
Training and fine-tuning AI models requires troves of data, but storing and processing that data is financially and environmentally costly. Here are three ways to address the problem....
Audio availableExplore first person business intelligence from top minds curated for a global executive audience