Now decides next. Is Europe ready for generative AI?

Generative AI: Transforming content generation, search and conversational interfaces

Generative AI, a specific type of AI known for creating human-like outputs,¹ is used to develop content across various formats like text, computer code, audio and/or visual output.² The most common applications reported by survey respondents globally included content generation, search/knowledge management, virtual assistants/conversational chatbots and content summarisation. In terms of integrated generative AI resources, the top categories are productivity applications, enterprise platforms, publicly available large language models (LLMs) and code generators.

European perspective on generative AI

Balancing caution with opportunity in adoption of generative AI

While there are broad similarities in use cases globally, European leaders show less interest and attention towards generative AI than their counterparts in the Americas and Asia-Pacific regions (figure 1). In line with this lower level of engagement, a significant portion of European respondents (over 20%) believe their industry and their own organisations are paying ‘too little attention’ to generative AI’s potential and implications. 

This could relate to less perceived pressure for European respondents to adopt generative AI, with only 26% reporting significant pressure compared to higher percentages in the Americas and Asia-Pacific. Additionally, they anticipate a more extended time frame for AI to significantly transform their organisations, with a higher proportion of European leaders believing it will take more than three years, and only 9% currently seeing transformative effects take shape. This contrasts with higher percentages in other regions where users believe in AI’s immediate transformative impact (figure 2). Research from The Deloitte Global Boardroom Program found that almost half (48%) of European leadership teams and board members identified their inability to show how technology enables growth as their biggest challenge when assessing the value of digital transformations.³ This reflects a broader technology-literacy predicament for European executives resulting in their belief that their organisations are not ready for generative AI.

When asked about the emotions leaders associate with generative AI, excitement and fascination are common responses across regions. Still, European leaders report notably lower trust concerning the technology. This mistrust may stem from cultural differences and concerns about AI-associated risks like biases and copyright issues.⁴ European companies are focused on developing this new technology responsibly and ensuring its trustworthiness. They aim to balance the potential and advantages of generative AI with the need for it to be regulated. This means ensuring that AI systems are fair, impartial and accountable. They also want AI to be responsible, robust and dependable, while being safe and secure and protecting privacy and confidentiality. Emphasising ethical AI practices could help organisations avoid reputational risk and enhance trust among customers and employees.

High expectations for productivity amidst slow adoption of generative AI tools 

European leaders in our study highlight efficiency, productivity, cost reduction, innovation and growth improvements as the benefits of generative AI, which mirrors global findings. These results are also consistent with previous reports such as the autumn 2023 edition of Deloitte’s European CFO Survey.⁵ A significant 91% of European respondents expect generative AI to increase productivity, aligning with global results. This is particularly significant for Europe, given the region’s recent productivity challenges, as highlighted by Deloitte Germany’s research into the economic effects of a shrinking workforce.⁶

Despite such acknowledged benefits, European leaders face implementation challenges. Lower interest levels, trust gaps, slow implementation of governing regulation and expectations of longer timelines for generative AI–driven change hinder organisational investment in and readiness for these technologies. Compared to other regions, European leaders report less preparedness for adopting generative AI in business areas like risk management, strategy, talent development and technology infrastructure.

Similarly, generative AI adoption in Europe is lower across all business functions compared to other regions (figure 3). Alongside regulatory considerations, this may stem from Europe’s challenging economic conditions and ongoing geopolitical tensions impacting interest and slow adoption. The survey took place against a backdrop of a US economy that had outperformed expectations and in which growth had accelerated. In contrast, European growth had slowed sharply and Germany, although not the euro area as a whole, had fallen into recession. The US has also enacted policies to enhance economic competitiveness, such as the Inflation Reduction Act and the CHIPS and Science Act. However, this does not necessarily explain the higher levels of adoption in the Americas as the NextGenerationEU programme could provide similar incentives for European organisations to adopt generative AI.⁷

Lower levels of generative AI adoption are certainly a result of European companies operating in a more complex and regulated environment than their counterparts in the Americas and Asia-Pacific regions. In December 2023, the EU provisionally agreed on the EU AI Act, its landmark, world-first AI regulation, which will introduce a comprehensive, legally binding, cross-sectoral framework for the technology to regulate its use and development.

Using a risk-based but prescriptive approach, the law will regulate AI, including generative AI, based on the potential risks of specific models or applications. Certain AI use cases, such as behavioural manipulation, will be banned altogether. For AI systems and models deemed high-risk, organisations providing or deploying them will be subject to stringent requirements, including pre-deployment fundamental rights impact assessments, pre-market conformity assessments and transparency obligations, to name but a few.⁸ While the compliance implications are likely to be substantial, the Act will also bring more accountability and fairer distribution of responsibilities across the AI value chain, as well as increased consistency across sectors.

The Act will also have global implications, as it will apply to any AI providers or deployers whose systems are marketed or affect individuals residing in the EU, regardless of their location. The final legal text, expected in early 2024, will give organisations further details to fully assess the Act's operational and strategic impacts.⁹ It will be interesting to observe whether further clarity on the EU regulations will speed up the pace of implementation of generative AI in Europe.

While there’s an expectation of comparable increased investment across the Americas, Asia-Pacific and Europe, European organisations in our survey reported allocating less budget to generative AI than their peers in other regions. The wait for the final legal text of the EU AI Act may account for the reluctance of European executives to move forward with investment as they wait to understand the regulatory trickledown of what the Act means for them in practice. Further, Europe does not have the same legacy of investing in digital transformation and disruptive technologies (figures 4 and 5). Historically, most external private investments in such technologies have been concentrated in the Americas, with the leading creators of generative AI and the most notable LLMs in the world being based mainly in the US.¹⁰

Globally, our report shows that leaders tend to prefer buying over building generative AI tools, a trend particularly noticeable in Europe, where 37% acknowledge this as their go-to strategy. In the Americas, it is 33%, and in Asia-Pacific 32%. This strategy is cost-effective for routine activities but offers limited control and lacks a disruptive competitive advantage.¹¹ However, this may not be a choice for many European organisations, who likely do not have the resources to create and experiment with LLMs and lack access to the high-specification hardware needed to train models. It has been widely reported that the graphic-processing units needed, for example, Nvidia A100/H100, have been stockpiled by various entities, especially in Asia.¹²

Talent strategies

European organisations are less active in reskilling workers, educating their workforce and recruiting technical talent (figure 6). The latter is partially due to the region’s more limited talent pool and existing skills shortages.¹³ More than a third of the EU’s labour force lacks necessary digital skills,¹⁴ and the UK seems to be in a similar position.¹⁵ These talent shortages, combined with modest efforts in educating and reskilling workers, are hindering Europe’s ability to leverage the benefits of generative AI fully.

Europe’s cautious approach to reskilling its workforce may be influenced by its strong labour protection laws and high unionisation rates. In the case of generative AI leading to job displacement, European businesses may perceive the immediate benefits of generative AI, like cost savings and productivity gains, as less substantial compared to regions with less stringent labour laws.¹⁶ Additionally, robust labour protection and trade unions require European companies to adopt a more deliberate approach when implementing technologies that could displace jobs as it can involve complex legal considerations.

Yet there is also the possibility of generative AI leading to job augmentation, rather than job displacement via automation. A recent Deloitte report on generative AI and the future of work¹⁷ suggests “there is a growing sense that generative AI will augment the human workforce rather than replace it.” In other words, generative AI can enhance the workforce experience by eliminating routine tasks, allowing employees to focus on more meaningful work and increasing employee job satisfaction and performance in the process.

As such, these rather limited efforts around talent might have adverse implications. The general-purpose nature of generative AI means that the demand for skilled labour could increase across a broad range of occupations and industries. In addition, in countries with ageing workforces or declining working-age populations, there’s often an increased drive towards automation to compensate for labour shortages.¹⁸ Firms in regions with a declining number of middle-aged workers have historically turned to automation to make up for this demographic shortfall. With many European countries dependent on declining working-age populations, the likelihood of widespread generative AI adoption increases.¹⁹

Completing such a transition means an increased demand for skilled workers at a time when demographic trends mean companies will be competing for an ever-shrinking labour pool.²⁰ This makes the lack of transparency of European businesses and reluctance to actively educate their workforce about AI’s capabilities, benefits and value puzzling. Still, organisations will only realise generative AI’s potential with the understanding and acceptance of employees. In particular, their fears about automation and job displacement need to be addressed.

Many European respondents in our study believe it will take up to two years to adjust their talent strategies for generative AI, with fewer feeling an immediate need for change than counterparts in the Americas or Asia-Pacific. This may indicate a more cautious approach to organisational change amid ongoing considerations of the technology’s risks, or it may simply be as a result of not yet knowing what the workforce implications will be as this technology rolls out. Will generative AI replace jobs or make jobs easier and more enjoyable?  Whether it plays more of a role in enhancing the employee experience and enabling people to be more productive at work or taking over entire tasks and roles is yet to be determined as the potential of this technology is explored.

Talent and skill gaps: Europe’s main challenge to maximising generative AI's potential

Across all regions, the technical talent shortage is a critical barrier to developing and deploying generative AI, with nearly 40% of European leaders selecting this as a key obstacle This is consistent with previous Deloitte analyses that identify talent resources and capabilities as the main challenge in Europe.²¹ European leaders also cite a lack of an adoption strategy and regulatory compliance concerns more than leaders in other regions (figure 7). This is even though European organisations have less difficulty identifying use cases than peers in different regions.

Concerns common across regions include intellectual property issues, regulatory compliance, a lack of confidence in AI results, transparency, data privacy and data misuse. European respondents more frequently see risk management as a barrier to implementing generative AI and are less convinced about their organisation’s efforts in governing AI adoption and mitigating potential risks. Effective governance of generative AI is likely to be an essential precursor to its scalable adoption across European organisations.

Respondents were also asked about strategies for managing generative AI risks. Top actions include monitoring and regulatory compliance, governance frameworks and internal audits. European respondents particularly emphasised regulatory compliance as important, tying back to the need for a clearer understanding of how the EU AI Act will impact organisations in practice.

Moreover, with generative AI, risk and regulation are no longer an exercise in technology management. Instead, when considered equally to other strategic levers they can realise significant value. The relative novelty of LLMs in business applications can be a challenge, and the risks of LLMs are dynamic and may change depending on their interactions with the user. However, development of guardrails, alongside proportionate deployment of testing, controls and monitoring mechanisms can empower organisations to use generative AI safely and confidently.²²

Generative AI: A strategic imperative for European businesses

This analysis shows that European leaders should prioritise preparing their organisations and workforce for the disruptive potential of generative AI. Recent Deloitte reports indicate that, although generative AI is a new technology requiring time for adoption and benefits realisation, aligning it with an organisational strategy is critical.²³

Europe’s cautious approach to this emerging technology, characterised by a wait-and-see attitude, contrasts with the more proactive stances reported in the Americas and Asia-Pacific regions. This difference in approach could see Europe lag in exploring the potential for generative AI, but it could also result in a more responsible deployment environment that considers new responsibilities that are created when technologies are invented.

Balancing the need for trust with the urgency to remain competitive in the global market is critical. This involves taking a multi-disciplinary approach to develop generative AI transformation strategies from the outset, and not just considering the technology potential itself. By approaching technology investment responsibly, while also investing in the necessary training and development of the workforce, European organisations can better position themselves to capitalise on the enormous benefits of generative AI, such as increased efficiency, innovation and competitive advantage.