Skip to main content

GenAI in the Netherlands: frontrunner or rearguard?

Naser Bakhshi, Partner AI & Data at Deloitte, on the shift from hype to pragmatism, on hallucinations and the importance of trust

Every quarter, Deloitte conducts global research on the state of AI in the business sector. The fourth edition of ‘The State of AI in the Enterprise’ also examined the Dutch situation.

Naser Bakhshi leads Deloitte's data and artificial intelligence (AI) team for clients in the consumer industry. Together with his team, he strives to push the boundaries of what is possible in the industry and tackle complex challenges such as poverty, food and climate risks, workforce skills of the future and new business models. In his passion for the transformative power of data and AI, he is guided by the belief that humans are and must remain the determining factor. 

Acceleration in an erratic market 

Developments in GenAI have accelerated over the past two years. Since the general public can use tools like ChatGPT and Gemini, we have also seen an influx of new applications in the business world. These developments are strongly influenced by an erratic and highly competitive market, by European regulations and by geopolitical shifts. Among other things, the positions taken by China and the United States present the Netherlands with a challenging competitive situation. Do we stand a chance to become frontrunners or will we remain in the rearguard?

Bakhshi: ‘We see a clear shift in sentiment around GenAI. Just a year ago, the general attitude was a lot more cautious. Back then it was mainly ‘We understand that this new technology offers opportunities, but what exactly can we do with it ourselves?’ Back then, people also found it a bit scary. Whereas now people say: ‘GenAI is here to stay, what are we going to do with it? Even at board level, the realisation is now dawning on people that GenAI is going to turn things upside down, which can seem threatening but which can also yield new processes and business models. So less hype and more level-headedness, you could sum it up.'

‘Yes and no. Right now we see this ambition to really start using this technology, everyone is curious to see what value can be added. But at the same time, we are all also looking at the risks and compliance. Isn't our brand at risk? Are we not jeopardising our loyalty programme? Are our customers not being inconvenienced? These are questions that companies are now emphatically asking themselves. Trust arrives on foot and leaves on horseback, so it is a very important topic. We at Deloitte therefore see Trust in GenAI as a spearhead in our approach to this topic.’

Any notable differences between the Netherlands and the situation worldwide?

‘On the one hand, we see a lot of enthusiasm and optimism: for instance, Dutch companies indicate that 88% of their employees are interested in GenAI, compared to only 50% worldwide. But on the other hand, as many as 40% in the Netherlands are uncertain whether they can comply with all relevant regulations. From Europe, new directives are coming at them at a rapid pace, deeply affecting all kinds of business processes. Dutch companies are also more concerned about transparency and misuse of data than companies in other countries. Half struggle with risk management, while globally barely a third see it as a challenge. In addition, we should not forget that GenAI applications also still hallucinate some of the time. None of this is good for confidence.’

‘We have to keep in mind: whether we are talking about employees, customers or citizens,the human factor must always remain at the centre.’ 
- Naser Bakhshi

‘Lots of companies struggle with this. On the one hand, 58% of Dutch companies put a lot of time and effort into training employees. That's more than in many other countries. But of course there is still a huge shortage of talent, and even though it may sound strange coming from me: you can't lean on external expertise forever. A recent study by the World Economic Forum and the University of Amsterdam reports that, in time, two-thirds of all jobs will be either fully or partially taken over by AI and robots. Incidentally, I do question that prediction. In IT and development, that seems more likely than in, say, a cocoa factory or a hospital. Plus: there will also be many new jobs. Jobs that require new skills, think LLM operators or prompt engineers. Moreover, it is also a matter of expectations and perception anyway, as is nicely shown in a recent infographic by Ipsos. In short, in my view it is a bit more nuanced than ‘GenAI is going to take over all the work’.’

Now that many companies have wrapped up the first pilots, are they succeeding in  scaling up working with GenAI?

‘That is not so easy for most companies. If you want to scale up throughout the organisation, you get to the core processes of the business. Then you discover things like ‘Hm, we are not as mature as we thought’, ‘We lack the right data’ and ‘Who has the right skills for this?’ Moreover, you then inevitably run into questions about data quality and security, mitigating risks and complying with regulation. And measuring the actual effects of GenAI is not always easy either. You need patience to properly integrate such a radical technology into your way of working — That's why it's worrying that 96% say ‘If it's going to take more than two years before we see tangible results, we'll pull the plug’.’

‘Everyone is looking for pragmatic translations of abstract regulations. We all know what a red traffic light means and what a green traffic light means. But GenAI still lacks such recognisable routines. How do users know they are interacting with a GenAI system? How can they safely and easily manage their own data? How do we see that data used is reliable? Such concerns call for 'privacy and security by design', not entirely coincidentally the motto of our Trustworthy AI framework. We all need to make sure that working with GenAI is done responsibly, that it actually produces correct results and that it meets the strictest ethics and privacy requirements.'

The survey involved companies. Do governments also benefit from the results? 

‘Definitely. For instance, take open source Large Language Models. Public organisations do not want to be tied to a specific provider, so they too are interested in solutions that fit their needs and allow them to remain flexible in the future. That's why it is also good that public and private organisations are collaborating more and more, as are companies and universities. At Deloitte, we are also active in the AI Coalition 4 NL, which promotes the development and application of responsible AI in the Netherlands. As a company, you don't make any money from this, but you do support the full breadth of civil society organisations moving forward.’ 

‘We have to keep in mind: whether we are talking about employees, customers or citizens, the human factor must always remain at the centre. Now that sounds nice, of course, but modesty befits us: we don't have all the answers either. So if the Netherlands does not want to remain in the rearguard, we will all have to discover and learn a lot in the coming years. But if we are prepared to invest enough time and energy, I am convinced we can be in the vanguard. Look, for example, at the surprising success of Mistral's flexible models. These have suddenly put France at the forefront of GenAI. The Netherlands may be a small country, but it was also the birthplace of ASML. So why couldn't we produce a hit in GenAI?’

Did you find this useful?

Thanks for your feedback