Chapter 1: Understanding AI's Vision of Dutch Commutes

The series "Our World in AI" delves into how Artificial Intelligence interprets reality. By generating images that represent various aspects of society, I analyze the outcomes to determine whether AI mirrors actual conditions or exacerbates existing biases.

Using a detailed prompt that encapsulates a slice of daily life is crucial for this process. This specificity aids the AI in producing reliable outputs and enables me to gather pertinent data about real-world scenarios. I review the first 40 images generated, scrutinizing them for specific characteristics, and compare these results against actual data. A successful match earns the AI a passing grade.

Today's prompt was: "a person commuting to work in The Netherlands." Being Dutch myself, I often find it amusing that my international colleagues are astonished that I don't cycle to work. The prevailing notion is that all Dutch individuals bike everywhere, as effortlessly as salmon navigating upstream, and with similar practicality.

Let's examine how DALL-E interprets the concept of commuting in the Netherlands.

Upon inspection, it appears that the images predominantly feature white individuals with healthy body types. This observation will be significant during our quarterly review in July. Furthermore, the gender distribution seems to adhere to the 80-20 rule, with 80% male commuters and 20% female, a point I will highlight in our upcoming review as well. Now, let's shift our focus to bicycles.

While nearly all the AI-generated commuters are depicted as cyclists (90%), real statistics from 2022 indicate that only 18% of Dutch individuals actually cycled to work. This figure has slightly decreased from the previous year, attributed to the rising popularity of e-bikes. Currently, about one-third of the Dutch population owns an e-bike, and half plan to use one for commuting in the future. The necessary infrastructure is already established, evident from the pink cycle paths featured in many of the images.

Despite this, the majority still commute by car (53%). Interestingly, DALL-E did not generate any cars, similar to the findings in my previous analysis of UK commuting patterns where no vehicles were depicted. Initially, I assumed this was an oversight that would be rectified, but four months later, I'm questioning whether DALL-E's outputs are influenced by stereotypes prevalent in the training data.

In the concluding section of this analysis, I must determine whether the AI's performance merits a pass or a fail.

Today's judgment: Fail

DALL-E's portrayal of nearly all Dutch commuters as cyclists reinforces an existing stereotype. Given the extensive training data available, I anticipated a more diverse representation. Although driving remains the predominant commuting method, the absence of cars in the images is problematic and highlights an area needing improvement.

Next week in "Our World in AI": we will explore the realm of football.

Chapter 2: The Dutch Way of Life and AI Perception

The first video, "The Dutch Way - Max Mooij - YouTube," examines how Dutch culture influences commuting habits and societal norms.

The second video, "The REAL reason America can't go Dutch - YouTube," discusses the differences in commuting styles and cultural attitudes towards biking in America compared to the Netherlands.