At The Markup, we’re proud of all the great investigations we published in 2023, and of their impact. 

But we are a relatively modest-sized news organization, and hardly the only entity whose work challenges technology to serve the public good. Throughout the year, we see stories in other publications that turn us green with envy.

We sometimes share these articles in the moment through our personal and organizational social channels, but, taking a cue from The 74 and Bloomberg Businessweek, we thought it would be worth gathering them on one web page here at the end of the year.

As you start thinking about 2024, it’s worth a look back at an incredible, head-spinning year in the world of technology, told through some of the best scoops, analyses, and narrative stories  from our colleagues throughout journalism.

This in-depth investigation showed how outsourced Kenyan laborers have been critical to the success of OpenAI, the Silicon Valley startup valued at tens of billions of dollars thanks to its breakout hit product ChatGPT. The story helped the public understand the importance of a largely invisible workforce of “data labelers,” who are tasked with reviewing some of the most disturbing content on the internet. Even cutting-edge artificial intelligence, it turns out, does not work without the help of many human beings.

This project showed how generative AI literally shows us a distorted reality that amplifies racial and gender stereotypes. A Bloomberg team prompted the AI image generator Stable Diffusion with the names of seven jobs considered high paying and seven considered low paying. The images returned showed men and people with lighter skin tones dominated the results for high-paying jobs; people with darker skin were more commonly shown on the low paying job searchers and women were underrepresented except in certain low-paying fields.

Another excellent visual investigation into AI and bias. In this story, Rest of World gave the image generator Midjourney prompts combining various countries across three continents with the concepts of “a person,” “a woman,” “a house,” “a street,” and “a plate of food” in those places. The output showed how starkly the tool flattened a diverse world into stereotypes.

A team at Wired—including Markup alum Todd Feathers—used public records to show how internet filtering software prevents children from learning about a wide range of important subjects on school-issued laptops. Given that the investigation focused on schools in a single city, Albuquerque, it presented a surprising quantity of damning detail, in part because the district’s censorship has been so extensive, blocking content more than 13 million times in less than two years.

This graphical look at changes in how the U.S. census collects information on the race and ethnicity of its population spans centuries. It also features a nuanced look at the reasons for these changes and the controversies they have sparked—weaving a surprisingly compelling story about data collection.

As we at The Markup have learned, It can be hard to explain how the intricacies of software algorithms affect people in their day to day lives. Writing on his independent Substack publication, BIG, economic policy researcher Matt Stoller connected the Federal Trade Commission’s antitrust complaint against Amazon to a broader pattern in which pricing algorithms and analytics services seem to be enabling collusion in a number of industries and thus drove inflation in 2021 and 2022. When it comes to the impact of software, it doesn’t get much more bread and butter than that.

You may have heard about the students at Westfield, New Jersey’s high school who earlier this year reportedly made deepfake pornography of more than 30 girls at the school. It made national headlines; made us all gasp and shake our heads at the dangers that today’s young people face at the hands of AI.

But chances are you did not hear what happened next. One of the girls who was targeted, Francesca Mani, stood up and spoke out. In an incredible display of poise and courage (she’s 15!), she turned a horrific, personal violation into an opportunity to fight for a solution to the problem of deepfakes. Her advocacy has gotten the attention of state and federal lawmakers, and resulted in a bill in the New Jersey legislature. Almost as impressive as Francesca and her family’s courage was the reporter, Tate Ryan-Mosley’s work in telling a hugely important story with tremendous sensitivity and nuance.

If you’re ever told a product you’re buying is “carbon neutral,” you should immediately be suspicious. Over the last few years, journalists and researchers have increasingly found that the forestry and agricultural (and in some cases, oceanographic) projects behind carbon offset schemes have a number of problems. This story charts the idea of offsetting from its humble, well-intentioned beginnings to a massive commercial operation where—ready for a surprise?—greed intervenes, and the communities affected by climate change rarely benefit. At the center of it you’ll find wealthy white guys engaged in an elaborate project to monetize an African forest.

As machine learning and generative artificial intelligence continue to permeate our civic lives, reporting on how they fail the most at-risk people is critical. Johana Bhuiyan builds this story around the experience of an Afro-Indigenous asylum seeker from Brazil who was unable to make himself understood to U.S. immigration officials, who had been instructed to use automated translation tools that weren’t up to the task. Bhuiyan digs into the technical details of how these tools fall short and steps back to map out the bureaucratic constraints that led to their use, showing us that this is one thread in a knot of systemic issues at the border.

Aiming to publish “society-shaping technology journalism,” 404 launched in August following the bankruptcy of VICE Media, where 404’s founders worked on the widely-read Motherboard vertical. The website and media company has since published at an impressive clip, regularly reporting out stories that catch our eye and make waves in the tech world and beyond. Just as importantly, the whole venture has added to a growing pile of evidence that reader support is a viable model for investigative- and beat journalism.