There is a dark side to today's shiny AI revolution. That was the warning from digital rights advocates and activists gathered in Berlin for the re:publica technology conference. 

"New patterns of discrimination are emerging from AI and algorithms," said Ferda Ataman, Germany's federal anti-discrimination commissioner. She warned that AI technology, if left unchecked, could exacerbate discrimination by perpetuating existing biases. 

The recent rise of AI technology has led to remarkable advances, from innovative ways to diagnose and treat cancer to new AI systems that allow anyone, regardless of technical expertise, to create images and videos from scratch. 

Less visible, however, are the significant downsides of the technology, several experts warned during the conference — and vulnerable communities on the margins of society and in the Global South would bear the brunt of these consequences first.

AI surveillance 

The way artificial intelligence is being used to take surveillance to unprecedented levels illustrates this, according to Antonella Napolitano, a technology policy analyst.

In a report published last year by the human rights organization EuroMed Rights, Napolitano argued that as part of a broader trend to "outsource" control of its external borders, the European Union is funding companies that test new AI-based surveillance tools on migrants in North Africa, whose privacy is not as protected as that of EU citizens. 

"It's easier for companies to test different technologies on these people without facing any consequences," Napolitano told DW. 

"But the consequences are real," she said, adding that AI-assisted surveillance can push migrants into more dangerous migration paths. 

'Data colonialism' 

Reports like these have led some digital rights advocates to draw parallels with the history of colonialism from the 15th century onwards, when European powers began to exploit large parts of the world. 

Another aspect of this new "data colonialism" is that many Western companies have turned to places in the Global South to collect and annotate data, according to Mercy Mutemi, a Kenyan lawyer specializing in digital rights. 

"History is repeating itself," Mutemi told DW. 

Most of today's cutting-edge AI systems rely on vast amounts of data. To meet that demand, Western companies are extracting data from individuals across Africa on a massive scale, often without consent and with little to no benefit in return, said Mutemi, who is involved in about a dozen ongoing cases challenging the role of Western tech companies in her home country. 

This phenomenon is likely to intensify in the coming years as AI technology's demand for data continues to grow, she added. 

Carbon footprint 

The breakneck pace of AI development raises concern about its environmental impact, too.  

Training and running complex AI models requires massive computing power in data centers, which are under scrutiny for their high energy consumption, water usage, and greenhouse gas emissions.

Already today, data centers account for over two percent of global electricity consumption, according to estimates by the International Energy Agency. 

The emergence of particularly energy-intensive AI tools will drive energy consumption even higher, according to Canadian author and technology critic Paris Marx. Marx warned of an "AI-fueled data center boom" driven by the commercial interests of a few powerful tech companies.  

"There has been a huge investment to build many more data centers in places around the world," Marx told the conference. 

 Edited by Rina Goldenberg