Issue No. 300

 

Earlier this year, a clip of an interview with Peter Thiel, a billionaire founder of the US cyber-arms and tech firm Palantir, went viral. In it, he was asked whether he would prefer the human race to endure, and Thiel pauses before stuttering through a half-answer in the affirmative. For those following Thiel and his brand of ultra-conservative Christian politics blended with surreal transhumanist tech-based visions of a future, though undoubtedly creepy, it came as little surprise. But Thiel and his ilk are the ones now directing much of our future, a handful of nearly exclusively white North American men, wealthy, vainglorious and battling out the future of humanity with their egos in the race for Artificial Intelligence (AI) and data supremacy.

We are living in an age of conquest, with land, gold, oil, and natural resources all up for grabs on the frontlines in this shifting multi-polar world order. Data is often overlooked, yet it is undeniably the defining resource of our age —the substance upon which the ever-more powerful learning machines and AI software are being built. Already remaking our world and economies, the speed and scale of change are likely to dwarf the Industrial Revolution, but remain poorly understood and regulated by most populations and governments. Everything from medical research to children's homework to artwork is being permanently changed, rewritten by a handful of mega-powerful companies predominantly based in San Francisco's Silicon Valley. Against this backdrop, the US-China tech war is advancing at a rapid pace, with Beijing investing in the African market as an offset against its San Francisco competitors.

In the AI and data wars, immense quantities of power and water are required to feed and cool kilometre-long data plants responsible for processing billions upon billions of requests every hour. By 2026, it is expected that global electricity consumption of data centres will approach 1,050 terawatt hours-- making them the 5th largest consumer ranked against countries, sandwiched between Japan and Russia. Looking to cut costs for such plants, Ethiopia has been increasingly touted as one location for tech companies, particularly following the inauguration of the hydroelectric Grand Ethiopian Renaissance Dam (GERD) earlier this month, which is nearly doubling the country's power output by 5,150 megawatts. The image of Silicon Valley techbros in Benishangul-Gumuz may jar the mind somewhat, but several companies have now set up shop in Ethiopia —particularly Chinese —to begin mining data and Bitcoin, the first and pre-eminent decentralised cryptocurrency. 

Last year, a memorandum of understanding was signed between a subsidiary of the Hong Kong-based West Data Group and Ethiopian Investment Holdings for USD 250 million to develop data mining and AI infrastructure. Lured by the promises of cheap power —and that Beijing banned crypto trading and data mining in 2021 —Ethiopia has become one of the favoured destinations, alongside Kenya. Yet the influx of these rapacious tech companies, which will gobble up soaring Ethiopia's energy production, contrasts with the government's rhetoric of the GERD dispensing electricity to 40% of Ethiopia's 120 million population that are without it. However, Addis is in dire need of foreign exchange following major economic reforms and is attempting to position itself as an attractive destination for foreign tech companies to invest, despite internal instability.

Amidst the dizzying scale of societal and economic change driven by the AI race, some have compared it to a form of 'digital colonialism.' Cheap labour forces in the developing world are being used to train AI on particular data sets, while mind-bogglingly wealthy companies like Meta are reaping the benefits. Even the profits from the data gleaned from African populations are stored in the US, while internet data is routinely routed through Europe. But when reports emerged of Kenyan workers being used to trawl through Facebook and Instagram to assess the cacophony of violent and abusive images on the platforms, Meta simply shut the factories. Neither Addis nor Nairobi are in control of AI development, involuntarily ceding their digital sovereignty and economies to the US and China. It is a similar kind of relocation of wealth and resources of developing countries to the dominant world powers of the colonial period, but rather than gold or lumber, it is the information economy that is being transferred.

That is not to say there are no benefits of AI or data for Ethiopia or the Horn more broadly. Rather, there are immense rewards to be gleaned in developing countries from AI as well, such as dramatically slashing costs for medical diagnoses, with models able to trawl every research paper ever published on cancer to advise a particular course of treatment. In Ethiopia and the Horn, where the climate crisis exacerbates food insecurity, AI models can provide guidance on crop selection during drier seasons and recommend actions for farmers using minimal fertilisers. And given the inefficiencies in any government, one might think that the bureaucratic-heavy Ethiopian state machinery might be well-primed for some AI productivity. There is a painful irony, however, that the data centres upon which AI models rely are some of the most energy-demanding infrastructure in the world. Countries with the least say over AI and tech innovation are likely to be most impacted by some of the societal transformations it drives, including accentuating the climate crisis.

Equally concerning is the political threat posed by such technologies, which lack sufficient guardrails against the spread of misinformation and hate speech that can be amplified by the generative texts and images of AI created by bad-faith actors. There is a chequered history of mega-tech companies in the Horn, not least Facebook, which has been accused of failing to act sufficiently to prevent hate speech being disseminated during the Tigray war and other conflicts. The Ethiopian government and others within the Horn may not have played a role in developing these platforms and technologies, but they have absolutely seized upon them and deployed them with devastating effect as well. The 20th-century promise of the Internet and social media as a kind of 'great social-democratic equaliser' where all the world has equal access to information and analysis has long faded. However, the suppression of dissenting voices, similar to Beijing's parallel Internet, is now commonplace across the Horn, with democratic activists being arrested for expressing their opinions online or facing outright internet shutdowns.

There are numerous other ways in which these rapidly developing image and language models could be used. Some have raised the chilling prospect of AI being used to identify particular ethnicities for targeting. And amidst an authoritarian slide in nearly every Horn country, the prospect of 'smart'-- and importantly, cheap-- AI surveillance software is on the horizon, expanding beyond 'Pegasus,' the premier but extremely costly programme created by the Israeli cyber-arms company NSO Group. Meanwhile, incorporating AI into drones is something that Palantir is pioneering-- perhaps a new development that we will see on the battlefields of Sudan, or the Oromia and Amhara regions in Ethiopia in the coming years. Cheaper knock-offs have an unerring way of finding their way into the marketplace.

None of these hyper-rich 'tech bros' like Thiel appear to care much about the societal or political fallout of their inventions; they consider themselves 'disruptors', dismissive of the liberal democratic state, and too often harbouring alt-right views. Hubris and extreme capitalistic instinct are their drivers. But these innovations are already reshaping our world, for positive and negative. With the tech firms dismissive of such concerns, channelling these innovations for public good and curbing their inevitable harms will thus have to fall upon the consumers in this part of the world, be it governments, multilaterals, or civil societies.