Aleph Alpha is Europe’s answer to OpenAI.
Aleph Alpha is Europe’s answer to OpenAI.
Europe’s Quest for a Homegrown AI Champion
Europe is determined to create its own Open AI, tired of regulating American tech giants from a distance. The continent wants to build its own generative AI, and all eyes are on Jonas Andrulis, a laid-back German entrepreneur with a carefully groomed goatee. As skepticism lingers over whether his company, Aleph Alpha, can truly compete with the likes of Google and OpenAI, many in the European Union are hopeful that Aleph Alpha can challenge American dominance in generative AI. The EU has a history of clashes with US tech giants over privacy and data security, and some Europeans believe that their values significantly differ from those in Washington DC.
Andrulis assures that his company is not a nationalist project, with Americans also working at Aleph Alpha. However, he is proud to be at the forefront of Europe’s AI efforts. Having spent three years working on AI at Apple, Andrulis left in 2019 to explore the technology’s potential outside of a large corporation. He founded Aleph Alpha in Heidelberg, Germany, with the goal of developing large language models (LLMs), a type of AI that generates text and analyzes documents by identifying patterns in human language.
Currently, Aleph Alpha has thousands of paying customers, including banks and government agencies, who use their LLM to write financial reports, summarize documents, and build specialized chatbots. The challenge for Andrulis is to make the AI customizable, allowing businesses to have control over its behavior. Even if a large international bank wants a chatbot that is sarcastic and insulting, Andrulis believes they have every right to it.
However, LLMs are just the beginning for Aleph Alpha. Their ultimate aim is to build artificial general intelligence (AGI), an artificial, humanlike intelligence that can be applied to a variety of tasks. The company’s success so far, with claims of 10,000 customers, indicates that it can compete with emerging giants in the field. According to Jörg Bienert, CEO of the German AI Association, this demand validates the need for developing and providing these models in Germany, particularly for governmental institutions that prefer solutions developed and hosted in Europe. The concern over data privacy and exclusion of European languages from AI developments are additional reasons why European AI is essential.
To address the issue of AI systems providing incorrect information, Aleph Alpha emphasizes transparent decision-making. Andrulis demonstrates how their AI explains its decisions by clicking on specific words in a generated sentence or an image, tracing what informed the AI’s response. This level of explainability is novel, even to AI experts. Nicolas Moës, director of European AI governance at the Future Society think tank, believes that these features could become more prevalent once the EU passes the AI Act, which is expected to include transparency requirements.
Some doubt Aleph Alpha’s underlying technology, believing it is not as advanced as its American competitors. Matthias Plappert, an AI consultant in Berlin, points out that Aleph Alpha’s model does not perform better than its counterparts in standardized tests. Nevertheless, many Europeans remain adamant about the need for a viable European contender in the AI industry. Apart from economic reasons, Europeans argue that European companies are more likely to prioritize privacy and avoid discrimination.
The concept of “European values” repeatedly surfaces when discussing the preference for AI made in Europe. Europeans fear that AI developed by US companies may not align with their values. Andrulis references a past incident where Facebook removed an image of Michelangelo’s David due to nudity, highlighting a disagreement in values. However, Andrulis acknowledges that it is not his role to determine how European values should be translated into AI. His focus is on building excellent, transparent, and controllable technology.
In order to cater to regulated industries with strict data requirements, Aleph Alpha established its first data center in Berlin. This move ensures customers in industries like government and security that their sensitive data remains hosted within Germany.
While doubts persist, Europe’s determination to create a homegrown AI giant, rooted in the continent’s values, is unwavering. The hope is that stricter regulations will prove beneficial, allowing European AI companies to build better products and establish a standard of quality, similar to other tightly regulated European industries. Europe wants to be an active participant in the AI revolution, shaping its future and ensuring that European values are embedded in this era-defining technology.
