At the TechBBQ conference in Copenhagen, the media claimed that when European investors, entrepreneurs and operators talked about AI, the focus was not just on products and applications, but rather on how to use it.Who controls models and infrastructureI don't know. This topic echoes the theme of this year's General Assembly “Emerging from Agency” and reflects the continued concern of the European scientific community over sovereignty over AI.
Model interrupts magnifying dependence on anxiety
The report mentions that the Anthropic AI model Mythos and Fable were once unable to provide services to users outside Europe earlier this year. This event brought a number of participants to a new look at whether Europe is “capable” in this round of AI waves, or simply renting it from the United States and China.
Senior executives of start-up companies indicated that the disruption of the model had clearly disrupted the pace of work of their software team; it was also felt that business could be maintained in the short term and that problems had not really erupted. But the general discussion on the floor went from “AI can do what” to “European capacity to decide what AI can do”.
The discussion has shifted to human judgment.
The angel investor, Novo Nordisk Foundation Cellerrator, said that this year’s discussion was more like asking how much judgement people would like to give to AI than simply making the machine smarter. In her view, the so-called AI proxy age was not just a process of machines gaining more autonomy, but a process in which humans had voluntarily given up part of the decision.
The report also mentioned that the President of Signal, Meredith Whittaker, in a discussion about privacy and the ability of AI to coexist, was clearly opposed to the deep integration of AI assistants and agents into the operating system. By name, she criticized practices such as ChatGPT access iMessage, arguing that the current AI development was developing a new data collection system and that laboratories and platforms were weakening external vigilance about the consequences of data through marketing.
She also stated that there was a clear market demand for privacy and that European users would remain interested in such products, especially in the context of rising sovereignty concerns.
Control of AI affects national capacity
In another discussion, Emad Mostaque, co-founder of Stability AI and Intelligent Internet, defined sovereignty as “the ability to resist external power to exert influence”. In his view, while currently AI powers were concentrated in the hands of a small number of laboratories, the future governance, economic and social functioning of States might be increasingly dependent on AI.
According to him, whoever controls AI may have a greater impact on a country's ability to operate. This judgement also raises broader questions: If intelligent agents take on more and more cognitive work, and robots take on more physical work, society may not simply embrace a technological upgrading, but rather a new discussion of ownership, labour, democracy and economic participation.
Finally, the report mentions that although the General Assembly also has topics such as women ' s health, cross-regional innovative cooperation and European growth capital, AI has been a bottom-up theme throughout the field. For the European scientific community, the key issue in the next phase is not just how to use AI, but how to retain more initiative at the modelling, data and infrastructure levels.