ALLAI input to EU Apply AI Strategy
The Apply AI strategy centers its vision for enhancing EU competitiveness on the premise that AI (particularly generative AI) will be a driver of productivity gains through automation, optimization, and improved decision-making. Specifically, it argues generative AI will increase efficiency and enable new types of products, services, and business models. As such, the strategy places significant emphasis on the integration of generative AI into key sectors, suggesting that such integration could ‘unlock’ up to EUR 3.4 trillion in economic value by 2030. This figure is used to justify the widespread deployment of generative AI technologies across public and private sectors. We argue that these premises provide too narrow a scope for a successful and sustainable Apply AI Strategy and suggest a number of elements that the Apply AI Strategy should at minimum address.
A realistic view of AI’s capabilities and limitations
First, the Apply AI Strategy should address the critical need for alignment of the capabilities and limitations of state-of-the-art AI with the actual application demands. The call for evidence assumes AI (particularly generative AI) to possess uncontested capabilities for full scale deployment. It should however address the question of suitability of AI for a desired purpose, the level of maturity of current AI, and require a critical assessment of its multiple limitations.
Grounded in Trustworthiness
Second, the Apply AI Strategy should be firmly grounded in the EU’s ethical, legal and societal vision for trustworthy AI: AI that is lawful, ethically aligned, and socio-technically robust[1]. This requires an underpinning of AI applications with the Ethics Guidelines for Trustworthy AI, their compliance with the AI Act and other applicable regulation and alignment with what is to be considered socially acceptable and technically robust.
Address Job Displacement, Prioritize Public Interest and Societal Well-being
Third, the risks associated with labor automation and displacement must be mitigated by prioritizing the public interest, ensuring that technological progress enhances societal well-being and long-term sustainable innovation. Additionally, the strategy must address data inclusion, ensuring that the datasets driving AI development are representative, diverse, and equitable.
Address the EU AI Divide
We believe there is an additional need to acknowledge and close the existing divide in AI capacity and access across EU Member States. If AI is to be applied to all sectors without supporting those Member States that lack robust verticals, this gap could be entrenched and undermine the EU’s collective strength. Adjacently, we recommend the strategic use and streamlining of the EU’s existing AI ecosystems, infrastructures and initiatives to help in this end.
Focus beyond Generative AI
From a technological standpoint, we argue that a shift away from an exclusive focus on generative AI is necessary to future-proof the Apply AI strategy. Specifically, we believe there should be greater attention to other AI paradigms that may be more effective and suitable for certain sectors or contexts.
Read the full paper by ALLAI President Catelijne Muller and junior associate Alice Teilhard De Chardin here.
