Bertin Martens on The Incongruent Economics of AI скачать в хорошем качестве

Bertin Martens on The Incongruent Economics of AI

Follow the link for the full summary: https://bcf.princeton.edu/events/bert... Link to sign up for the webinar series: https://bcf.princeton.edu/markus-acad... On Thursday, December 5, Bertin Martens joined Markus’ Academy for a conversation on The Incongruent Economics of AI. Bertin Martens is a Senior Fellow at Bruegel. Introductory remarks by Markus Brunnermeier. A summary in four bullets The talk highlighted four key incongruencies in the economics of AI: (1) the tension between growing fixed costs of AI and the productivity growth it will deliver, (2) new questions in competition policy, (3) the challenge that centralized learning poses for individual data rights, (4) the EU’s AI policy dilemmas Even under the most optimistic scenarios investment costs will be a chokepoint for the development of AI. However developers are already working on shifting workloads from upfront investments to operational costs “Co-opetition deals” between startups and big tech both promote competition and promote the market power of big tech. Are these necessary to promote AI development? The EU faces several disadvantages in AI, including the absence of major tech companies and high regulatory and electricity costs. Leveraging open-source and “derived” LLMs could help unlock productivity gains Timestamps: [0:00] Markus’ introduction [9:52] Exponential cost growth and the productivity growth from AI [27:50] Three challenges in competition policy [50:35] Growing tension between the private and social value of data [1:01:32] EU AI Policy Options [1:08:58] Non-economic drivers of AI