OpenAI & Anthropic vs. China Kopien
In this episode, we contextualize the development of AI in the digital ecosystem and e-commerce, beginning with a discussion of how frontier models like OpenAI and Anthropic differ from distilled models. We explain that distillation involves training a smaller, less expensive model with the output of a larger one and discuss why this is technically difficult to prevent. We then discuss the economic aspects of these models and whether the high valuations of OpenAI and Anthropic are justified. In our view, these valuations are overly ambitious because the costs of new models are rising sharply, many use cases don't require AI-generated intelligence (AGI), and less expensive models are often sufficient. Furthermore, many users and companies have limited willingness to pay. Another key focus is who ultimately controls the value chain. We discuss the hypothesis that large platforms like Google, Apple, and Microsoft could ultimately benefit more as distribution and app store providers, while the model providers come under pressure. This also involves considering the small number of paying users and the importance of a better user interface for AI applications. We also discuss a short-term access freeze to a new anthropic model and what this reveals about dependencies on US technology. We see this primarily as an example of how quickly access to key tools can be restricted and why European capabilities are becoming increasingly important. Preparation document by Julian: https://www.kassenzone.de/wp-content/... Partner in this episode: https://linktr.ee/kassenzone Community: https://kassenzone.de/discord Feedback on the podcast? Email [email protected] Disclaimer: https://www.kassenzone.de/disclaimer/ "Kassenzone" is marketed by Podstars by OMR. Interested in advertising on "Kassenzone"? Visit https://podstars.de/kontakt/?utm_sour... Alexander Graf: / alexandergraf / supergraf YouTube: / kassenzonede Blog: https://www.kassenzone.de/ E-Commerce Book 2019: https://amzn.eu/d/5Adc1ZH Platform Book 2024: https://amzn.eu/d/1tAk82E
