Author:
Paaß Gerhard,Giesselbach Sven
Abstract
AbstractFoundation Models emerged as a new paradigm in sequence interpretation that can be used for a large number of tasks to understand our environment. They offer the remarkable property of combining sensory input (sound, images, video) with symbolic interpretation of text and may even include action and DNA sequences. We briefly recap the process of pre-training, fine-tuning or prompting of Foundation Models and summarize their main properties. For the different application areas presented in the book, we summarize the performance levels of the models and delineate different promising economic applications. A section is devoted to discussing the potential harm that can be caused by Foundation Models, including bias, fake news, but also possible economic monopolies and unemployment. There is an urgent need for a legal regulation of the construction and deployment of these models. The last section considers advanced artificial intelligence systems and the shortcomings of current systems. Foundation Models have significantly improved performance in recent years and have the potential to reduce the gap to a truly general AI.
Publisher
Springer International Publishing
Reference104 articles.
1. M. Abadi, A. Chu, I. Goodfellow, H. B. McMahan, I. Mironov, K. Talwar, and L. Zhang. “Deep Learning with Differential Privacy”. In: Proc. 2016 ACM SIGSAC Conf. Comput. Commun. Secur. 2016, pp. 308–318.
2. M. Ahn et al. Do As I Can, Not As I Say: Grounding Language in Robotic Affordances. Aug. 16, 2022. arXiv: 2204.01691 [cs].
3. F. Alam et al. “A Survey on Multimodal Disinformation Detection”. 2021. arXiv: 2103.12541.
4. W. An, Y. Guo, Y. Bian, H. Ma, J. Yang, C. Li, and J. Huang. “MoDNA: Motif-Oriented Pre-Training for DNA Language Model”. In: Proc. 13th ACM Int. Conf. Bioinforma. Comput. Biol. Health Inform. BCB ’22. New York, NY, USA: Association for Computing Machinery, Aug. 7, 2022, pp. 1–5. isbn: 978-1-4503-9386-7. doi: https://doi.org/10.1145/3535508.3545512.
5. W. Apt and K. Priesack. “KI und Arbeit – Chance und Risiko zugleich”. In: Künstliche Intelligenz: Technologie — Anwendung — Gesellschaft. Ed. by V. Wittpahl. Berlin, Heidelberg: Springer, 2019, pp. 221–238. isbn: 978-3-662-58042-4. doi: https://doi.org/10.1007/978-3-662-58042-4_14.