24 Nov 2021
At the "KI Lectures" Frauke Kreuter shows how AI and Big Data are used to automate administrative processes and study demographic changes - and what the pitfalls are.
Algorithms can also harbor biases, says Prof. Dr. Frauke Kreuter. And they do so when they learn them from historical data. | © LMU
As part of LMU’s virtual “KI Lectures” series, statistician Professor Frauke Kreuter, Chair of Statistics and Data Science in Social Sciences and the Humanities, spoke about use cases for artificial intelligence in economics and the social sciences and the relationship between Big Data and AI applications. She pointed out that large pools of data alone are not enough to make algorithms intelligent.
“Historical datasets that are used to train AI applications often contain data that do not represent the whole of society,” says Kreuter, summarizing one of the challenges of using artificial intelligence in the social context. One example she cites is the almost complete lack of non-binary gender identities in datasets to date.
The public sector is one area where Professor Kreuter sees great potential for AI applications to be used. It’s not about getting rid of jobs, says the LMU statistician, it’s about making better use of the resources we have, making people’s jobs easier, and doing more to ensure that decision-making processes facilitate equal opportunities.
“Companies are already using AI applications for processes that take a long time, such as checking documents,” explains Kreuter. “In the future, self-learning algorithms could be able to supply decision documents that can be used to offer unemployed people suitable measures to get back into the job market.” But, she says, this would require a paradigm shift on data privacy, along with innovations in the infrastructure for research data, in order to create a reliable and freely available data basis.
91 Min. | 24 Nov 2021
The next KI Lecture will take place on November 30 with Prof. Dr. Enrique Jiménez. Register now!
To the program of the KI Lectures