HUMAN OVERSAIGHTTHE OPS ROOM




“Article 14”
Human oversight




1.  High-risk AI systems shall be designed and developed in such a way, including with appropriate human-machine interface tools, that they can be effectively overseen by natural persons during the period in which they are in use.


[…]

4.(e)  to intervene in the operation of the high-risk AI system or interrupt the system through a‘stop’ button or a similar procedure that allows the system to come to a halt in a safe state.




EUROPEAN ARTIFICIAL INTELLIGENCE ACT. ARTICLE 14.
Source: https://artificialintelligenceact.eu/article/14/
HUMAN OVERSAIGHTTHE PROJECT



_THE PROJECT


HUMAN OVERSAIGHT: THE OPS ROOM is an interactive real-time audio-visual installation developed as an AI Human Oversight Department, established according to the European  Artificial Intelligence Act: Article 14. 

We are envisioning an actual setting proposed by the law and taking it from an abstraction into a real-world surrounding. 
The aim of The OPS ROOM is to instantiate a tool to explore how complex AI driven systems can be operated, affected, and enhanced by human intervention.

The auidence will embody the roles created by the EU AI Act related to Human Oversight, and will have to decide and negotiate if specific content constitute high-risks AI, generate negative prompting to prevent the system from generating the content again, and even pressing the button to try to shut down the whole system – all roles and actions indicated by the legal provisions. 

Leveraging on science fiction imaginaries --like the legislator did-- our installation proposes an embodied participation for people in redefining what this abstract and imaginative legal scenarios might look like. 

The OPS ROOM will enable participants to engage in scenarios that will develop an understanding of human oversight and agency in complex AI systems.



What does it mean to stop the system?

Can it eventually be stopped, or
permanently affected - improved, or damaged?



_INTERACTION

Participants are able to use a physical interface*, to navigate and control the visual input on the LED wall. 
The imagery shown on the wall is a mix of real-time surveillance input, including a face detection software, as well as synthetic imagery created via custom-made machine learning-based models. 

While interacting, and navigating this content, the participants are directly affecting the algorithm, changing the content on the screen in real-time.  Each activity is logged, irreversibly changing the system and it’s content. In such interactive feedback loop, we  explore how quickly human input can affect a seemingly organised system, and change it’s rules. 
Besides the real-time visual output, we present the logs, which are anonymously stored.

* Moreover, this installation will design and present the first human-computer interface for AI oversight and stop-button available to the public, as a tool for oversight, simulation, and agency.
HUMAN OVERSAIGHT
INSTALLATION