AI, over 70 years old, has seen periods of intense interest followed by disinterest. After a lull in innovation, AI has recently entered a new era marked by breakthroughs such as autonomous vehicles, virtual assistants, generative AI models…
Artificial intelligence promises a better world, but this vision often clashes with the harsh reality of incidents caused by these “intelligent” systems, rapidly deployed in our environment without adequate preparation, both individually and at the state level.
That’s where the AI Incident Database (AIID) comes in, a true collective memory cataloging bugs, crashes, and other AI incidents. It already lists over 1000 cases! Some are relatively harmless, while others have more significant impacts…
Here are a few examples of incidents:
- A chess robot at a tournament in Russia broke the finger of a child who reached onto the board before the robot had completed its move
- Popular Female Twitch Streamers Targeted In Deepfake Pornography Scandal
- Tom Hanks alleges dental company used AI version of him for ad: ‘Beware!!’
- …
In the realm of AI incidents, there’s a diverse range to explore, making this inventory quite intriguing. This database expands daily with community reports, open for contributions from anyone submitting new incidents.
The objective of the AIID is to glean insights from these errors in order to construct safer AI systems, mirroring the practices observed in critical sectors such as aerospace and cybersecurity.
Leave a Reply