関連インシデント
インシデント 211 Report
Tougher Turing Test Exposes Chatbots’ Stupidity (migrated to Issue)
インシデント 293 Report
Image Classification of Battle Tanks
インシデント 422 Report
Inefficiencies in the United States Resident Matching Program
インシデント 621 Report
Bad AI-Written Christmas Carols
インシデント 851 Report
AI attempts to ease fear of robots, blurts out it can’t ‘avoid destroying humankind’
インシデント 1591 Report
Tesla Autopilot’s Lane Recognition Allegedly Vulnerable to Adversarial Attacks
インシデント 2871 Report
OpenAI’s GPT-3 Reported as Unviable in Medical Tasks by Healthcare Firm
インシデント 2981 Report
Student-Developed Facial Recognition App Raised Ethical Concerns
The following former incidents have been converted to "issues" following an update to the incident definition and ingestion criteria.
21: Tougher Turing Test Exposes Chatbots’ Stupidity
Description: The 2016 Winograd Schema Challenge highlighted how even the most successful AI systems entered into the Challenge were only successful 3% more often than random chance.
Why Downgraded? This is an academic finding showing a weakness of the technology rather than a harm event.
Former Reports: These reports were formerly associated with the incident.
Migrated Reports: Reports associated with this incident are now associated with other incidents as context.
- none
62: Bad AI-Written Christmas Carols
Description: Janelle Shane, an AI research scientist, used 240 popular Christmas carols to train a neural network to write its own carols.
Why Downgraded? Was designed to be humorous and is in fact humorous.
Former Reports: These reports were formerly associated with the incident.
Migrated Reports: Reports associated with this incident are now associated with other incidents as context.
- none
159: Tesla Autopilot’s Lane Recognition Allegedly Vulnerable to Adversarial Attacks
Description: Tencent Keen Security Lab conducted security research into Tesla’s Autopilot system and identified crafted adversarial samples and remote controlling via wireless gamepad as vulnerabilities to its system, although the company called into question their real-world practicality.
Why Downgraded? The reports surface a vulnerability with projected harms rather than a report of harms in the real world.
Former Reports: These reports were formerly associated with the incident.
- Tencent Keen Security Lab: Experimental Security Research of Tesla Autopilot
- Three Small Stickers in Intersection Can Cause Tesla Autopilot to Swerve Into Wrong Lane
Migrated Reports: Reports associated with this incident are now associated with other incidents as context.
- none
287: OpenAI’s GPT-3 Reported as Unviable in Medical Tasks by Healthcare Firm
Description: The French digital care company, Nabla, in researching GPT-3’s capabilities for medical documentation, diagnosis support, and treatment recommendation, found its inconsistency and lack of scientific and medical expertise unviable and risky in healthcare applications.
Why Downgraded? The reports indicate the insufficiency of ChatGPT for several tasks for which the system was not deployed in the real world.
Former Reports: These reports were formerly associated with the incident.
- Doctor GPT-3: hype or reality?
- Researchers made an OpenAI GPT-3 medical chatbot as an experiment. It told a mock patient to kill themselves
- Medical chatbot using OpenAI’s GPT-3 told a fake patient to kill themselves
- This bot actually suggests patients to kill themselves
Migrated Reports: Reports associated with this incident are now associated with other incidents as context.
- none
298: Student-Developed Facial Recognition App Raised Ethical Concerns
Description: TheFaceTag app, a social networking app developed and deployed within-campus by a student at Harvard raised concerns surrounding its facial recognition, cybersecurity, privacy, and misuse.
Why Downgraded? The harms are predicted to occur, but have not yet occured.
Former Reports: These reports were formerly associated with the incident.
- Can I Scan Your Face?
- A Harvard freshman made a social networking app called 'The FaceTag.' It's sparked a debate about the ethics of facial recognition
- An app by a Harvard student caused controversy about its ethics
Migrated Reports: Reports associated with this incident are now associated with other incidents as context.
- none
85: AI attempts to ease fear of robots, blurts out it can’t ‘avoid destroying humankind’
Confirmed migration by editors - Incident 85 - likely does not fit current incident criteria, where GPT-3 was manipulated by human editors to exaggerate harm, borderline sensationalism. https://incidentdatabase.ai/cite/85
Description: On September 8, 2020, the Guardian published an op-ed generated by OpenAI’s GPT-3 text generating AI that included threats to destroy humankind.
Why Downgraded? Unclear who was harmed, if anyone, via the events described.
Former Reports: These reports were formerly associated with the incident.
Migrated Reports: Reports associated with this incident are now associated with other incidents as context.
- none
Candidates for Migration
The following incidents may also be migrated in the future based on discussion among the AI Incident Database editors:
Does not meet current definition and criteria https://incidentdatabase.ai/cite/42
Incident 29 concerns the "tank story," which may be apocryphal. https://incidentdatabase.ai/cite/29