# AI Causing Chaos

> So much AI turns out to be low-waged people in a call center in the Global South pretending to be robots that Indian techies have a joke about it: “AI stands for ‘absent Indian’” - [Cory Doctorow](https://pluralistic.net/2024/01/31/neural-interface-beta-tester/)

I am collecting examples of AI fails by people who either deliberately or naively mislead about the capabilities of AI.

## 2024

- **11/02/24 -** [System set up to detect cheating students found 97% of students were cheating and UK officials just accepted this](https://www.theguardian.com/uk-news/2024/feb/11/english-test-scandal-students-renew-fight-to-clear-names-after-10-years)
- **09/02/24 -** [Gemini LLM conflates unsafe in the C# memory management context with unsafe as in unethical.](https://archive.jamesravey.me/archive/1707468320.73205/singlefile.html)
- **05/02/24** - [Using neural networks to generate images of fake ids for online verification](https://www.404media.co/inside-the-underground-site-where-ai-neural-networks-churns-out-fake-ids-onlyfake/)
- **04/02/24** - via [Yoav Goldberg](https://twitter.com/yoavgo/status/1753556086534070706) - [GEM using models to predict protected characteristics of job applicants](https://archive.jamesravey.me/archive/1707030622.46187/index.html) when such information is not supplied
- **28/01/24** [Dudesy Podcast claimed that they trained an AI to impersonate the late comedian George Carlin](https://arstechnica.com/ai/2024/01/george-carlins-heirs-sue-comedy-podcast-over-ai-generated-impression/) but [Cory Doctorow](https://pluralistic.net/2024/01/29/pay-no-attention/#to-the-little-man-behind-the-curtain) and [Simon Willison](https://www.theregister.com/2024/01/24/willison_ai_software_development/) both call bullshit. IT stinks of the "We made an AI watch 100 hours of X..." meme that went round a few years ago.
- **20/1/24** [DPD introduce a ChatGPT-powered bot which users jailbreak and get to swear and call DPD 'useless'](https://www.theguardian.com/technology/2024/jan/20/dpd-ai-chatbot-swears-calls-itself-useless-and-criticises-firm)

## 2023

- **20/12/23** [Chevrolet integrated a ChatGPT-powered bot into their site](https://gizmodo.com/ai-chevy-dealership-chatgpt-bot-customer-service-fail-1851111825) but because there is no such thing as input sanitization for [LLMS](https://wiki.jamesravey.me/books/working-with-llms "Working with LLMs") people used prompt injection attacks to make the bot agree to sell users cars for $1.
- **24/10/23** [Cruise robo-taxi dragged a person 20ft along the floor](https://www.wired.com/story/cruise-robotaxi-self-driving-permit-revoked-california/) and had their license to operate in San Francisco suspended.
- **17/01/23** it emerged that [a 2016 demo of Tesla's self-driving capabilities was completely staged](https://www.reuters.com/technology/tesla-video-promoting-self-driving-was-staged-engineer-testifies-2023-01-17/)
-