Disclaimer: This is not a universal test for strong AI!
Background
Three and a half years ago, just as new AI techniques were clearly about to propel generative AI into the public domain, a post ‘“ERIC”: How a Bad Think-Thing Destroyed the World’ was published on this blog. There was a threefold purpose:
- Because (it was felt) it made a decent enough point in its own right, but also …
- To get it established on the Internet and available for the initial OpenAI/ChatGPT cutoffs, and in particular …
- To serve as a simple test for levels of human-like cognition in emerging AI and generative AI.