FT Tech Tonic Superintelligent AI: can chatbots think?
27 snips
Nov 28, 2023 
Prof. Emily Bender and Pablo Arredondo discuss the intelligence of generative AI systems like ChatGPT and their role in legal reasoning. The limitations of chatbots in language understanding and the concept of 'stochastic parrots' are explored. The challenges of AI systems in explaining reasoning and the debate on AGI funding and anthropomorphization are also discussed.
AI Snips
Chapters
Transcript
Episode notes
Clever Hans Effect Explained
- Clever Hans the horse appeared to do math by tapping his hoof but was actually responding to subtle cues from his trainer.
- This is known as the "clever Hans effect," where correct answers correlate with unintended signals, not real understanding.
GPT-4 Aces Bar Exam
- GPT-4 passed the US bar exam, demonstrating impressive legal reasoning and understanding complex cases.
- It even identified nuanced evidence admissibility in hypothetical trials, showcasing advanced application of legal rules.
GPT-4 Detects Fraud Nuances
- GPT-4 analyzed Enron scandal documents and detected nuanced references to fraud, like a Sesame Street cookie analogy.
- It excelled in document review tasks, spotting hidden meanings beyond simple keyword searches.