The Savant Syndrome: Is Sample Recognition Equal to Intelligence? | by Salvatore Raieli | Oct, 2024
|LLM|INTELLIGENCE|REASONING|
Exploring the boundaries of synthetic intelligence: why mastering patterns might not equal real reasoning
I’ve hardly recognized a mathematician who was able to reasoning. — Plato
Reasoning attracts a conclusion, however doesn’t make the conclusion sure, except the thoughts discovers it by the trail of expertise. — Roger Bacon
Large Language Models (LLMs) have proven outstanding capabilities, particularly for classical duties in natural language processing (akin to query answering). Surprisingly, they confirmed enchancment in complicated duties requiring reasoning (akin to coding and arithmetic). These capabilities have lengthy been thought of unique to people. So claiming that LLMs can resolve duties that require reasoning has opened a heated debate.
Can Giant Language Fashions (LLMs) actually purpose? Or are they only subtle sample matchers?