@emollick The paper absolute does make this claim. Right in the conclusion:
"LLM ... convenience came at a cognitive cost, diminishing users' inclination to critically evaluate the LLM's output or ”opinions”... This highlights a concerning evolution of the 'echo chamber' effect..."
@MatthewJBar Back then we thought scaling parameters was all that was required and it truely felt like AGI was around the corner. Now we have these painfully slow "reasoning models" that don't actually reason, and agents that kinda work sometimes.