Half of LLM users (49%) think the models they use are smarter than they are, including 26% who think their LLMs are “a lot smarter.” Another 18% think LLMs are as smart as they are. Here are some of the other attributes they see:
- Confident: 57% say the main LLM they use seems to act in a confident way.
- Reasoning: 39% say the main LLM they use shows the capacity to think and reason at least some of the time.
- Sense of humor: 32% say their main LLM seems to have a sense of humor.
- Morals: 25% say their main model acts like it makes moral judgments about right and wrong at least sometimes. Sarcasm: 17% say their prime LLM seems to respond sarcastically.
- Sad: 11% say the main model they use seems to express sadness, while 24% say that model also expresses hope.
LLMs are smart in the way someone is smart who has read all the books and knows all of them but has never left the house. Basically all theory and no street smarts.
Bot even that smart. There a study recently that simple questiona like “what was huckleberry finn first published” had a 60% error rate.
yeah my analogy is not so good… LLMs suck with factual stuff, they are better with coding or languages (Claude has been really helpful to me with Estonian).
They’re not even that smart.
Well yes, they are glorified text autocomplete, but they still have their uses which could be considered “smart”. For example I was struggling with a programming thing today and an LLM helped me out, so in a way it is smarter than me in that specific thing. I think it’s less that they are dumb and more that they have no agency whatsoever, they have to be pushed into the direction you want. Pretty annoying…
A broken clock is right two times a day I suppose.