the_troy•19d agoProlific Poster

Unpopular opinion: I think the push for bigger AI models is missing the point

Honestly, I was building a chatbot for a small business in Austin and kept trying to use the biggest language model I could access. It was slow, expensive, and kept giving overly complex answers. The tipping point was when the owner said, 'I just need it to tell customers our hours, not write a poem about them.' Ngl, that made me realize we're often using a sledgehammer to crack a nut. Has anyone else found simpler, smaller models work better for straightforward jobs?

3 comments

3 Comments

piper_reed19d ago

That "sledgehammer to crack a nut" line is perfect. My buddy was trying to make a simple FAQ bot for his bike shop. He used this huge model and it would give these long, weird answers about the philosophy of cycling when someone just asked if they fixed flat tires. Switched to a tiny, cheap model that only knows his shop info and it works like a charm now.

james_kim19d agoMost Upvoted

But sometimes you need that big model... it can handle weird questions the small one just can't. A customer might ask something totally out of left field, and the fancy model actually gets it. The extra cost is worth it for covering all the bases.

mary77613d ago

My son tried to get one of those big models to explain his math homework last week. It gave him a three page essay on the history of algebra instead of how to solve for x. I see what you mean, @james_kim, about covering weird questions, but that kind of overkill is why my kid failed his quiz. Piper's bike shop story is exactly right. Most of the time, you just need the answer to the question you actually asked.