6 ms·
How well does that work? I would imagine recipes iate really hard to entrust to chatgpt. At least for baking, where ratios and timing is crucial. LLMs have very
by Too 3y ago
How well does that work? I would imagine recipes iate really hard to entrust to chatgpt. At least for baking, where ratios and timing is crucial. LLMs have very bad sense of numbers generally. How would you verify it’s not all bogus hallucinations? At best it’s a source of inspiration, for ingredients you can freestyle from yourself.
- Roark66 3y agoI didn't mean recipes specifically, but this type of question where I'm looking for a very specific short answer. For example, I don't know if some software has some specific feature, or I ask, write me a short python script fragment that does this or that. Or how do I reverse a git commit that was pushed etc. Or what is the thermal conductivity of kingpin extreme paste. All are very specific things years ago I would use Google to search for, but today Google is almost useless on this. It does hallucinate sometimes, it writes programs using non existent apis etc, but I've found asking it "is what you said correct" or "is it true?" or "how did you came up with this answer" tend to produce replies like : I'm sorry, this is wrong, this is good instead. On rare occasion it also produces a bad answer and when you ask it will alternate betetwo bad answers, but most of the time it does pretty well. Here I tried asking it for a recipe for chocolate brownies. I also tried to find out if it gave correct answers. Based on my past use I'd say it's pretty likely the recipe is OK. What do you think? https://chat.openai.com/share/2b36e42c-b6f2-42f3-b67e-f4a5a7aa8b8a https://chat.openai.com/share/2b36e42c-b6f2-42f3-b67e-f4a5a7...
- Too 3y agoA programming snippet and a baking recipe are completely different. The first can be statically linted, fact checked against the docs, code reviewed and tested in a few seconds. Mainly it saves you the typing and finding the right name in the docs. The later: a trip to the store, an hour of rising, then 1h in the oven, then all your guests arrive, not worth the gamble.
- lesostep 3y agoI found original "The Joy of Cooking" brownie recipe https://www.cbsnews.com/news/brownie-recipes-from-joy-of-cooking/ https://www.cbsnews.com/news/brownie-recipes-from-joy-of-coo... and compared it with the recipe Chat-GPT gave you. Now, I'm not a great baker, so I'll relay heavily on "The Joy of Cooking" explanations. chat-GPT: 2. Mix melted butter and sugar. Add eggs, one at a time you can see how its a bad idea. Butter melting point is somewhere in 82.4–96.8 °F (28–36 °C). Eggs could be cooked at that temperature. Some of the proteins would harden prematurely. Also, who uses "1 cup of butter" as a measure. Original recipe calls for 1 stick of butter, but google says that 1 cup is 2 sticks. Maybe chat-GPT switched chocolate from original recipe with 1 stick of butter and cocoa powder? Chat-GPT doesn't mention beating mixture up either. Introducing more air to the mixture would make brownies better. Based on my experience, this chat-GPT recipe would make you some nice cookies, but not brownies.
- nullc 3y agoThe internet is full of false recipes that can't possibly work, probably increasingly LLM generated themselves. Consider: Persimmon jam. Not possible short of some secret chemical process-- heating reverses the chemical reaction that eliminates the astringency of the unripe persimmon. Yet there are now countless recipes online that will lead you to inedible muck, all effectively bogus hallucinations. Not necessarily machine ones, but also just content farms manually making every conceivable permutation of every existing recipe, the author of those persimmon jam recipes has probably never seen the fruit first hand. At least if you LLM them on your own you know to be dubious and also you're more likely to use a 70B model (or GPT4) which is a little more likely to get them right, vs whatever garbage the spammer used. It would probably be pretty interesting for someone to make a concerned effort to digitize every recipe in print prior to 2005 or so-- the relevant threshold time isn't when LLMs came on the scene but when google made low effort content farming so profitable. It wasn't guaranteed that every recipe printed pre-internet had actually been cooked by anyone, but it was a lot more likely.