Soon there won't be any blogs anymore written by humans. Perplexity, Kagi, Google, ChatGPT will kill ads revenue for them. There will be only synthetic generated content. I have generated some answers with hallucinated AI, and then Kagi scanned these pages and returned them as the answer for another generated page, which will be scanned again and again. Never ending loop with synthetic and fake content. Same I can do with Recipes. This is the end of the Internet, goodbye.
Some of the best blogs I've ever read were not written for the sake of "revenue". Ad revenue on blogs has basically been in the toilet for a decade anyway; there used to be a time when you could publish a niche blog and make good money from unobtrusive AdSense ads but those days are long gone.
> This is the end of the Internet, goodbye.
Beg to differ.
We have to consider the fact that human prompting + selection of outputs is essentially RLHF, so the models can and will continue to get better over time.
It's not the end of the Internet, it's the beginning of a new era.
The same SEO industry that crapped up recipes with a 300 word irrelevant personal story is now upset that we don't want to read their blogspam to see the recipe.
The same search engine that also runs a display ads network that incentivizes low quality SEO blogspam has so throughly killed the golden goose they have one more year left of summarizing human content before they fully enter the oroborous end state that is summaries of summaries of pure LLM Markov vomit.
Unfortunately it's not just Google's display ad business that incentivized blogspam, but pretty much every display ad business I've ever seen. So I'm not sure we can attribute the decline in search quality directly to this conflict of interest. It feels more to me like Google is just losing its mojo and running out of ideas, while the arms race of "SEO" continues.
Useful. I've been throwing recipes in chatgpt to spit out recipe ingredients in grams scaled to whatever X grams of protein I bought bulk for meal prep. It's saving me a lot of time.
Alternatively, a calculator? And with a calculator you don't have to worry about ChatGPT getting it completely wrong.
Copy paste + convert to grams for 2.74kg of pork shoulder takes 2 seconds instead of crunching numbers for every ingredient. It rarely gets it completely wrong. If the old way was more convenient I wouldn't be using new method.
Is chatgpt performing more reliability on unit conversion and math related questions?
I recall this being very hit or miss even a few months ago.
It is still very hit or miss for me. It often says nonsensical things (both ChatGPT4 and -4o) like “Yes, because $SMALL is larger than $LARGE…” when trying to convert units.
I haven't noticed anything obviously/frequently amiss dealing with recipes. I've seen it mess up when doing shoddy math with unit conversion i.e. barrels of oil to litres -> convert to energy units and prices per unit of energy in different currencies over various time period, calculations off by magnitudes enough that I have to ask it to show math step by step. I do remember chatgpt convertion fail telling me to add 3 litres of soy sauce intead of 30ml / 2 tablespoon, but with cooking it's pretty obvious when something is off.
You can always ask it to write and run code to validate the conversions in the same prompt. I believe Gemini is already doing this behind the scenes.
There might be more to this than Google bypassing the site. Quote from a rep was added to the article:
"We’re always experimenting with different ways to connect our users with high-quality and helpful information. We have partnered with a limited number of creators to begin to explore new recipe experiences on Search that are both helpful for users and drive value to the web ecosystem. We don’t have anything to announce right now.
Last week I made a chicken tortilla soup from a recipe on a random site. The directions were to cook the chicken for a suspiciously short time and sure enough, following the instructions verbatim the chicken was still mostly raw. Fortunately it was obvious enough that no one would have eaten it without cooking it longer. But it was probably my first encounter with an AI recipe. It doesn't seem like a mistake a person would make.
I made the mistake of forgetting to stick to sites with some level of reputation. Taking information directly from the search engine robs people of important context. Even if the site the recipe is taken from is cited, without going to the site you won't be able to get that sense of whether the site's look and feel is spammy or AI slop. And of course that applies to any other type of information too.
Google disgusts me more and more every day.