The real reason is because recipes consisting of just the list of ingredients can't be copyrighted, but the articles and all the other stuff are copyrightable.
Ingredients lists and basic instructions ("place potatoes in boiling water") are not copyrightable. Only the additional text is copyrightable. This has long been a 'problem' for cookbook writers and chefs. Unlike cookbooks, blogs have no particular constraints on how much text they can add, and as pointed out here, perverse incentives to add lots of extraneous material.
I don’t see this. If somebody’s scraped just the recipes from all these long winded sites, I’d be thrilled! Adding the extra nonsense and narrative doesn’t prevent that.
I also wonder how much of it is a perception among the blog authors that they have to make their recipes long winded for SEO reasons. They see everyone else doing it, there's an information cascade, and they believe (perhaps incorrectly) that they also must do it.
I suppose you could have a "conspiracy theory" level argument that Google publishes what it claims it will do only to turn around and do the opposite, but that seems unlikely.
Unfortunately if you publish guidelines that you'll promote pages that are not auto-generated DB dumps, and prefer pages that promote the idea the author put in a lot of effort and is highly authoritative and E-E-A-T and all that you can read in the guidelines, you're going to get generic recipes wrapped in a VERY thick blanket of fake filler, good enough to fool the average reader (whom doesn't read very well) but ridiculously inadequate compared to where the pros and semi-pros operate.
Speaking as someone who likes to cook, and hates wading through pages of SEO bait to find the actual recipe, it seems like in principle it wouldn't be too hard to write a search engine specialized for recipes that heavily penalizes longwindedness.
Made the cod chowder last night. Came out great, but there's one thing I noticed: the recipe didn't specify how much water or salt to use. I eyeballed it and it came out fine, but that probably wouldn't work for something touchier like baked goods.
Sometimes they could still do with more detail, especially when using imprecise imperial measurements. The one that I find most irritating is using volume measurements for non-liquids, e.g. cup of flour (sifted or not?), or ratios, where it's not specified if it's a weight:weight or volume:volume ratio.
You can try plainoldrecipe.com[0] for de-cluttering recipes. It doesn't work for a lot of the bloggy recipe sites, but when it does[1], it's great. And it works for all the popular sites like AllRecipes.
There's also the app CopyMeThat, which will capture recipes from darn near anything, let you tag and otherwise organize them, let you make a mealplan, and more. We have 3000+ recipes in it now (of which we've made probably a couple hundred lol).
I would love to see a crowdsourced cookbook with user submitted variations on recipes and vote ranking and comments. search by ingredients, dish, macros, calories. create meal plans, meal prep, shopping lists. then you can integrate with grocery store apis to pick up your shopping list or have it delivered with Instacart.
there are lots of avenues for monetization but it's a crowded space and hard to build all these features in and gain momentum.
My little dream project is to build a yelp for recipes. Books, internet or custom from users. I don't like Google results and also don't think search engines should be the place to find a good recipe or judge quality or authencity etc..
Searching for recipes turns out to be a great use for ChatGPT. You can give it pretty fuzzy search parameters and have conversations about adjusting the recipes and it will just give you a simple list of ingredients and instructions - no BS.
Conversely, if you do want to use regular searches and recipes from internet sites, the Paprika app [0] is amazing. It will download and parse the recipes and get rid of all the BS. You can then sync recipes between devices and people as well as manage grocery lists and meal plans based on the recipes.
I would honestly guess that the blog text is point of the process and not the recipe for many of the writers. Maybe the recipe is just there to justify the process of writing the blog post. Or they would eventually be asked for the recipe, so get away having to repeatedly answer to same question.
ChatGPT also doesn't really understand the recipes it's giving you, so I don't know if I would trust it to make something tasty: https://www.youtube.com/watch?v=yT8KoWpqUgg
ChatGPT is the last thing I would use for recipe help (0). Instead I'd go with a reliable source like SeriousEats, NYTimes, or as you suggested, Cooking for Engineers.
An alternative opinion with a logical underpinning:
1) Chili is a dish most closely associated with the southwestern United States.
2) Among the most important cultivars of that region, dating back thousands of years, are: peppers, maize, squash, beans. This combination provides a complete mix of nutrients for a healthy diet.
3) Therefore, adding all those ingredients to chili in some form or other is entirely justifiable, as well as being historically accurate.
Can you help me corroborate your theory of historically accurate chili? I can not find any document supporting your "thousands of years" claim.
It not hard to believe that native people dumped some of the earth's best vegetables into a pot and let them steep, but, I can not find _any_ support for your claim.
I genuinely hope you enjoy your historically-accurate, bean-laden chili, but now I am suspicious; I know that LLMs can not really enjoy the richness in any chili.
This is the problem in a nutshell, do you define a "good recipe" as one that is historically accurate, that tastes good, or that follows various belief systems like not culturally appropriating or only culturally appropriating when you do a good job of it?
Meanwhile the ad sales platform is going to define a "good recipe" as one that results in high click thru on ads they have contracts for.
And add a side dish of the people doing the rating of how good the page is, specifically have zero skill in the field and are going to have intense normie middle of the road bias because on average they don't know anything.
If you magically crowdsourced the averaged opinion of average folks about the General Theory of Relativity in 1900, would that result be useful and have beaten Einstein before he wrote it in 1915? The AI people think so; especially the ones who's paychecks depend on it; personally I have my doubts.
No one in this comment thread made any claim of any recipe or particular chili recipe being "good".
Your summary is confusing to me, because whatever the "normie middle of the road" idea we are supposedly now discussing does not have anything to do with what I said or had asked questions about.
Most Texas Red Chili's originate from a dish called Chile Colorado with deep roots in mexican cuisine. Chile Colorado is served with tortillas and frijoles (pinto beans), and usually some pickled veggies. At some point its not impossible to believe someone went "To hell with it" and just added all of the ingredients to the pot.. but you are correct, there's really no good clear history of how Chili with meat became Chili with beans.
Fun fact, chile colorado translated just means colored red chile.
the lore that know-nothings told themselves about seo. the think seo is about content, its about users. they screwed this pooch up over a decade ago and it noticeably changed the content of the internet for the worse, much like what people are now consciously thinking about chatgpt. its shocking that we finally have enough people to notice this but it wont be enough.
Google wills it so. Why? IDK, ask them. Probably yet another case of something net-harmful making them more money.
Luckily it's also one of the only cases where a microformat is actually nigh-universally used. Outside a few sites where the "fluff" isn't fluff and is actually good content itself (e.g. Serious Eats) I just throw the URL into Paprika 3 (no affiliation, and there are tons of other programs and probably some browser plugins to do the same thing) and let it strip it down to the ingredients and instructions by reading the microformat markup. Works flawlessly nearly every time.
[EDIT] Credit where it's due, I'm also pretty sure the microformat is only popular because of Google. I think they downrank sites in recipe searches if Google can't "see" the recipe, using the microformat. They may be both the cause of, and solution to, the problem.
Trying to mentally parse these long winded recipes = takes longer to read = "greater engagement". These crappy recipes are all about seo and add placement as you'll notice these recipes are loaded with Google ads. So it is a self fulfilling prophecy that the pages where users spend more time and see more ads are ranked highest.
I’ve completely given up on web recipes, and only use YouTube ones now. Most of the good recipes have a comment that has a condensed version of ingredients and instructions.
Yes. I think Google deserves a ton of credit for microformats, and you can just use a browser extension like Repibox that shows you only the recipe when you go to a recipe site.
I think the reason Google wanted microformats is for Google Assistant recipe answers.
^ Colors in that space only just recently started getting supported by displays/browsers. And while p3 colors work in css on Safari, I had to use a single-pixel background png to make it work in Chrome.
I tried that and it was super buggy... Thankfully I built the whole thing in SwiftUI so that I can port it to Mac fairly quickly. Moving as fast as I can!
But yes, you are right, as far as native apps go. FWIW, I just got an Android test phone so I can start on that version, should be ready in a few months.
Looks cool. As a cook in my fifth decade, may I humbly offer some suggestions? Im on Android and PC so perhaps not everything will apply, but here goes:
1. Import from other recipe apps.
Recipe apps and sites have a finite lifetime. Its great that you already support export, but import from others would be great. Pepperplate for example is sqlite on android. Other apps might be harder, eg the New York Times recipe box. But without import, its hard to commit to a new system.
2. Cooking mode
This is a mode that avoids that situation where you are following a recipe and need to manipulate your phone but you cant because you have hands covered in flour/chicken juice or whatever. Voice control would be awesome too: "scroll down", "set a timer for 20"
3. Notes
Some place where you can comment on the recipe, like "use peanut oil not canola"
Definitely agree. A couple days ago, someone asked for a Paprika recipe importer, which I've added here: https://www.umami.recipes/paprika-import
I'll have to take a look at Pepperplate and NYT recipe box (though, you can already import any NYT recipe link!).
> Voice control
Love this idea.
> 3. Notes
There is indeed a notes section! Though I'm guessing you mean inline comments, like google docs style?
On #1, there are so many potential import sources that you probably cant do them all solo...but attacking the top few and documenting what your import format is would be a win. In short, let motivated members of the community with an itch to scratch help. Good ones could be folded into the mainline if you agree with the submitter, maybe.
On 3, notes can either be for personal use (eg in Cozi's recipe handler) or a shared conversation in a recipe (like what the NYT does). I feel the former is more important, but your users might have different opinions.
I'm open to it! Right now it runs on firebase and a unary gRPC service (nodejs). Probably firebase could be swapped out for supabase, and gRPC is easy enough to host.
Finding a result on Google, clicking on it, then leaving and clicking something else on Google means you probably didn't find the thing you were looking for. Staying on one site for a while probably means you found what you were using for. Presumably Google uses this information to try to determine if you engaged with the site, and then it ranks sites that you stay on longer higher in search results.
This works great for things like StackOverflow posts where you can quickly judge if the answer is correct, and if it's not, you can quickly bounce. It's not great for recipes, which are almost always "correct" (in that they are a recipe of the type you're looking for), so spending longer probably just means you scrolled for longer.
How does it interplay with opening multiple search results in new tabs at once? It's been ages since I've personally visited search results sequentially - I open them in batches.
Not a problem, you're an outlier, the average user either doesn't use tabs or doesn't open multiple of them to let them load in the background because they haven't experienced the web on a modem and "I click, page loads in 1-5s, I read" is what they're used to.
Not sure experiencing the web on a modem is a real factor here. Anecdotally, at least, I've certainly experienced the web on a modem -- and living in a rural area, broadband wasn't available for me until maybe 2005-2006 (and even then, only 1.5Mbps until around 2014) -- but I open multiple tabs not to let them load in the background, but to avoid unnecessary context switches. I can load ten tabs async, never context-switch, and then process the tabs one by one. Half the context switches of opening a page, going back, etc.
I feel for your prolonged and unnecessary pain with rural broadband, a friend of mine lived just a few hundred meters too far from the great access and suffered for years until the cable company finally relented.
Regarding context-switching: that's even more of an outlier! The vast majority never worries about that, and from my experience even among knowledge workers, that's rare.
I've never seen a recipe that was "correct". I don't even know what that would mean. As far as i'm concerned, a recipe is a suggestion. If I don't know how to make a dish, I read several recipes, and then brew my own.
I don't think most online recipes are "correct", especially chili recipes that call for Heinz baked beanz. Wut?
You are clearly skilled at cooking. For people like myself, who are not so blessed, rigorously following recipes is the only realistic path to producing decent dishes.
So, for me, a recipe can indeed be correct or not. My definition of "correct" is narrow, though -- if it produces a dish I enjoy, it's correct. Otherwise, it's not.
Yeah I wish you could hard nope our of videos and never see them again. It should be possible without viewing the shit. I could ban all the conspiracy and fake shit
Google presents what its customers (people doing searches) want to see, so shouldn't we ask the people who click the results that make it to the top (whether they leave the page dissatisfied or not is a different story, right?) over and over?
What incentive does Google have for prioritizing succinct quality posts versus blogspam posts? What incentive does Google have for prioritizing blogspam posts over succinct quality posts? Why is it the way it is if we're talking about the unsponsored (not paid) organic results?
As a blogspam author, why do you want people on your page? To get more AdSense views/impressions? You typically aren't selling anything on a website that gives away free long-winded recipes (so conversion rate/cost per click/paid advertising isn't really a concern?)
> Google presents what its customers (people doing searches) want to see
Google presents what it thinks people want to see, which it infers based on some number it can collect, like time-on-page, or scroll depth. The argument is that its proxy metrics for "the user liked this" are ill-considered, especially in this particular context.
Even when Google acts like a rational actor, it is not an oracle. It rationally optimizes an observable metric, which may or may not correlate to its actual goal.
I have no direct knowledge, but Google is known to penalize duplicate pages. Without the lead in, one bread recipe is pretty much the same page as 1,000 others.
Nearly everything online today is long-winded. Hell, most media is long-winded. You can't just know what happened, where, and why; you've got to read a creative-writing thesis to "put you in the shoes" of the subject du jour. You can't simply watch a lecture; you need fancy graphics and fast cuts with memes woven in. Want a compelling story? You better watch all 6 seasons, including the remaining 4 without the original writers, and be prepared to be disappointed. How about learning to fix your car? In that case, be prepared for 3 minutes worth of info in 20 minutes of some schmo complaining about why Hyundais suck while he constantly jiggles the camera and gloats about much more clever he is than other mechanics.
News, blogs, recipes, shows, instructional videos... all pretty much blow chunks. Any time I simply want straight forward information it becomes a matter of frequent fast-forwarding and combining of info from disparate sources when it shouldn't be necessary. Get. To. The. Point.
It's easy to think all of the things you wanted to say were important, so you leave them in. But often they weren't so important to your eventual audience.
Honestly I think this is part of the reason why Twitter was so successful for so long. The rest of the internet has become so padded for length (even streaming TV content -- how many shows have you watched where you thought "this would be better as a movie"?) one platform that demands you focus on being short, punchy, and getting straight to the point would obviously ascend.
And instead people make tweetchains a dozen long, making Twitter the worst of all worlds. It got better when they raised the cap from 140, but it's still too low to saying anything of substance or nuance.
Yes, but tweetchains are notable in that they're exceptions to the normal flow of chatter on twitter. And a good tweetchain will have people retweeting-out the good parts.
Should every site be twitter? Hell no. But is it a good thing that one or two major social media sites demand brevity and distilling your thoughts? Yes.
A different thing, certainly. An interesting thing, probably. A good thing? Just because something is different and interesting doesn't mean it's good. Twitter exists as a site on the internet, which is probably the most charitable thing that deserves to be said about it.
Honestly this is why TikTok ended up being so successful. Short videos, straight to the point, with the Reddit vibe of "this is from an actual human". I routinely try out new recipes I've seen on TikTok (not the trendy influencer recipes that just look good though, actual recipes from people) and am never disappointed. It takes like 30 seconds to get the gist of the recipe, and the details are in description or comments without any fluff.
Google measures the amount of time you spend on a webpage before you go back to the search page (if you dont open in a new tab) and it’s used as an important ranking signal. That’s pretty much it.
All the long winded recipe bullshit just keeps you on the page longer before you realize the recipe is crap.
The problem will not be that the value of content will drop to zero, but that the value of asking someone else to ask ChatGPT for me and then wrap it in ads for me to view is very low compared to the inevitable result of I'll just ask ChatGPT or similar directly.
Its the LMFGTFY or whatever the acronym for "let me google that for you" brought to ChatGPT topics.
If, for example, as already seems to be the case, the news becomes nothing more than someone else generated algorithmic fluff mixed with censored propaganda that somehow ends up with lower trust metrics than actual email spam, why do I need someone to make that for me, just ask for my own copy. The future will not be CNN trying to lie to you and sell you stuff, it'll be the AI directly lying to you and trying to sell you stuff, CNN will no longer be a provider of value. Or MSNBC or Fox or BBC or your choice of propaganda outlet.
The other side problem is if you want culinary excellence today you'll eventually find guidance to buy "modernist cuisine at home" or similar technical cookbooks or maybe some old Alton Brown TV show memes, but search engines are required by corporate to have total randos provide opinions about page quality and needs met that are going to be uneducated and thus generic/bland and are designed to select inoffensive bland middle of the road content, not excellent recipes. What appeals to children more, an animated skinner box / loot box paradise designed to attract their attention and lure them in, or a calculus textbook? If corporate goal is to maximize the former you'll get more of the former, but what if the latter is the only way to learn calculus? Well, you aint going to learn calculus, thats for sure.
How AI / ChatGPT will avoid bland normie ignorance is an excellent question. How to provide bland middle of the road answers to bland middle of the road questions is essentially a solved problem; how to cook the best hamburger possible cannot be handled by that subculture where groupthink conforming to average blandness is the primary criteria enforced.
I'm reminded of people eating Tide pods. Who makes 'better' memes? Social media or ChatGPT? Is something like this situation proof that Dead Internet Theory is real and social media is mostly bots?
279 comments
[ 5.9 ms ] story [ 254 ms ] threadThe reason is SEO; this is Google’s fault.
(I am not affiliated with it)
https://support.google.com/websearch/answer/9281931?hl=en
I suppose you could have a "conspiracy theory" level argument that Google publishes what it claims it will do only to turn around and do the opposite, but that seems unlikely.
Unfortunately if you publish guidelines that you'll promote pages that are not auto-generated DB dumps, and prefer pages that promote the idea the author put in a lot of effort and is highly authoritative and E-E-A-T and all that you can read in the guidelines, you're going to get generic recipes wrapped in a VERY thick blanket of fake filler, good enough to fool the average reader (whom doesn't read very well) but ridiculously inadequate compared to where the pros and semi-pros operate.
IANAL
[0] https://plainoldrecipe.com/
[1] https://plainoldrecipe.com/recipe?url=https%3A%2F%2Fwww.spen...
https://based.cooking/
2. Keywords and content for SEO
3. Some bloggers plan on compiling their posts into a cookbook
there are lots of avenues for monetization but it's a crowded space and hard to build all these features in and gain momentum.
Conversely, if you do want to use regular searches and recipes from internet sites, the Paprika app [0] is amazing. It will download and parse the recipes and get rid of all the BS. You can then sync recipes between devices and people as well as manage grocery lists and meal plans based on the recipes.
[0] https://www.paprikaapp.com/
The main difference being that those anime articles seem entirely generated from templates.
Also I have to mention this http://www.cookingforengineers.com/
(0) - https://i.redd.it/8xvhkhxc495a1.png
1) Chili is a dish most closely associated with the southwestern United States.
2) Among the most important cultivars of that region, dating back thousands of years, are: peppers, maize, squash, beans. This combination provides a complete mix of nutrients for a healthy diet.
3) Therefore, adding all those ingredients to chili in some form or other is entirely justifiable, as well as being historically accurate.
Meat, on the other hand, is entirely optional.
It not hard to believe that native people dumped some of the earth's best vegetables into a pot and let them steep, but, I can not find _any_ support for your claim.
I genuinely hope you enjoy your historically-accurate, bean-laden chili, but now I am suspicious; I know that LLMs can not really enjoy the richness in any chili.
Meanwhile the ad sales platform is going to define a "good recipe" as one that results in high click thru on ads they have contracts for.
And add a side dish of the people doing the rating of how good the page is, specifically have zero skill in the field and are going to have intense normie middle of the road bias because on average they don't know anything.
If you magically crowdsourced the averaged opinion of average folks about the General Theory of Relativity in 1900, would that result be useful and have beaten Einstein before he wrote it in 1915? The AI people think so; especially the ones who's paychecks depend on it; personally I have my doubts.
No one in this comment thread made any claim of any recipe or particular chili recipe being "good".
Your summary is confusing to me, because whatever the "normie middle of the road" idea we are supposedly now discussing does not have anything to do with what I said or had asked questions about.
Fun fact, chile colorado translated just means colored red chile.
Edit: I fixed some typos.
Luckily it's also one of the only cases where a microformat is actually nigh-universally used. Outside a few sites where the "fluff" isn't fluff and is actually good content itself (e.g. Serious Eats) I just throw the URL into Paprika 3 (no affiliation, and there are tons of other programs and probably some browser plugins to do the same thing) and let it strip it down to the ingredients and instructions by reading the microformat markup. Works flawlessly nearly every time.
[EDIT] Credit where it's due, I'm also pretty sure the microformat is only popular because of Google. I think they downrank sites in recipe searches if Google can't "see" the recipe, using the microformat. They may be both the cause of, and solution to, the problem.
Like how select cuts of beef slow roasted with veggies and served on a bed of glazed hand cut potatoes.
Is more appealing to a potential cook than cut beef in a slow cooker with some wedges.
I think the reason Google wanted microformats is for Google Assistant recipe answers.
Umami has Chrome/Firefox extensions that you can use while on desktop, which will sync your recipes to the app. Disclaimer: I built it :)
^ Colors in that space only just recently started getting supported by displays/browsers. And while p3 colors work in css on Safari, I had to use a single-pixel background png to make it work in Chrome.
But yes, you are right, as far as native apps go. FWIW, I just got an Android test phone so I can start on that version, should be ready in a few months.
1. Import from other recipe apps.
Recipe apps and sites have a finite lifetime. Its great that you already support export, but import from others would be great. Pepperplate for example is sqlite on android. Other apps might be harder, eg the New York Times recipe box. But without import, its hard to commit to a new system.
2. Cooking mode
This is a mode that avoids that situation where you are following a recipe and need to manipulate your phone but you cant because you have hands covered in flour/chicken juice or whatever. Voice control would be awesome too: "scroll down", "set a timer for 20"
3. Notes Some place where you can comment on the recipe, like "use peanut oil not canola"
Thank you
> 1. Import from other recipe apps.
Definitely agree. A couple days ago, someone asked for a Paprika recipe importer, which I've added here: https://www.umami.recipes/paprika-import I'll have to take a look at Pepperplate and NYT recipe box (though, you can already import any NYT recipe link!).
> Voice control
Love this idea.
> 3. Notes
There is indeed a notes section! Though I'm guessing you mean inline comments, like google docs style?
On 3, notes can either be for personal use (eg in Cozi's recipe handler) or a shared conversation in a recipe (like what the NYT does). I feel the former is more important, but your users might have different opinions.
This works great for things like StackOverflow posts where you can quickly judge if the answer is correct, and if it's not, you can quickly bounce. It's not great for recipes, which are almost always "correct" (in that they are a recipe of the type you're looking for), so spending longer probably just means you scrolled for longer.
Regarding context-switching: that's even more of an outlier! The vast majority never worries about that, and from my experience even among knowledge workers, that's rare.
I don't think most online recipes are "correct", especially chili recipes that call for Heinz baked beanz. Wut?
So, for me, a recipe can indeed be correct or not. My definition of "correct" is narrow, though -- if it produces a dish I enjoy, it's correct. Otherwise, it's not.
longer times on a page means longer enagement == more ads revenue.
is it possible that google is incentivizing longer engagements via adwords by using search priority that ranks by average time spent on page?
http://microformats.org/wiki/h-recipe
But it looks like there might be one or two others active in the wild.
Google presents what its customers (people doing searches) want to see, so shouldn't we ask the people who click the results that make it to the top (whether they leave the page dissatisfied or not is a different story, right?) over and over?
What incentive does Google have for prioritizing succinct quality posts versus blogspam posts? What incentive does Google have for prioritizing blogspam posts over succinct quality posts? Why is it the way it is if we're talking about the unsponsored (not paid) organic results?
As a blogspam author, why do you want people on your page? To get more AdSense views/impressions? You typically aren't selling anything on a website that gives away free long-winded recipes (so conversion rate/cost per click/paid advertising isn't really a concern?)
Google presents what it thinks people want to see, which it infers based on some number it can collect, like time-on-page, or scroll depth. The argument is that its proxy metrics for "the user liked this" are ill-considered, especially in this particular context.
Even when Google acts like a rational actor, it is not an oracle. It rationally optimizes an observable metric, which may or may not correlate to its actual goal.
News, blogs, recipes, shows, instructional videos... all pretty much blow chunks. Any time I simply want straight forward information it becomes a matter of frequent fast-forwarding and combining of info from disparate sources when it shouldn't be necessary. Get. To. The. Point.
Should every site be twitter? Hell no. But is it a good thing that one or two major social media sites demand brevity and distilling your thoughts? Yes.
A different thing, certainly. An interesting thing, probably. A good thing? Just because something is different and interesting doesn't mean it's good. Twitter exists as a site on the internet, which is probably the most charitable thing that deserves to be said about it.
All the long winded recipe bullshit just keeps you on the page longer before you realize the recipe is crap.
Th primary reason being generating long form has become so cheap with AI that it has essentially rendered it worthless.
Its the LMFGTFY or whatever the acronym for "let me google that for you" brought to ChatGPT topics.
If, for example, as already seems to be the case, the news becomes nothing more than someone else generated algorithmic fluff mixed with censored propaganda that somehow ends up with lower trust metrics than actual email spam, why do I need someone to make that for me, just ask for my own copy. The future will not be CNN trying to lie to you and sell you stuff, it'll be the AI directly lying to you and trying to sell you stuff, CNN will no longer be a provider of value. Or MSNBC or Fox or BBC or your choice of propaganda outlet.
The other side problem is if you want culinary excellence today you'll eventually find guidance to buy "modernist cuisine at home" or similar technical cookbooks or maybe some old Alton Brown TV show memes, but search engines are required by corporate to have total randos provide opinions about page quality and needs met that are going to be uneducated and thus generic/bland and are designed to select inoffensive bland middle of the road content, not excellent recipes. What appeals to children more, an animated skinner box / loot box paradise designed to attract their attention and lure them in, or a calculus textbook? If corporate goal is to maximize the former you'll get more of the former, but what if the latter is the only way to learn calculus? Well, you aint going to learn calculus, thats for sure.
How AI / ChatGPT will avoid bland normie ignorance is an excellent question. How to provide bland middle of the road answers to bland middle of the road questions is essentially a solved problem; how to cook the best hamburger possible cannot be handled by that subculture where groupthink conforming to average blandness is the primary criteria enforced.
https://chrome.google.com/webstore/detail/recipe-filter/ahlc...
More real estate for you to scroll through for display ads?
I agree, though. It's why I try going to websites run by people who make their money elsewhere. Like Food Network.