Sure but the problem with the rocket scientists is they think about how good things can be, not how evil they can be. Sure, a chat bot could be made to give very valuable-to the end user results but as it will free like Google, results yielding-profits-to-company will be given, in the fashion you describe Google doing.
For example, I just got curious and asked Google how many oranges in a cup of juice. The main result is a conversion of a cup to cubic inches, for no reason. Scrolling down a bit, the first usable answer was a quora blurb.
ChatGPT told me it depends on the size and type of orange, but generally 2 to 4.
Guess which one I'll use next time?
I'm not sure why you didn't get this, but it seems better than ChatGPT.
Interestingly, if I add 'are' as in 'how many oranges are in a cup of juice', it doesn't do that weird cu in conversion and gives a reasonable answer.
That seems like an odd parsing choice for lack of a, in this case, fluff word, as the former is how I would ask the question... though maybe I'm the odd one.
And there used to be sites that answered the question on the search page. Google pushed them to the bottom.
Every single hour spent putting work on a website that doesn't go towards any of its users is better not happening. If AI is what it takes so a website doesn't have to "feature" whatever amount of useless stupid paragraphs riddled with keywords to tell Google what the site is about then so be it.
1st result: https://www.apple.com/watch/
Google worked perfectly for me when I wanted to buy a product.
Case in point. The fact that this works really well for this type of content means that Google has failed. Now whether they failed deliberately or not is a different question.
In either case, search needs to be demonopolized. The truth is we don’t even know what search would be like without the Google hegemony. Even if all language models do is force Google to innovate and improve it’s worth every penny.
If you simply search "best frypan", you'll get lists of affiliate links to overpriced Chinese crap, written by someone who has never even looked at any of the products in person and frankly doesn't even care if they're any good.
Google shouldn't be expected to know everything, but it should know that someone searching for "best frypan" is looking for real recommendations and not literal spam pumped out by a content mill.
Searching for services like appliance repair or removalists are even worse, to the point where I'll now just ask on private Facebook groups for recommendations.
Pages and pages of SEO spam, generally ranked by how big & established the site pushing said spam is, rather than any unique insights contained within.
I guess the market responded with brand loyalty. If you have an iPhone, you just buy the Apple Watch, you don’t look further. If you have a pixel phone, you just buy whatever google sells.
You can't really tell what the user wants.
Followup question - is LLM tech more or less likely to end up replacing search when those same incentives really saturate the LLM product itself.
One big difference seems to be attribution. LLMs don’t tell you where their info comes from. They just say what is (but can be asked to cite works - with mixed results).
Will LLMs get good at citing sources and if not will people care or will they give in to LLMs as a source of information that’s “good enough because it mostly works”
From a product placement / ads perspective, being able to persuade people to fully accept everything an ad-infused LLM says because it’s good enough seems like an incredibly lucrative product bundling opportunity if they can get it right. Esp. If they can use that to convince regulators they can’t annotate the difference between ads and non ads.
Seems pretty dystopian from a disinformation perspective though.