Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s • Hacker News Board

Re: Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s

#201

Post by whimsicalism » Sat, Dec 28, 2024, 7:07 PM UTC

lubujackson wrote: post #66

toddmorey wrote: post #31

Earlier quoted context omitted.

For quick simple steps like how to get a Bluetooth keyboard into pairing mode, it seems to work really well. I hated the prior world where everyone attempted to hide the real answer 3/4ths of the way through a useless blog post or YouTube video.

Which, we should note, didn't happen 10 years ago before the accountants took over search at Google. Those good, lean, helpful pages still exist. Google incentives websites to have pages of slop on everything now because they track how long you spend on a site as a "metric of a good match". Forrest for the trees...

i think if anything by google actively penalizes long slip articles with lots of affiliate links

Re: Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s

#202

Post by whimsicalism » Sat, Dec 28, 2024, 7:08 PM UTC

Schnitz wrote: post #27

binkHN wrote: post #7

> Enter 2024 with AI. The top 20% of search results are a wall of text from AI... I'll be the contrarian here and say I actually like Google's AI Overview? For the first time in a long time, I can search for an answer to a question and, instead of getting annoying ads and SEO-optimized uselessness, I actually get an answer. Google is finally useful again. That said, once Google screws with this and starts making sear…

The AI answers are nowhere near good enough to always be at the top, without any clear indication that they are just a rough guess. Especially for critical things like visa requirements or medical information. When you search Google for these sort of things, you want the link to the authoritative source, not a best guess. It’s very different for queries like say “movies like blade runner”.

agreed and frankly I am a big fan of LLMs in general… it just doesn’t seem like the one behind google search is all that smart

Re: Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s

#203

Post by xbmcuser » Sat, Dec 28, 2024, 7:08 PM UTC

I don't think it will be as easy to supplant google search as it was in the past with Yahoo and Alta Vista. In the past internet search was just starting we were all used to trying different search engine and the results were different enough with all the different ways of indexing and searching that moving from 1 to another made a difference. Most of the people using the internet were a bit of pioneers navigating through something brand new and unique.

Today majority of the internet users ie 10-32 years old or 60%+ of the internet users grew up with using Google and how to get best answers our of it. Chat gpt might bring in some churn but as long google is close enough it won't get replaced easily.

Re: Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s

#204

Post by whimsicalism » Sat, Dec 28, 2024, 7:09 PM UTC

asdff wrote: post #28

binkHN wrote: post #7

> Enter 2024 with AI. The top 20% of search results are a wall of text from AI... I'll be the contrarian here and say I actually like Google's AI Overview? For the first time in a long time, I can search for an answer to a question and, instead of getting annoying ads and SEO-optimized uselessness, I actually get an answer. Google is finally useful again. That said, once Google screws with this and starts making sear…

Were you not googling before? They had a bullet point summary that was actually more accurate because it scraped direct quotes from the website. Now I am getting wrong info from the ai summary. Its a huge step back from just what was there previously but its sold as some advancement.

…i’m no fan of the google AI feature but it is way more accurate than the scraped bullet point predecessor which would often scrape things while missing something key like a “here is the opposite of what we are talking about:” in the webpage

Re: Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s

#205

Post by griomnib » Sat, Dec 28, 2024, 7:10 PM UTC

beejiu wrote: post #117

echelon wrote: post #70

Earlier quoted context omitted.

I get very wrong and dangerous answers from AI frequently. I just searched "what's the ld50 of caffeine" and it says: > 367.7 mg/kg bw This is the ld50 of rats from this paper: https://pubmed.ncbi.nlm.nih.gov/27461039/ This is higher than the ld50 estimated for humans: https://en.wikipedia.org/wiki/Caffeinism > The LD50 of caffeine in humans is dependent on individual sensitivity, but is estimated to be 150–200 milli…

Perhaps a more common question: "How many calories do men need to lose weight?" Google AI responded: "To lose weight, men typically need to reduce their daily calorie intake by 1,500 - 1,800 calories" Which is obviously dangerous advice. IMO Google AI overviews should not show up for anything (a) medical or (b) numerical. LLMs just aren't safe enough yet.

Funny thing is you can train a small BERT model to detect queries that are in categories that aren’t ready for AI “answers” with like .00000001% of the energy of an LLM.

Re: Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s

#206

Post by HarHarVeryFunny » Sat, Dec 28, 2024, 7:10 PM UTC

binkHN wrote: post #7

> Enter 2024 with AI. The top 20% of search results are a wall of text from AI... I'll be the contrarian here and say I actually like Google's AI Overview? For the first time in a long time, I can search for an answer to a question and, instead of getting annoying ads and SEO-optimized uselessness, I actually get an answer. Google is finally useful again. That said, once Google screws with this and starts making sear…

An answer is only as good as the expertise behind it. When searching I always pay attention to the source, and will skip ones that look less trustworthy.

One major advantage of Google's original pagerank was that originally it worked well and number of links to a page was a good proxy for trustworthiness and authority on a subject. It used to be that you'd find what you were looking for in the top few Google search results, which was a massive improvement to Alta Vista which was the existing competition where you'd have to wade though pages of keyword match sites listed in no particular order.

Anyway, source is critically important, and if I'm looking to find something authoritative then the output of an LLM, even if RAG based, is not what I'm looking for! Increasingly people may be looking to search to verify stuff suggested by an LLM, which makes a search engine that puts LLM output as it's top result rather unhelpful!

It doesn't help that with Google in particular their AI output is all heavily DEI biased, and who knows what else ... I just don't trust it as objective.

Re: Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s

#207

Post by griomnib » Sat, Dec 28, 2024, 7:11 PM UTC

chimpanzee wrote: post #85

echelon wrote: post #70

Earlier quoted context omitted.

I get very wrong and dangerous answers from AI frequently. I just searched "what's the ld50 of caffeine" and it says: > 367.7 mg/kg bw This is the ld50 of rats from this paper: https://pubmed.ncbi.nlm.nih.gov/27461039/ This is higher than the ld50 estimated for humans: https://en.wikipedia.org/wiki/Caffeinism > The LD50 of caffeine in humans is dependent on individual sensitivity, but is estimated to be 150–200 milli…

For whatever it’s worth, in response to the same question posed by me (“what is the ld50 of caffeine”), Google’s AI properly reported it as 150-200 mg/kg. I asked this about 1 minute after you posted your comment. Perhaps it learned of and corrected its mistake in that short span of time, perhaps it reports differently on every occasion, or perhaps it thought you were a rat :)

LLM are non deterministic by nature.

Re: Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s

#208

Post by imperialdrive » Sat, Dec 28, 2024, 7:11 PM UTC

potsandpans wrote: post #42

Just sharing an anecdote: I was searching for a quote that I'd heard in an audiobook the other day. I just had the general paraphrase, and didn't feel like scanning through the chapters to go find it. This was a somewhat obscure source. Google had just straight garbage for me. The quote was political in nature, and I felt like the results were fighting general tone-policing filters and were tuned for recent events. o…

I had a similar experience last week. Searched and searched for a blurb from Middlemarch, without luck. o1 returned exactly what I was hoping for in one fell swoop.

Re: Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s

#209

Post by nixosbestos » Sat, Dec 28, 2024, 7:11 PM UTC

georgeplusplus wrote: post #139

I think I’ve had such a bad experience with openAIs charGPT answering a question falsely and sounding very convincing that I skip the AI overview or double check to confirm it’s correct. Am I mad?

I already regularly am helping someone with something, they'll Google it at the same time as me and they'll authoritatively regurgitate the AI summary. More often than not, it's been misleading, or a partial answer to the point of being detrimental.

I get it though. I sort of am the news guy of my family and they constantly ask me where I go for news. It's an impossible answer. I grew up with the internet in it's early days, using it for policy debate research. It's a non-precise method of consuming various sources, judging their answers on plausibility and commonalities, and a healthy dose of constant skepticism.

But people don't want that. They want the answer spoodfed to them.

Re: Google's Results Are Infested, Open AI Is Using Their Playbook from the 2000s

#210

Post by Beijinger » Sat, Dec 28, 2024, 7:11 PM UTC

jayd16 wrote: post #190

Beijinger wrote: post #183

This! "I don't want to watch a 10-minute video for a quick answer." And this: "OpenAI's search is becoming Google in the 2000s, if it can remain trustworthy." The problem I see: People use OpenAI/Perplexity for knowledge. Not to seek website. I think sooner or later, most website will block AI crawlers. What does a website gets out of it?

I'm still waiting for the AI service that turns those 10 minute videos into a text tutorial with photos. I think it's pretty damning that it's not a built in YouTube feature by now.

https://chromewebstore.google.com/detail/youtubedigest-summa...