Claude can now search the web

Post by importantstuff » Fri, Mar 21, 2025, 7:51 AM UTC

shortrounddev2 wrote: post #416

ggm wrote: post #407

So in many respects, search the place it used to construct the model? Isn't that functionally bias-reinforcing? "Look what I synthesise is correct and true because when I use the same top 10 priming responses which informed my decision I find these INDEPENDENT RESULTS which confirm what I modelled" type reasoning. None of us have a problem with an LLM which returns 2+2 = 4 and shows you 10 sites which confirm. What w…

We will very quickly enter a Kepler effect of information on the internet. All text on the internet will become AI slop being parsed by AI. Real information and human beings will be drowned out by the garbage. The internet will cease to be useful and we will retreat to corners of the web or to walled gardens. I'm seeing more and more online communities these days enforce invite only because there's just too much AI s…

Post by hansmayer » Fri, Mar 21, 2025, 8:02 AM UTC

simonw wrote: post #167

hansmayer wrote: post #157

So, referring specifically to the example they show on the front-page, what value does this bring actually? The best example they could come up with is Typescript migration ? Really? Weren't the LLMs supposed to be a superior alternative to searching the web? Why do we need to produce more CO2 to do the same we could have done at the fraction of the cost, of course at the time when the google search was still working…

The CO2 concerns of using LLMs are massively overblown these days (with the exception of o1-pro and GPT-4.5 at least). The energy efficiency of most models has improved by an order of magnitude since the most widely cited CO2 usage papers were published. (It remains frustratingly difficult to get accurate numbers though: at this point I think more transparency would help rather than hurt the big AI labs)

Post by theshackleford » Fri, Mar 21, 2025, 8:07 AM UTC

mvdtnz wrote: post #68

_2d30 wrote: post #39

Earlier quoted context omitted.

I don't think it should. If a user asks the AI to read the web for them, it should read the web for them. This isn't a vacuum charged with crawling the web, it's an adhoc GET request.

No thank you, when I define a robots.txt file I expect all automated systems to respect it.

Post by input_sh » Fri, Mar 21, 2025, 8:10 AM UTC

UnreachableCode wrote: post #510

tcdent wrote: post #102

Searching the web is a great feature in theory, but every implementation I've used so far looks at the top X hits and then interprets it to be the correct answer. When you're talking to an LLM about popular topics or common errors, the top results are often just blogspam or unresolved forum posts, so the you never get an answer to your problem. More of an indicator that web search is more unusable than ever, but inte…

> web search is more unusable than ever I’m curious why I’m seeing a lot of people thinking this lately. Google definitely made the algorithm worse for customers and better for ads, but I’m almost always able to find what I’m looking for in the working day still. What are other people’s experiences?

Post by collyw » Fri, Mar 21, 2025, 8:12 AM UTC

tcdent wrote: post #102

Searching the web is a great feature in theory, but every implementation I've used so far looks at the top X hits and then interprets it to be the correct answer. When you're talking to an LLM about popular topics or common errors, the top results are often just blogspam or unresolved forum posts, so the you never get an answer to your problem. More of an indicator that web search is more unusable than ever, but inte…