We need a way to collapse comments on HN
OMG!
Claude can now search the web
741–750 of 758 posts
Re: Claude can now search the web
#742Earlier quoted context omitted.
Yeah sure, some people do this. But average humans understand the limit of their knowledge. LLMs cannot do that. You can find the right person for a space where this knowledge of limitations is necessary. Can't find an LLM which does that
I will grant you that there are at least some of us capable of this, where you’ll find no LLM capable. > average humans understand the limit of their knowledge. We’ll have to agree to disagree here. I’d call it a minority, not the average. Which is why we live in a world where huge numbers of people think they know significantly more than they do and why you will find them arguing that they know more than experts in…
I can accept what you're saying, while knowing the exact measure of how many people truly know their limits is unknown, but at least it is nonzero
Re: Claude can now search the web
#743I usually find Claude to be my favourite flavor of LLMs, but I still pay for ChatGPT because their voice offering is so great! I regularly use it as an "expert on the side" when I do other things, like doing bike repairs. I ask it things like "how do I find the min/max adjustments on my particular flavor of front derailleur", or when cooking, and my hands are dirty, I can ask stuff like "how much X do I usually need…
Is it possible to use ChatGPT voice feature in a similar manner to Alexa where I only need to say an activation word? I’m aiming to set up a system for my 7-year-old son to let him engage in conversations with ChatGPT as he does with Alexa.
Re: Claude can now search the web
#744I usually find Claude to be my favourite flavor of LLMs, but I still pay for ChatGPT because their voice offering is so great! I regularly use it as an "expert on the side" when I do other things, like doing bike repairs. I ask it things like "how do I find the min/max adjustments on my particular flavor of front derailleur", or when cooking, and my hands are dirty, I can ask stuff like "how much X do I usually need…
Is it possible to use ChatGPT voice feature in a similar manner to Alexa where I only need to say an activation word? I’m aiming to set up a system for my 7-year-old son to let him engage in conversations with ChatGPT as he does with Alexa.
I assume it would be possible to make yourself with the OpenAI api together with a locally run voice model to only detect the activation word. There might be of the shelf solutions for this, but I am not aware of any.
Re: Claude can now search the web
#745Earlier quoted context omitted.
“Prompting” is kind of a myth honestly. Think about it, how much marginal influence does it really have if you say OP’s version vs a fully formed sentence? The keywords are what gets it in the area.
I find that providing more context and details initially leads to far more success for my uses. Once there’s a bit of context, I can start barking terms and commands tersely.
I find more hallucination - like when you're taught as a child to reflect back the question at the start of your answer.
If I am not careful, and "asking the question" in a way that assumes X, often X is assumed by the LLM to be true. ChatGPT has gotten better at correcting this with its web searches.
I am able to get better results with Claude when I ask for answers that include links to the relevant authoritative source of information. But sometimes it still makes up stuff that is not in the source material.
Re: Claude can now search the web
#746Earlier quoted context omitted.
That is absolute bonkers. I hope you’re having fun, because that kind of logic won’t lead you anywhere where people get paid to reason.
You do understand you're on a site composed mostly of people who get paid to reason all day, right?
I’m sure this site is an outlet for a lot of straight thinkers, but that doesn’t mean we’re being entirely rational in this discussion.
Re: Claude can now search the web
#747Earlier quoted context omitted.
You do understand you're on a site composed mostly of people who get paid to reason all day, right?
I’m sure this site is an outlet for a lot of straight thinkers, but that doesn’t mean we’re being entirely rational in this discussion.
I agree, but I'm sure you can guess why?
Re: Claude can now search the web
#748Earlier quoted context omitted.
> This isn't a vacuum charged with crawling the web, it's an adhoc GET request. Doesn't matter. The robots-exclusion-standard is not just about webcrawlers. A `robots.txt` can list arbitrary UserAgents. Of course, an AI with automated websearch could ignore that, as can webcrawlers. If they chose do that, then at some point, some server admins might, (again, same as with non-compliant webcrawlers), use more drastic m…
In the limit of the arms race it's sufficient for the robot to use the user's local environment to do the browsing. At that point you can't distinguish the human from the robot.
That's not how many of these services work though. The websearch and subsequent analysis of the results by an LLM are done from the servers of whoever supplies the solution.
Re: Claude can now search the web
#749Earlier quoted context omitted.
No thank you, when I define a robots.txt file I expect all automated systems to respect it.
Then you’ve fundamentally misunderstood what a robots.txt file does or is even intended to do and should reevaluate if you should be in charge of how access is or is not prevented to such systems. Absolutely nothing has to obey robots.txt. It’s a politeness guideline for crawlers, not a rule, and anyone expecting bots to universally respect it is misunderstanding its purpose.
> Absolutely nothing has to obey robots.txt
And absolutely no one needs to reply to every random request from an unknown source.
robots.txt is the POLITE way of telling a crawler, or other automated system, to get lost. And as is so often the case, there is a much less polite way to do that, which is to block them.
So, the way I see it, crawlers and other automated systems have 2 options: They can honor the polite way of doing things, or they can get their packets dropped by the firewall.
Re: Claude can now search the web
#750Earlier quoted context omitted.
While robots.txt is not there to directly prevent automated requests, it does prevent crawling which is needed for search indices. Without recursive crawling, it will not possible for a engine to know what are valid urls[1]. They will otherwise either have to brute-force say HEAD calls for all/common string combinations and see if they return 404s or more realistically have to crawl the site to "discover" pages. The…
> Robots.txt does absolutely apply to LLMs engines and search engines equally. It does not. It applies to whatever crawler built the search index the LLM accesses, and it would apply to an AI agent using an LLM to work recursively, but it does not apply to the LLM itself or the feature being discussed here. The rest of your comment seems to just be repeating what I already said: > Whatever search index they are using…
> It does not.
Yes it does. I am the one controlling robots.txt on my server. I can put whatever user agent I want into my robots.txt, and I can block as much of my page as I want to it.
People can argue semantics as much as they want...in the end, site admins decide what's in robots.txt and what isn't.
And if people believe they can just ignore them, they are right, they can. But they are gonna find it rather difficult to ignore when fail2ban starts dropping their packets with no reply ;-)