TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Is AI the solution to search's problem?

21 点作者 lucaserb超过 2 年前

15 条评论

baxtr超过 2 年前
Maybe I’m an outliner. But I’m not buying it.<p>To me, it misses to analyze and address a fundamental problem of why google “got so bad”.<p>I believe this is mainly a problem of the top 5%. There is reason why google presents their results like they do. It’s not like they’re a bunch of idiots not knowing their craft.<p>Yes, for those of use around the internet since 1996, the results got worse. But since then the entire population got online. They’re just fine with the top results. They’re good enough. Google’s machine knows and sees this in the data.<p>I am sure there is value in summarizing and querying the text. But this will increase the convenience not necessarily the quality of results.
评论 #34515456 未加载
评论 #34515364 未加载
isaacfrond超过 2 年前
Honestly, &#x27;google has gotten so bad&#x27;, it is just a meme.<p>It is just not true.<p>Use any of the competing search engines for even a day and you&#x27;ll come running back to Google. Believe me, I tried it. I really wanted to switch to Bing or the Duck, but they just don&#x27;t compare. If the search results of Google are not to your liking it is because search is hard. Also login, if you are a super user, Google will catch on to this as well.
评论 #34517369 未加载
评论 #34516947 未加载
评论 #34518475 未加载
评论 #34517138 未加载
Barrin92超过 2 年前
No. Generative AI at least won&#x27;t. ChatGPT is Borges Library of Babel, it produces any reasonably sounding sentence in existence if you ask it to, of course useful information by definition is somehow in there but so is everything else. If your problem is that too much crap is on the internet, well that crap is in ChatGPT, and then some.<p>Even if you only get fiction 20% of the time that&#x27;s still absolute poison for a search engine that sounds completely authoritative. It&#x27;s like having sand on your toothbrush.<p>The solution to abundance of bad info on the internet are two things. One, rudimentary search engine skills, i.e. filtering by source, keyword search and some basic literacy because all the good sources are still there and growing. Secondly, trusted sources curated by people that you have confidence in to produce high quality output.<p>The discourse increasingly reminds me of Alexa&#x2F;&quot;smart&quot; home assistants which people thought would replace everything because the natural language interaction seems futuristic.
LastTrain超过 2 年前
This isn&#x27;t a solution to the search problem but a good option would be to add the ability to search by original index date. Google and DDG have operators for searching by date, but they are limited and seem to be looking for dates declared in the indexed documents, not on an internal date. One advantage to searching by original index date is that it cannot be manipulated by SEO&#x27;s retroactively, so searches can be crafted to get results from before a time when SEOs started manipulating the topic you are interested in.
sovnwnt超过 2 年前
First the LLM needs to know what it doesn&#x27;t know, and it looks like except for the explicit limitations embedded by the developers it isn&#x27;t capable of doing that. When I asked ChatGPT about its own model variants, it gave me wild answers for things that made sense but didn&#x27;t exist.<p>&quot;&quot;&quot;<p>There are several variants of GPT-3 model, each fine-tuned for specific tasks or industries. Here are a few examples:<p>- GPT-3 &quot;davinci&quot; is fine-tuned for creative and imaginative writing tasks, such as poetry, short stories, and song lyrics.<p>- GPT-3 &quot;curie&quot; is fine-tuned for conversational and dialog generation.<p>- GPT-3 &quot;babbage&quot; is fine-tuned for code generation and language understanding.<p>- GPT-3 &quot;einstein&quot; is fine-tuned for question answering tasks.<p>- GPT-3 &quot;jules&quot; is fine-tuned for summarization.<p>- GPT-3 &quot;parliament&quot; is fine-tuned for legal text generation.<p>- GPT-3 &quot;bronte&quot; is fine-tuned for creative writing tasks like storytelling and fiction-writing.<p>&quot;&quot;&quot;<p>I&#x27;d almost assume &quot;parliament&quot; it&#x27;s a leak of some sort if any of the other examples (except for davinci) were accurate.
评论 #34515129 未加载
looseyesterday超过 2 年前
AI is not the solution. The root issue here is Goodheart&#x27;s law. SEO marketers are hacking rankings with mediocre content, this leads to lower returns on time invested for generating valuable content, so writers &#x2F; creators divert their attention elsewhere. Kagi.ai is a good example of how google could be fixed.<p>Generative models dont know what they dont know, they will always struggle on new topics and worse still they wont provide feedback to the user to adjust their query as they will simply hallucinate details
tracerbulletx超过 2 年前
I think the next major leap will be when it accurately recommends real world books, videos, excerpts, papers, and products that naturally flow from your interaction.
AussieWog93超过 2 年前
Anecdotally, I&#x27;ve been using ChatGPT to do the legwork of copying-and-pasting from Stack Overflow, and it&#x27;s honestly really quite good. Have asked it to write maybe a dozen programs, even asked it why a particular line of code won&#x27;t compile, and it does a really good job of explaining how things work.
acadapter超过 2 年前
Google search from 2010-2015 was a great tool. Google search today is a sad shadow of its former self.
swiftpoll超过 2 年前
Very interesting! And the bear-themed solution is cute! I would love to use such a tool to summarize and to ELI5 things like employment contracts or fine prints on a bank statement on the web. This tool might save folks a lot of time and energy.
friend_and_foe超过 2 年前
I think this line of reasoning is probably correct: these language models aren&#x27;t intelligence, they&#x27;re just the next iteration on information lookup systems.
flohofwoe超过 2 年前
I totally read that as &quot;Is AI a solution in search of a problem?&quot; at first, but I guess that summarises the article pretty well too.
评论 #34517179 未加载
danbulant超过 2 年前
This solves the problem of pages full of ads and keywords and irrelevant content. It doesn&#x27;t solve the problem that&#x27;s talked about in the article - there are other solutions, usually alternative search engines (kagi[0], or if more AI=more better, the kagi labs with contextual answers with sources[1]. Not sponsored, just like using them. Similar service offered by Hey.com, but I didn&#x27;t try them much).<p>[0]: <a href="https:&#x2F;&#x2F;kagi.com" rel="nofollow">https:&#x2F;&#x2F;kagi.com</a> [1]: <a href="https:&#x2F;&#x2F;labs.kagi.com&#x2F;ai&#x2F;contextai" rel="nofollow">https:&#x2F;&#x2F;labs.kagi.com&#x2F;ai&#x2F;contextai</a>
lucaserb超过 2 年前
Using OpenAI to research and learn faster online
评论 #34514833 未加载
fedux超过 2 年前
Ads will not go away. Wait until the free-tier model is fine-tuned to promote certain products in its output.