We're sunsetting PodQuest on 2025-07-28. Thank you for your support!
Export Podcast Subscriptions
People
J
Jaeden Schafer
Topics
Jaeden Schafer:Anthropic 公司终于为 Claude AI 模型添加了网页浏览功能,这是我一直期盼的功能。这项更新解决了 Claude AI 模型长期以来缺乏网页浏览功能的问题,此前由于缺乏此功能,我在许多用例中都推荐用户不要使用 Anthropic 或进入其生态系统。 Anthropic 公司之前没有添加此功能的原因是他们将 Claude AI 模型定位为一个安全的、自包含的 AI,不希望它接触到真实世界。然而,由于缺乏网页浏览功能,Claude AI 模型的实用性受到很大限制,我个人也因此没有购买 Anthropic 的高级帐户。 目前 Claude AI 模型的网页浏览功能仅限于付费用户预览版,且仅在美国可用。其功能与其他 AI 模型类似,会在 AI 响应下方显示网络搜索结果摘要。Anthropic 公司提供的用例很多其他 AI 模型已经具备,这让我觉得他们添加此功能的真正原因可能是为了应对竞争压力,而非仅仅出于安全考虑。 为 AI 模型添加网页浏览功能在技术上并不复杂,Anthropic 公司此举可能是为了避免落后于竞争对手,因为许多竞争对手的 AI 模型已经具备了网页浏览功能。Claude AI 模型的网页浏览功能会提供直接引用,方便用户核实信息来源。 虽然 Claude AI 模型的网页浏览功能带来了便利,但也存在信息准确性方面的争议,因为搜索结果可能会被误导生成完全错误的摘要。然而,对于开发者来说,这项更新弥补了 Claude AI 模型最大的缺点之一,使其更加实用。

Deep Dive

Chapters
Anthropic finally added web browsing to Claude AI. This was a highly requested feature, and its absence was a major drawback for many users. The reasons behind the delay are discussed, including safety concerns and competitive pressures. The release is currently in preview for paid US users.
  • Claude AI now has web browsing capabilities.
  • The feature was previously unavailable due to safety concerns and is now in preview for paid users in the US.
  • Competitor pressure likely influenced the decision to release the feature.

Shownotes Transcript

Translations:
中文

We have some amazing news out of Anthropic. This is something I have been begging for for ages, and that is that they've finally added

web browsing to Claude. Now, Claude is like one of the best models for a lot of different things. And this web browsing that was missing was something I have personally complained about for years at this point, ever since I believe like over a year ago, when they first added browse, like Anthropic didn't. This was literally to the point where I would recommend people not use Anthropic or get into its ecosystem for a lot of use cases because it didn't have features. So

They've added the feature. There's some caveats to this. And they kind of talked about why they didn't have it for so long, why it's coming now. But there's also some other theories. I'm going to be getting into all of that on the podcast today. The first thing I wanted to say is that if you're not already a member, I would love to have you as a member of my AI Hustle school community. So every single week, I create an exclusive podcast.

podcast content that essentially teaches people how to grow and scale their businesses using AI tools. So there's a paid community. It's $19 a month. In the past, I've had this thing at like $100 a month, but it's at a discount right now. If you'd like to join, I break down all of the AI tools, strategies, and workflows that I use to grow and scale my businesses using AI tools. So an amazing community. I'd love to have you as a member. There's over 300 members and everyone shares and

incredible feedback and content. So it's a great place to be. There's a link in the description. All right, let's get into what Anthropic is doing here. So the first thing that I thought was really interesting in regards to all of this, essentially, is the fact is kind of the reason why they said that they, you know, haven't had

They haven't had it available in the past. And the reason why is because they essentially said that this thing was, it's like this self-contained AI and they didn't want out in the real world. So it almost came at like a, it was sort of like a safety thing. And because Anthropic is branded as like the safe AI, they're like, look, we have this contained AI. It's not going to go out and like search the internet.

That made a lot of people happy. It also made a lot of people upset because it was just very unusable. So who has access to this? I hopped on today and was like, sweet, let me use Anthropic. I don't have a premium account with Anthropic if I'm being 100% honest because of this fact, because it didn't have access to this. So I have a premium Grok and premium chat GPT and premium like everything else. But this I was like, ah, you know, it doesn't have that, you know, use perplexity to search the web if I need to. So I was like, kind of got replaced and you can use Anthropic and perplexity together. Oh, um,

I don't have a premium account. When I hopped onto a cloud to try it out today, I realized it wasn't available. So it is only available in preview. So there's, they're not even saying this is full. There's just preview for paid cloud users and only in the United States. So they have a whole blog post where they talked kind of all about this, but in any case,

It is out now. You can actually get access to it. Essentially how this looks is when you do a search, it's going to kind of like chat GPT or a lot of the others pull up a web search below its response. So if you ask it a question, it'll give you kind of its chat GPT or its AI response. And then it goes, scrapes the web and gives you a summary. This is similar to what we're seeing with other things. They gave a few use cases. They're like, I don't know. It's kind of funny to me because they're like, here's some great use cases you can use it for. And it's like, as

as if people that actually need any of those use cases haven't just been using chat GPT up at this point. They're like, for your sales team, it can transform account planning and drive higher win rates through informed conversations with prospects. A financial analysis can assess current market data, earning reports, and industry trends to make better investment decisions. Like, these are kinds of things that was like a non-starter for me, if I'm being 100% honest. It's like, I want to know, okay, what's going on with like X, Y, and Z today? And like, it doesn't know. So I would use Grok for that because it kind of has access to like the

the X API and could get you like all the tweets for the day or whatever. And then chat GPT, which I'm like, hey, like I'm trying to plan a trip. Like what's the weather like in Barcelona today? Blah, blah, blah. And, you know, Claude would be like, oh, I don't have access to it. So like there's all these use cases that I think just got shipped other places. And I think this is the real reason why Anthropic has added this, right? Because originally this was all about safety and

Now it seems like they don't really, I mean, it's not like they don't care about safety. I think they realize that that safety concern of an AI be connected to the internet is probably overblown. Every other competitor is doing it. Every other competitor is, you know, gaining ground on them. This isn't something fundamentally difficult to do. Google has APIs. There's a lot of these web browsers that have APIs that you can plug in directly to, or, you know, they probably built their own, but like there's a lot of options out there. This isn't like a technically complicated,

insanely challenging thing. And so I think they knew they kind of needed to do this or they're going to lose a lot of ground, you know,

You can ask it questions like, what major event is taking place this week in the tech industry? And it's going to give you an actual rundown on this, whereas before it couldn't. So this is what they said in their blog post. They said, when Cloud incorporates information from the web in its responses, it provides direct citations so you can easily fact check sources. This is something everyone else is doing, but I'm happy that they're doing this as well. They also said, quote, instead of finding search results yourself, Cloud processes and delivers relevant sources individually.

Of course, right? It's a no-brainer at this point. This makes perfect sense why they would be doing this. And this is coming after we have not just ChatGPT, but Google Gemini and even Mistral, right? The open source French AI model, their LeChat model has...

access to the internet. So like we're getting every competitor in the book, tiny. I'm not, I mean, Mistral, I hate to call it a tiny company, but like compared to how much Mistral might've raised a billion dollars or so, Anthropica has raised like way more than that. So,

These are very different scales of companies. And when the smaller ones start to do it, like, you know that Claude really has to do it. So just because they were, quote unquote, designed to be self-contained, I think because there's so much competition, it's so competitive that. Now, what are the controversies around this? Well, if you go and ask TechCrunch, what is going on with Anthropics Claude? Is this good or bad? This is the paragraph you'll get out of TechCrunch. And I quote.

A separate report from the Guardian found that ChatGPT's search-focused experience can be fooled into generating completely misleading summaries. Cloud's so much more usable, so I'm excited to actually dig in and use Cloud when it has new features. My development team has specifically told me this is one of the big drawbacks. They love Cloud, and they still do use it as their number one coding source, but they have told me the biggest drawback is it doesn't have access to the internet. So now this is out in the open. This is fantastic. Great news, and kudos to everyone at Anthropic.

Thanks for tuning into the podcast today. If you enjoyed the episode, I would love it if you could leave me a review for the episode. And if you're interested in joining the AI Hustle School community, check that out in the link below. I would love to talk to you about growing and scaling your business or your career with AI tools. So much exciting stuff happening there and we have a phenomenal community. Thanks so much for tuning in and I will catch you next time.