{"id":1371,"date":"2024-05-25T09:35:39","date_gmt":"2024-05-25T07:35:39","guid":{"rendered":"https:\/\/eltecnologuia.com\/?p=1371"},"modified":"2024-05-25T09:39:19","modified_gmt":"2024-05-25T07:39:19","slug":"doubts-about-lam-threaten-rabbit-r1","status":"publish","type":"post","link":"https:\/\/eltecnologuia.com\/en\/doubts-about-lam-threaten-rabbit-r1\/","title":{"rendered":"Doubts about the LAM threaten the Rabbit r1"},"content":{"rendered":"

First news about the Rabbit r1 Teach Mode arrives<\/a><\/h4><\/header><\/div>
<\/div><\/div><\/article>\n

The Rabbit r1 is updated: Beta Rabbit, timers and alarms<\/a><\/h4><\/header><\/div>
<\/div><\/div><\/article>\n

Cybersecurity experts put the Rabbit r1 to the test<\/a><\/h4><\/header><\/div>
<\/div><\/div><\/article>\n<\/div><\/section>\n\n\n

If there's one thing the Rabbit r1 community is asking for, it's proof that LAM works and is real.<\/h2>\n\n\n\n

It's been a little over a month since the folks at Rabbit released what is known as Pickup Party <\/a>its r1 to the first buyers. The delivery of the devices became a reality and can be used for the first time <\/strong>in the hands of someone other than Jesse Lyu.<\/p>\n\n\n\n

And until that moment the only demos <\/a>and therefore the only sign that everything that had been announced worked came from the hands of Jesse and part of his team, insisting on show us how Spotify worked<\/strong> and leaving aside the others.<\/p>\n\n\n\n

This raised some doubts about Doordash, Uber or Mdjourney, given that there was no trace of the other applications<\/strong> which in theory were going to be managed thanks to the LAM<\/a>. After all, they had already been delivered and it was just a matter of waiting.<\/p>\n\n\n\n

The lack of functionality of the LAM on a day-to-day basis <\/h2>\n\n\n\n

Once the first units reached the clients, media and influencers who had purchased it, that was when It was possible to see in the testing field how well they worked<\/strong> those applications, the LAM and if it really was something polished even if it were in those 4 apps.<\/p>\n\n\n\n

Spoiler: It wasn't.<\/strong><\/em><\/p>\n\n\n\n

Some Youtubers like Unbox Therapy<\/a> either iJustine<\/em> apparently They didn't have any problems<\/strong>, or at least it is not shown in their videos, and in fact in the tests they do they seem quite functional in this regard.<\/p>\n\n\n\n

But it was shortly after when The networks were filled with accusations about the LAM<\/strong> and that it was useless in its state, the accusations, the fights and the lack of critical sense began that have led us directly to the point where we are.<\/p>\n\n\n\n

Internet divided with the LAM<\/h2>\n\n\n\n

This is where several groups emerge regarding this issue. On the one hand there are those who They see that if the Large Action Model does not work<\/strong> Nowadays it is reason enough to return the product because they have been deceived.<\/p>\n\n\n\n

On the other hand, there are the defenders who, even knowing that everything that surrounds the Apps<\/em> managed by automation need to improve, they deny the reality that It's not even close to being something that works.<\/strong> optimally.<\/p>\n\n\n\n

And on the other hand, where I like to place myself: I accept that it is not finished or functional today, but I understand that it is There is a great project behind it and they are working<\/strong> in implementing the necessary improvements.<\/p>\n\n\n\n

The Coffeezilla video and the criticism of Rabbit<\/h2>\n\n\n\n

He Youtuber<\/em> coffeezilla<\/a> published 2 videos talking about Rabbit<\/strong> throughout these last few weeks. The first related to Jesse Lyu's past and the second, which is what interests me, about the LAM and if it really was what they promised.<\/p>\n\n\n\n

In short, the video accused the company that the reality of the Rabbit R1 is far from the promises made. According to him, the investigation revealed that the device does not use its own advanced AI<\/strong>, but relies heavily on existing technology like ChatGPT and pre-built scripts using Playwright, a web automation tool.<\/p>\n\n\n\n

On the one hand ChatGPT<\/strong> for Rabbit R1 responses that do not come from an internally developed AI and from another Predefined scripts<\/strong> whose functions are executed through predefined scripts instead of adaptive AI.<\/p>\n\n\n\n

This particularly surprises me on the part of Coffeezilla, since it was never said that the answers would be given by its own AI, but rather that it would use Anthropic, Perplexity or indeed ChatGPT, so I don't understand the surprise.<\/p>\n\n\n\n

Another criticism was Web automation<\/strong>, and it seems that the Rabbit R1 uses Playwright to automate interactions with websites, which is not a true use of AI. And I insist, this was public from the beginning.<\/p>\n\n\n\n

The problems of the r1<\/h3>\n\n\n\n

The investigation highlighted several significant issues with the Rabbit R1, both in terms of functionality and privacy and accuracy.<\/p>\n\n\n\n

Functionality Issues<\/h4>\n\n\n\n

In these terms he highlighted the Inability to adapt<\/strong> and that predefined scripts cannot adapt to changes in application user interfaces, resulting in frequent crashes. Besides, the Lack of precision<\/strong> and that the device was unable to provide accurate locations or perform basic tasks reliably.<\/p>\n\n\n\n

Privacy Concerns<\/h4>\n\n\n\n

Regarding privacy using the Rabbit r1, he comments that there are Security risks<\/strong> and that the review of the source code revealed possible vulnerabilities that could allow unauthorized access to user data.<\/p>\n\n\n\n

Privacy was also discussed on Discord and in fact even It is one of the topics we have discussed on this website for a long time.<\/a>, as it was something that was of particular concern to the community and was being corrected.<\/p>\n\n\n\n

Rabbit Inc's response to the LAM<\/h3>\n\n\n\n

Rabbit has responded to these statements with the following announcement that can be read on Discord:<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n
We appreciate everyone's interest in LAM and our security practices, as well as your desire to learn more about how it works.

To this end, we are pleased to share a couple of new posts on our blog that explain both aspects. We have also shared this information with Coffeezilla and other media that have made inquiries.

You can read more about how LAM works here:

LAM Security Architecture Overview<\/a>

And you can learn about our Vulnerability Disclosure Program (VDP) here, and how we look forward to working with any security researchers or members of the general public to help improve our security.

Vulnerability Disclosure Program (VDP)<\/a>

At Rabbit, we stand behind our product, technology, and vision of building a custom operating system that can understand users' intentions and intuitively help them accomplish their tasks.

We firmly believe that we are building something new in a frontier category and that requires a different approach from the personal devices already on the market.

Native AI products are rapidly evolving and improving at exponential rates, and we are excited to see what we can build in this new industry.<\/div>