Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Rabbit: LLM-First Mobile Phone

rabbit.tech

Post reply on HN
Print view View original

341–350 of 547 posts

Re: Rabbit: LLM-First Mobile Phone

#341

Post by matthewfcarlson »

gyomu wrote: post #333
dvt wrote: post #321

Earlier quoted context omitted.

> regex compatible Even a typical Product Manager, let alone a regular layperson, would have no idea where to start with regex, so it's a super weird thing to bring up. Of course grep is a great tool, just how the hammer is a great tool (invented around 3 million years ago); but we've moved past that. You can't use a hammer to do EUV lithography.

You're right, most people just need basic word search and replace, which pretty much any text editor supports. Supporting regex covers that extra mile that some advanced power users might need, since you were referencing obscure grep flags. What kind of search and replace are you doing that is akin to EUV lithography?

A good example of something I was doing today. Taking some text and formatting it as a markdown table with proper spacing. Copilot made it super to highlight the line, type out what I wanted, then kept moving on. Manually it was lots of adding spaces, deleting spaces, and adding |s.

Re: Rabbit: LLM-First Mobile Phone

#342

Post by BoppreH »

dvt wrote: post #287

I feel like the R1 is getting a lot of weird hate here on HN. I get it, it's a bit pretentious and wants to be very Apple-like, yadda yadda. Probably not that much value generation, it's overpriced, and no one wants to carry another thing in their pocket. And yes, the company will likely go bankrupt (or pivot into something less exciting). It's also all in the cloud (no local inference) so it doesn't really even "do"…

Assistants are cool, but I want better tools. Give me a timeline with all my emails, chats, documents, and pictures shown together. Full text search on my entire browsing history. Exporting my Booking.com search results as a spreadsheet. Controlling my phone from my desktop, and vice versa.

As a bonus, those capabilities don't require uploading your data to the cloud or risking LLM hallucinations.

Re: Rabbit: LLM-First Mobile Phone

#343

Post by vineyardmike »

BoppreH wrote: post #342
dvt wrote: post #287

I feel like the R1 is getting a lot of weird hate here on HN. I get it, it's a bit pretentious and wants to be very Apple-like, yadda yadda. Probably not that much value generation, it's overpriced, and no one wants to carry another thing in their pocket. And yes, the company will likely go bankrupt (or pivot into something less exciting). It's also all in the cloud (no local inference) so it doesn't really even "do"…

Assistants are cool, but I want better tools . Give me a timeline with all my emails, chats, documents, and pictures shown together. Full text search on my entire browsing history. Exporting my Booking.com search results as a spreadsheet. Controlling my phone from my desktop, and vice versa. As a bonus, those capabilities don't require uploading your data to the cloud or risking LLM hallucinations.

> As a bonus, those capabilities don't require uploading your data to the cloud or risking LLM hallucinations.

Ignoring that half those services are dependent on the cloud already, this is such a bold and hard problem. A solution would be magical but LLMs are our best chance.

Without clearly defined protocols (eg SMTP, CalDav) and a ton of boring to write and use case specific glue code, you’re really dependent on LLMs to make those M*N connections from one service/format to another. An LLM can turn any current or future input format into any current or future output format. That’s really powerful and basically unsolvable at scale with the tools we have now.

Re: Rabbit: LLM-First Mobile Phone

#344

Post by qup »

ryanklee wrote: post #98
muricula wrote: post #70

Earlier quoted context omitted.

Is there a specific paper or something you can point me to? Or are you talking about like llama.cpp? Because I thought that referred to the fact that it was originally one c++ file named llama.cpp?

I assumed it was in reference to llama.cpp. It's a weak assumption, though.

The guy meant CCP, I'm pretty sure.

Re: Rabbit: LLM-First Mobile Phone

#345

Post by fxd123 »

This keynote really makes it seem like a scam. For example the "plan travel" section hardly shows anything. He just says "It's all been planned out, I just confirm, confirm, confirm..."- while the camera cuts to him tapping the device without showing anything on the screen.

Re: Rabbit: LLM-First Mobile Phone

#346

Post by joshspankit »

andybak wrote: post #299
dvt wrote: post #287

I feel like the R1 is getting a lot of weird hate here on HN. I get it, it's a bit pretentious and wants to be very Apple-like, yadda yadda. Probably not that much value generation, it's overpriced, and no one wants to carry another thing in their pocket. And yes, the company will likely go bankrupt (or pivot into something less exciting). It's also all in the cloud (no local inference) so it doesn't really even "do"…

The strangest thing about AI is that a lot of it's value comes from the fact that it's better at using the tools that we designed for humans than we are. It's essentially an admission of failure. I'm not sure what conclusions to draw from this observation however.

> is so broken

I would argue that it’s not that we’re bad at using tools that we designed for humans, but more that multiple shifts in the tech world over the last 50 years have intentionally (and unintentionally) broken those tools. AI does not need to use the tools at all, it knows the knowledge that the tools gate-keep and it can write code that does the data manipulation that underlies the most powerful software.

It essentially side-steps the whole mess that is the current tech industry (and to be honest a bunch of other industries).

Re: Rabbit: LLM-First Mobile Phone

#347

Post by fxd123 »

rchaud wrote: post #193
supafastcoder wrote: post #47

These demos are always the same: “ find me the closest pizza” Who actually interacts with a device like that? Is it an important problem enough to be paying $199 while I can do the same thing (often even better) with my phone?

People that buy Alexas in droves.

"Amazon Alexa is a “colossal failure,” on pace to lose $10 billion this year" (2022)

https://arstechnica.com/gadgets/2022/11/amazon-alexa-is-a-co...

Re: Rabbit: LLM-First Mobile Phone

#349

Post by pjfitzpatrick »

jonplackett wrote: post #307
covercash wrote: post #290

Earlier quoted context omitted.

Right after you stopped watching, he books an entire vacation - flights, hotel, car rental, activities - and says he is given multiple options to choose from but shows none of the options, just an itinerary that he quickly scrolls through.

The idea that anyone is going to book a vacation this way is ridiculous. Vacations take extensive thought. There’s no way a voice interface would work for this even if it was the Star Trek computer.

I think is true. I'd want a far better understanding of the details before committing to spend the time and money.

Re: Rabbit: LLM-First Mobile Phone

#350

Post by pjfitzpatrick »

cal85 wrote: post #322
squarefoot wrote: post #283

This is pretty much what everyone saw coming when LLM became a reality: that is, eliminating the usual human-machine interface because now the machine can understand what we say and instructing it through keypresses and clicks becomes unnecessary. In a few years everyone will carry a device like this one, and "call Jim, the one I met the other day, not my brother" or "start writing an email to all my associates" etc.…

I mostly agree but think you underestimate the “contexts where we actually want visual interaction” - a screen is not just for consuming visual media. It might be that all input is spoken, but output will always need a screen, because spoken word is just incredibly low bandwidth compared to a good text layout. Imagine an order confirmation page for a complex food delivery order, where you’ve made a couple of item mod…

great examples—that makes a lot of sense
Post reply on HN