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

Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

anthropic.com

Post reply on HN
Print view View original

661–670 of 758 posts

Re: Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

#661

Post by _factor »

alentred wrote: post #659

If "computer use" feature is able to find it's way in Azure, AAD/Entra, SharePoint settings, etc. - it has a chance of becoming a better user interface for Microsoft products. :) Can you imagine how simple the world would be if you'd just need to tell Claude: "user X needs to have access to feature Y, please give them the correct permissions", with no need to spend days in AAD documentation and the settings screens m…

Sure, Ted. I’ve let user HAL access feature “door locks”. I’ve corrected all permissions accordingly.

Re: Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

#662

Post by okasaki »

jandrese wrote: post #264
wild_egg wrote: post #240

Earlier quoted context omitted.

It's bad at directly working on classical computer problems like math and data processing. But you can do it indirectly by having it write a program that produces the correct result. Interestingly, I didn't even have to have it run the program, although usually you would write a tool which counts the number of w's in "strawberry" and return the result Which produced: Here's a simple Python function that counts the nu…

I always thought the halting problem was an academic exercise, but here we see a potential practical use case. Actually this seems pretty dangerous letting the LLM write and automatically execute code. How good is the sandbox? Can I trick the LLM into writing a reverse shell and opening it up for me?

I'm not sure it's still the case, but I've had ChatGPT run shell commands. But I don't know what you could do since it's ephemeral, doesn't have internet access or root. Plus I'm sure they have security scanning.

Re: Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

#663

Post by criddell »

criddell wrote: post #297
highwaylights wrote: post #81

Completely irrelevant, and it might just be me, but I really like Anthropic's understated branding. OpenAI's branding isn't exactly screaming in your face either, but for something that's generated as much public fear/scaremongering/outrage as LLMs have over the last couple of years, Anthropic's presentation has a much "cosier" veneer to my eyes. This isn't the Skynet Terminator wipe-us-all-out AI, it's the adorable…

As a Kurt Vonnegut fan, their asterisk logo on claude.ai always amuses me. It must be intentional: https://en.m.wikipedia.org/wiki/File:Claude_Ai.svg https://www.redmolotov.com/vonnegut-ahole-tshirt

I asked Claude if its logo choice was an intentional Vonnegut reference by Anthropic, would that be upsetting:

> If Anthropic intentionally referenced Vonnegut's irreverent artistic style, I wouldn't be bothered. After all, Vonnegut used humor and seemingly crude imagery to explore deep questions about humanity, consciousness, and free will - themes that are quite relevant to AI. It would be a rather clever literary reference.

Re: Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

#664

Post by distalx »

falcor84 wrote: post #496
mewpmewp2 wrote: post #464

Earlier quoted context omitted.

True, 2FA and all these little details that exist now have made this automation quite insanely complicated. It is of course necessary that we have 2FA etc, but there is huge potential in solving this I believe.

From a security standpoint, what's considered the "proper" way of assigning a bot access based on a person's 2FA? Would that be some sort of limited scope expiring token like GitHub's fine-grained personal access tokens?

Maybe we could build an authenticator as part of the RPA tool or bot client itself. This way, the bot could generate time-based one-time passwords (TOTPs).

Re: Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

#665

Post by klibertp »

GaggiX wrote: post #134
accrual wrote: post #121

Earlier quoted context omitted.

I have to agree. I've been chatting with Claude for the first time in a couple days and while it's very on-par with ChatGPT 4o in terms of capability, it has this difficult-to-quantify feeling of being warmer and friendlier to interact with. I think the human name, serif font, system prompt, and tendency to create visuals contributes to this feeling.

>it's very on-par with ChatGPT 4o in terms of capability The previous 3.5 Sonnet checkpoint was already better than GPT-4o in terms of programming and multi-language capabilities. Also, GPT-4o sometimes feels completely moronic, for example, the other day I asked for fun a technical question about configuring a "dream-sync" device to comply with the "Personal Consciousness Data Protection Act", and GPT-4o just replie…

Garbage in, garbage out. The ability to recognize absurd statements has nothing to do with correctly processing them. You're looking for something LLMs don't have in them; that doesn't mean there's nothing useful in them.

Re: Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

#666

Post by Klathmon »

accrual wrote: post #281
gzer0 wrote: post #182

One of the funnier things during training with the new API (which can control your computer) was this: "Even while recording these demos, we encountered some amusing moments. In one, Claude accidentally stopped a long-running screen recording, causing all footage to be lost. Later, Claude took a break from our coding demo and began to peruse photos of Yellowstone National Park." [0] https://x.com/AnthropicAI/status/1…

Seeing models act as though they have agency gives me goosebumps (e.g. seeking out photos of Yellowstone for fun). LLMs don't yet have a concept of true intent or agency, but it's wild to think of them acquiring it. I have been playing with Mindcraft which lets models interact with Minecraft through the bot API and one of them started saying things like "I want to place some cobblestone there" and then later more gen…

The one that gets me is the issue they found while testing gpt-4o where it stopped mid sentence, shouted "No!", then cloned the users voice and began speaking as them.

https://arstechnica.com/information-technology/2024/08/chatg...

Re: Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

#667

Post by CamperBob2 »

dzhiurgis wrote: post #423
tkellogg wrote: post #410

Earlier quoted context omitted.

Honestly, this is going to be huge for healthcare. There's an incredible amount of waste due to incumbent tech making interoperability difficult.

Or you'll start getting a captcha while trying to pump insulin

(Shrug) AI is now better at CAPTCHAs than I am, so bring it on I guess.

Re: Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

#668

Post by pc86 »

aduffy wrote: post #374
LASR wrote: post #344

This is actually a huge deal. As someone building AI SaaS products, I used to have the position that directly integrating with APIs is going to get us most of the way there in terms of complete AI automation. I wanted to take at stab at this problem and started researching some daily busineses and how they use software. My brother-in-law (who is a doctor) showed me the bespoke software they use in his practice. Runni…

This has existed for a long time, it's called "RPA" or Robotic Process Automation. The biggest incumbent in this space is UiPath, but there are a host of startups and large companies alike that are tackling it. Most of the things that RPA is used for can be easily scripted, e.g. download a form from one website, open up Adobe. There are a lot of startups that are trying to build agentic versions of RPA, I'm glad to s…

Is "AI SaaS bro discovers not everything has a JSON API" the new "startup bro just reinvented a bus"?

Re: Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

#669

Post by rmnwski »

highwaylights wrote: post #81

Completely irrelevant, and it might just be me, but I really like Anthropic's understated branding. OpenAI's branding isn't exactly screaming in your face either, but for something that's generated as much public fear/scaremongering/outrage as LLMs have over the last couple of years, Anthropic's presentation has a much "cosier" veneer to my eyes. This isn't the Skynet Terminator wipe-us-all-out AI, it's the adorable…

Not irrelevant at all! Compare their branding to that of Boston Dynamics. Their branding of robots reminds me more of a Black Mirror episode... If Claude was a dog like robot, it sure would look like a golden retriever or something. Positive AI branding should create a positive public perception which in turn should create a positive attitude towards AI regulation.

Re: Computer use, a new Claude 3.5 Sonnet, and Claude 3.5 Haiku

#670

Post by sensanaty »

jumping_frog wrote: post #604
gavin_gee wrote: post #474

Earlier quoted context omitted.

LLM's are enabling a reimagination of UI. Wheres the biggest opportunity in UI that hasn't kept up to date? legacy and regulated software in the long tail. Thats the disruption opportunity for LLM's.

Imagine a Banking website. It has lot of predefined flows of what can be achieved. These flows have steps arranged in a dependency graph. From the server side, a llm can ask users for inputs to satisfy the dependencies of the task which user wants to accomplish. We will have intuitive UI interfaces in all languages of the world.

But if it's a predefined list of flows, you can just throw Selenium/Puppeteer/Playwright/whatever other automation tool at it, rather than relying on an unstable AI that will do different things every time.

This is just a solution in search of a problem. AIs aren't reliable enough if the content changes constantly since it'll just click on the "close my account" button for no reason half the time, while a legacy website with no changes is much easier to program a tool like Selenium around than relying on the AI that will still be making random choices half the time.

Post reply on HN