I've been using it for more than 3 years now. The charge usually lasts for a week. I normally put it for a charge during the night, so it's not that inconvenient for me.
I think it has become a meme now that Apple can't even design practical things, but people will still buy it. And this just adds to it.
> On one evaluation created to test developers’ attempts to have models use computers, OSWorld, Claude currently gets 14.9%. That’s nowhere near human-level skill (which is generally 70-75%), but it’s far higher than the 7.7% obtained by the next-best AI model in the same category.
Here, "next-best AI model in the same category" referes to which model.
On their "Developing a computer use model" post they have mention
> On one evaluation created to test developers’ attempts to have models use computers, OSWorld, Claude currently gets 14.9%. That’s nowhere near human-level skill (which is generally 70-75%), but it’s far higher than the 7.7% obtained by the next-best AI model in the same category.
Here, "next-best AI model in the same category" referes to which model.
I'm curious, what does your human verification process look like? Does it involve a separate interface or a generated report of some kind? I'm currently working on an tool for personal use, that records actions and triggers them at later stage on when specified event occurs. For verification, generating a CSV report after the process is complete and backing it up with screen recordings.
It's a separate interface where the output of the LLM is rated for safety, and anything unsafe opens a ticket to be acted upon by the medical professionals.
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).
Hey, I know the feeling! It can be frustrating when we can't find something we're sure we saw on HN.
When default search failed me in the past, I've had some luck with experimental search projects that people have created. I'm sharing them hoping one might help you find what you're looking for:
Thanks for these! I tried a few of these and https://hn.lixiasearch.com/ turned up exactly what I was looking for as the first result. Gonna have to bookmark that one. I also had a HN story I've been looking for that I couldn't find via HN's algolia search, nor any major search engine.
This was last night so I'm forgetting exactly what I searched but it was something like: "blog about a developer helping his mom with her chromebook she couldn't update".
HN's algolia search is sometimes helpful but you need to know exactly what was either in the title or in the comments. I tried multiple keywords like "ChromeOS", "mom", "javascript error", "chromebook", and so on. I couldn't remember the exact details of the story. Now that I know "optional chaining" is in the title, I can find it right away.
All of them except the last one were nice. Turned up a lot of results I wasn't getting from Algolia, DDG, or Google. Unfortunately, even after digging through pages and pages, I couldn't find what I was looking for. :-(
I might just have to say "I messed up by not bookmarking this" and accept that I'll likely never see it again.
This discussion isn't going in the direction I was hoping for, ha ha!
But as a developer of these sites I've used the majority to check my own domains and the most reliable one I've found is https://www.ipqualityscore.com/
I strongly recommend that you take another look at your signup process though and don't require users to give up their email addresses, instead make it optional and present an alternative like a captcha.
I've never evaluated any myself, but became acutely aware of them after launching my own temporary mail service as a learning exercise. I think there's a github repo somewhere that's regularly updated with a block list of these cycled domains though.
I remember using it at a startup around 2017-18. At the time, I thought it was just another tool they were experimenting with. But after moving on to another job, I found myself missing it.
