HN2new | past | comments | ask | show | jobs | submitlogin

It's funny because the moment this is available to run on your machine you realize how useless it is. It might be fun to test its conversational limits, but only Siri can actually set an alarm or a timer or run a shortcut, while this thing can only blabber


Can you run Siri outside of iOS? Can you work on it? FLOSS can help there, I could run this locally on a RasPi or old laptop if I want


This is not a deterministic assistant like Siri, this is a ChatGPT conversational tool that might act up if you ask it to do anything


I don't want to sound dismissive, but 3rd party integration is part of the roadmap and any project has to start somewhere. I will admit I am kinda excited to have an alternative to commercial options.


It's pretty bad at baking a cake too.

It's a chatbot, not a home automation controller. It's a research&writing assistant, not an executive assistant.


How can it be a research assistant if it keeps making up stuff?


How can humans be research assistants if they make shit up all the time?


If I tasked an assistant to provide 10 papers, and 8 of them turned out to be made up they would be fired instantly. Unless someone wants to actively scam you, they will always provide 10 real results. Some of them might not be completely on topic, but at least they would not be made up


I don't see why you couldn't integrate this kind of thing with some kind of command line, letting it integrate with arbitrary services.


it's not deterministic, I don't want it to interpret the same command with <100% accuracy


It's deterministic. They throw in a random seed with online services like chatgpt.

If it wasn't deterministic for some reason thar wouldn't be because it's magic, it would be because of hardware timing issues sneaking in (same reason why source code compiles can be non-reproducible), and could be solved by ordering the results of parallel computation that doesn't have a guaranteed order.

To the best of my knowledge it's not a problem though.


Are humans deterministic? Hell, I wish my plain old normal digital computer was 100% deterministic, but it ain't due to any number of factors from bugs and state logic errors all the way to issues occurring near the quantum level.

You're setting the goal so high it is not reachable by anything.


I'm already doing this. I currently only accept a subset of possible commands.

The accuracy is a problem, but I think it's my prompting. I'm sure I can improve it by walking it through the steps or something.

You can also just work in human approval to run any commands.


To be fair, Siri’s success rate at setting an alarm is about 3/10 in my household. Let’s give open source a chance here




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: