Hacker News new | ask | show | jobs
by snowwrestler 778 days ago
I’m confused. Which of Apple’s lunch is OpenAI eating? OpenAI does not offer a voice assistant that competes with Siri. Google and Amazon do, and their usage is way down as well; voice assistants in general have lost consumer preference.
2 comments

You must not have tried the voice feature in the ChatGPT app yet. Try it out, it’s amazing. It pairs the power and functionality of a conversational voice assistant with all the power of ChatGPT and LLMs. Obviously the integration isn’t as convenient as Siri’s yet, but the utility is infinitely higher.

My point is that it seems pretty clear that the future is in the space that OpenAI is right now. And that isn’t a bet that Apple was investing in very heavily.

Apple is also not in the business of losing money. It's much easier to make something new and shiny when you can run it at a loss and your investors are happy for you to light piles of money on fire year after year.

That's also probably why most AI research published by Apple is about on-device inference. It's expensive to run inference servers at scale. Apple is a hardware company, so it makes sense they want to focus on what you can do on a local device (or more accurately, how they can sell you a new piece of hardware).

On-device also preserves privacy. I think that's the main reason Apple prefers it, not the cost of servers.
Not really, for those of us that rather speak our native languages.
FWIW I use ChatGPT’s voice feature in French and it works extremely well
Interestingly, for my elderly Russian-speaking mom’s benefit, I had an English conversation with it and asked it to reply only in Russian. It complied, and spoke and reasoned no worse than usual.
Not yet. Voice assistants will massively benefit from LLM integration which will benefit OpenAI for sure.

It's just that nobody has built one yet, I'm surprised because it's a very suitable application. But I think the cost is much higher than the current scripted models, which means there must be a payment model attached. Right now all the major voice assistants are free and I have a feeling they're all waiting to see who makes the first paid LLM-based product, and how the market reacts.

I think there is an important distinction to make. Siri is actually two things — there are voice informers (“Please explain how photosynthesis works”) and voice assistants (“Remind me to call Dave when I reach my work location”).

The former is really straightforward to implement with an LLM — it’s basically what an LLM is.

The latter is a whole different story.

It's not really. I've done the latter.

You just prompt it with "If I'm asking you to call someone, please output only "<CALL>" and the name of the person". Then capture that keyword.

It works fine like that.

I mean, the latter is basically OpenAI’s assistant, that has some APIs it can autonomously call.

The open problem here is making it on-device, and privacy preserving. Though I’m optimistic about this, as Apple has bought up a huge number of AI startups in the last couple of years, so they are probably onto something.