Skip to main content
MANIFOLD
Will an AI voice assistant with GPT-4 level capabilities be available on a smart device with voice activation by 2025?
31
Ṁ630Ṁ1k
resolved Oct 30
Resolved
YES

Approximately GPT-4 level sufficient, needs to be similar to Siri or Alexa in that can be activated by a voice command at any time.

Market context
Get
Ṁ1,000
to start trading!

🏅 Top traders

#TraderTotal profit
1Ṁ33
2Ṁ11
3Ṁ9
4Ṁ9
5Ṁ8
Sort by:

It's highly plausible that a GPT-4-powered AI voice assistant will be integrated into smart devices with voice activation by 2025. Given the rapid advancements in AI and language models, a GPT-4 system could offer even more natural and context-aware interactions, significantly enhancing user experience. The demand for seamless voice-operated smart devices and the technology's evolving trajectory suggest a feasible timeline for this integration. However, technical challenges and ethical considerations may influence the actual deployment and widespread adoption. Overall, the convergence of technology trends makes the scenario likely, though specific developments will depend on various factors and industry progress.

-Hines, K. (2023, November 6). OpenAI Introduces GPT-4 Turbo, Custom GPTs, Assistants API, And More. Search Engine Journal. https://www.searchenginejournal.com/openai-introduces-gpt-4-turbo-custom-gpts-assistants-api-and-more/500485/

predictedYES

Did you find anything that qualified, or just decided it was guaranteed to happen

@VAPOR I found this to qualify.

@dominic Yes! Thank you!

What would the definition of GPT-4 level capabilities include? Does the voice assistant need to have all the capabilities GPT-4 has? If not, which ones are required and is assessing benchmark tests necessary to determine if the capabilities the voice assistant has are actually on par with GPT-4? My assumption is that smartphone voice assistants will be augmented utilizing LLMs, but that capabilities will be more limited than is the case with GPT-4. Alternatively, would just having the ChatGPT app on a smartphone with a voice control option count?

@TP8ac2 Doesn't need to have all the capabilities, basically just a more-or-less equally good text generation based on general opinion. I'm open to any measurement that can be more objective. If the ChatGPT app can be voice activated without separately opening it first, then it will count.

It's got WiFi yes? And it can do Bluetooth? So it can go online and log into chatgpt 4 and do voice over Bluetooth? Besides this being an likely yes for me, can a tiny raspberry pi with a wide wrist strap be put on your wrist, a video taken and posted to the comments and it resolves to yes?

@VAPOR Voice-activated like 'Hey Siri' and that's native to the software.

predictedYES

@Nostradamnedus right, which (Siri etc) calls to a chatgpt clone to use that web services voice feature to do your functionality, as in it's not a sophisticated LLM running as software entirely in the smart watch, right?

@VAPOR If you mean whether it can be web-based, yes it can be.

predictedYES

@Nostradamnedus someone could do it by strapping a raspberry pi to their wrist right now, and it would resolve yes for you?

There should be a commercial one like this within a year I guess...

@VAPOR I don't understand how that includes the voice activation part without any additional steps.

predictedYES

@Nostradamnedus it would be a mini computer strapped to the wrist capable of everything a smart watch could, that would qualify as YES on your question, and could do it today. It would have an internet connection and audio in/out. It would be a chunky mess on the wrist, and probably cost <$200.

https://platform.openai.com/docs/models/whisper here is audio in > speech recognition > to text, available in the open AI APIs, which you can then put into chatgpt 4 API, but I don't see how to get speech audio back out, maybe find some speech synthesis thing run on the output text. The API might be expensive to use, but it's just a proof. Some big company can make it all cheaper i guess.

The additional steps are additional steps whether it's done by the maker of the smartwatch company, or some hobbyist, define "additional steps" it's all the same, just a fancy smartwatch hides that same functionality from you with a slick UI

@VAPOR Hmm.. I guess that sounds like it would technically fit the criteria but someone would have to make one.

predictedYES

@Nostradamnedus chatgpt 4 voice has not been out long, maybe someone will make one soon.

Just looked around some more, I found an Apple Watch Series 9, has cellular and Bluetooth, but the chatgpt app is not compatible with it yet, but any day now it might be, as pure voice mode solves the UI difficulties of the watch face that's holding that development back.

There are chatgpt clones available for the Apple watch but none of them are likely chatgpt 4 level or 2way voice enabled like chatgpt 4 is now.

Few months YES