3 ms·
I know where you are coming from, but I don’t agree. Voice Assistants were always a toy. It was pretty clear from day 1. Using LLM as a replacement for search i
by dougmwne 4y ago
I know where you are coming from, but I don’t agree. Voice Assistants were always a toy. It was pretty clear from day 1. Using LLM as a replacement for search is an intriguing possibility. I have started using it alongside Google and Wikipedia to try to break into new topics. It is useful in a new way, but its ability to make up info is a flaw that may prevent it from ever being a truly helpful tool. Nevertheless, it has a proven value and a place in the world right this very moment.
I can certainly imagine a more refined and truthful version of this that can cite its sources and access real-time data taking over Search, but we seem to be a long way off from that dream still.
- lagrange77 4y ago> I have started using it alongside Google and Wikipedia to try to break into new topics. It is useful in a new way, but its ability to make up info is a flaw that may prevent it from ever being a truly helpful tool. Exactly! Getting an introduction to new subjects is also my favourite use case, while it's this use case, that suffers a lot from wrong information, since the users lacks domain knowledge to tell. > I can certainly imagine a more refined and truthful version of this that can cite its sources and access real-time data taking over Search, but we seem to be a long way off from that dream still. Right, truthfulness (as per the training data of course) and citation would fix the most significant shortcomings.
- dougmwne 4y agoI have seen some really fascinating recursive approaches to GTP where formal logic and flow are used to conditionally re-prompt GTP to keep refining the answer. I think there will be an incredible power in combining the “intuitive” abilities of GTP with more traditional analytical AI approaches.