I am struggling to understand the differences between AI Unit and ChatGPT’s API. Since ChatGPT can extract information such as dates and amounts from PDFs, I am wondering what the key distinction is between it and UiPath’s Document Understanding.
Additionally, I am somewhat concerned that ChatGPT’s intelligence is growing exponentially, which could lead to a situation where UiPath’s AI might appear less capable in comparison. Could you please provide clarification on this matter?
There is no comparison between AI units and Chatgpt API as both are totally different things.
AI units is a kind of currency your bot will require to use AI Center models and their capabilities.
Chatgpt API is used to integrate OpenAI model into your bot for various purposes.
Regarding UiPath AI will fall behind Chatgpt,
UiPath AI is specialised AI build for specific tasks whereas Chatgpt is general purpose.
Even if UiPath AI falls behind any other AI models out there in market, you always have integration feature available in UiPath ecosystem which will help you to use that models in your bot.
If you have open ai subscription already then ypu can use that connector
If not you can take ai units and use UiPath ai units with its own model…also ai units are consumed for various other purposes as well like generation of xamls apps etc
Chatgpt is a general llm model and UiPath model is curated and more specific to the task done on UiPath…may be chatgpt can be better in few in the same way UiPath ai might be better in few…
If there is a specific example of this statement it would be easier to imagine.
ChatGPT can also be introduced to ChatGPTs to give specialized replies in specific areas.
Yes chatgpt can be done like that…that level of grounding is already done for UiPath llms…where as for chatgpt we have to do it again and then it can be consumed
Document Understanding model in UiPath is specialised in understanding documents like invoice,bills etc. when we use it for extracting data from invoice it will give more accuracy.
Where as Chatgpt will require context grounding about what we need and how we need etc…