Jarvis (Joint AI Research for Video Instances and Streams), not exactly the one from Ironman, but with similar tasks is well known as the conversational AI framework developed by Microsoft. Its target is the improvement of customer engagement and provision of personalised services to their users. Particularly, AI Jarvis, also called, uses natural language processing (NLP) and machine learning to understand user inputs and respond with the proper actions or information.
Now, this might sound kind of similar with Chat GPT, nonetheless, Jarvis can use ChatGPT as the system’s controller so that it can use various different models as necessary to respond to your command. In fact, you can create your own Jarvis adjusted to your needs. As Eric Peterson wrote in BetterProgramming, by using his own “Get Jarvis” (personal controller shortcut), Eric sophisticated his Apple Homekit and Siri to control their smart home devices and create custom voice commands to interact with their virtual assistant. This process is by making a request to Jarvi and then this request is combined with a prompt and sent to the OpenAPI for processing. A JSON response was employed to send and receive the data from the appliances.
What is interesting here is how tailored the system can result with the help of Jarvis rather than Siri commands, a simple example that was shown is a comparison between Siri and “Get Jarvis” about whether to go out based on the weather. Siri replied “I obtained this information”, while GetJarvis advised “if you decide to go outside, you should wear a coat because there is a chance of rain today”. And from there Jarvis can follow-up the conversation if it is asked again.
In another blog, Avram Piltch mentioned that Jarvis can answer standard text questions, queries asking about images, audio or even videos. It has not only the ability of answering but it can also generate images, sounds or videos as well. Moreover, the article provides step-by-step for setting up and using Project Jarvis to obtain an OpenAPI key until “enter your prompt”. Finally, it is relevant to remember some advanced features of Jarvis, such as its ability to integrate with other apps and services, and its use of machine learning algorithms to provide personalised recommendations and suggestions. And, we have to be careful with the “quality of feedback” – if we teach it how to reply with proper information the perceived value will be higher.