Get the App
SLTechnology News&Howtos  ›  IT Information  › 

Smarter than Siri and Alexa, developers use GPT-3 to develop voice assistants

Shulou Source: shulou.com Published: 2023-11-24 14:10:52 10月02日 Update

CTOnews.com, January 20 / PRNewswire-Asianet /-- A web developer recently launched GPT-3 voice assistant, which is smarter than Apple's Siri and other voice assistant functions. Developer Mate Marschalko posted a blog post on Medium with the goal of replacing Siri with GPT-3.

He said in his blog post:

Ever since I tried ChatGPT and GPT-3, I feel that everything else is creepy. This includes Siri, Alexa, Google Assistant, and all other intelligent assistants.

If I could replace Siri with GPT-3, that would be great. I made GPT-3 the smart brain of my family, explaining in detail what it could access around the house and how to respond to my request.

CTOnews.com small class: birth shaping pre-training Transformation Model 3 (GPT-3) is an autoregressive language model designed to use deep learning to generate natural language that humans can understand. GPT-3 is trained and developed by OpenAI, an artificial intelligence company based in San Francisco, and the model design is based on a language-changing model developed by Google. GPT-3 's neural network contains 175 billion parameters, which is the largest neural network model in history.

What is impressive in the blog post is that Marschalko does not need to develop any code to achieve this goal. He simply described to GPT-3 what he wanted to do and asked GPT-3 to reply in English and provide an appropriate JSON response.

As you can see, I explained everything in simple English. I described the type of request, the exact structure of the response, and asked it to behave like a perceptual artificial intelligence, and even give advice on personal problems.

I also provided some details about the time and place, as well as the equipment and rooms in the house. As a result, we will receive a perfectly structured response. That's what programming it is all about!

He tested it in his blog post, and the test results made him very satisfied.

For example, one of our requests is: "I just noticed that I recorded this video in a dark office. What can you do about it?"

The response from GPT-3 is:

{"action": "command" # Action: command "location": "office" # location: office "target": "light" # Target: lighting "value": "on" # value: open "comment": "Turning the light on for you", # comment: turn on the lights for you "scheduleTimeStamp": "# timing}

Tags: Intelligence development model assistant goal network language developer voice thing artificial intelligence office parameter house lighting neural network structure English Apple Docker Huawei Linux macOS MariaDB Microsoft MySQL NVidia OPPO Reno Apple MariaDB NVidia OPPO Reno Xiaomi