New GPT Services for the University Community

(December 2023)

To continue enhancing productivity and improve the academic pursuits at CityU, the following services are newly available or being investigated:

  1. CityU IT Chatbot
  2. Copilot: GPT-4V with Current Data & Vision
  3. Microsoft Copilot for Microsoft 365

CityU IT Chatbot

The chatbot is designed to answer enquiries about IT services, which helps you quickly locate and respond to your problems with IT services. You can access this in the CityU Portal:

  1. Login CityU Portal: https://www.cityu.edu.hk/portal/
  2. Selected "IT and Library Services"
  3. Locate the chatbot icon in the column of "IT services"
  4. Click to talk with the chatbot
 
 

Copilot: GPT-4V with Current Data & Vision

The limitation of having knowledge up to September 2021 may discourage you from using the CityU GPT chatbot. How about an AI companion connected to the live World Wide Web? Copilot from Microsoft, also advertised as Bing Chat, has access to the Web and uses the information to answer prompts, improving the accuracy of the replies. The vision in GPT-4V enables the chatbot to analyze the uploaded image for visual question answering, optical character recognition, object detection, and more.

  1. Copilot (microsoft.com)
  2. Bing Chat

CityU staff can access Copilot with the links provided above after signing in with the account <EID>@um.cityu.edu.hk (<EID>-c@my.cityu.edu.hk for students). Since enterprise-level service is offered to staff users, conversions with Copilot are protected. The tool is available to CityU students as a free (non-enterprise) service.

Note (1): Please use Microsoft Edge to access Copilot / Bing Chat as this is the only supported browser.

Note (2): When using Copilot / Bing Chat to analyze an image, the detected human faces will be blurred for privacy, so the chatbot cannot recognize faces.

 

Some use cases of ChatGPT with vision

  1. Photo Analysis

    Prompt

    Can you analyze the attached photo, and briefly describe a possible scenario during the photo captured in around 200 words.

    Input Photo


    (Photo generated with Midjourney)

    Note: When using Copilot / Bing Chat to analyze photo, the detected human faces will be blur for privacy and so the GPT response will reply the person cannot recognized.

    GPT Response

    Sure, I can analyze the photo you sent and generate a possible scenario based on the image content. Here is what I came up with:

    The photo shows a person with their face obscured, holding a plate of cupcakes and a kitten in a cozy room. The person is dressed in formal attire, suggesting a special occasion or a preference for elegance. The kitten is animated and seems to be either yawning or vocalizing. The room is warmly lit by chandeliers and has wooden interiors and shelves with various objects. The scene evokes an atmosphere of comfort, leisure, and whimsy.

    A possible scenario during the photo captured is:

    The person is a famous author who has just finished writing their latest novel. They have been working hard for months, researching, drafting, and editing their masterpiece. They have finally sent the final manuscript to their publisher and are waiting for their feedback. To celebrate this milestone, they have decided to treat themselves to some delicious cupcakes from their favorite bakery. They have also adopted a kitten from a local shelter, who has become their loyal companion and source of inspiration. They are enjoying this moment of relaxation and indulgence in their home office, which reflects their personality and taste. They are holding the kitten in their arms, listening to its cute noises and feeling its soft fur. They are also savoring the cupcakes, which are moist, fluffy, and topped with creamy frosting. They feel a sense of accomplishment and satisfaction, as well as anticipation and excitement for their future. They smile and think that life is good.

    This is the end of my scenario. I hope you liked it. 😊

  2. Generate HTML Code with Hand Drawing Draft

    Prompt

    Can you generate html code with the draft layout in the upload photo?

    Input Photo


    (Hand drawing draft of website layout)

    GPT Response


 
 

Microsoft Copilot for Microsoft 365

You may be amazed by the capability of Microsoft Copilot in PowerPoint to generate stunning visuals and propose structured presentation content with just a few words of instruction. This demo video showcases functions in the GenAI package of Microsoft Copilot for Microsoft 365.

The cloud services are currently available for commercial customers only. The university is working closely with the Microsoft Hong Kong representative to explore options to acquire Copilot for Microsoft 365.

Demo Video
Microsoft Seminar 2 - Demo of Microsoft Copilot in PowerPoint [CityU user only, 1:00:50-1:04:50]