Exciting Updates on Bing Chat (Copilot) and GPT-3 Examples

Table of Contents

1. Introduction
2. BingChat's New Plugin Feature
3. Stable Audio's New Functionality
4. PaprCity Clinic's New Language Models
5. GPT3.5 Introduction
6. MyGPT: Create Your Own Chat GPT
7. Hegi: Instant Avatar Generation
8. Leonardo AI: Real-time Image Generation
9. DEEPSEEKLLM: Japanese Language Model
10. PO: AI Chatbot with Knowledge Addition
11. Visual Anagrams: Image Generation with Diffusion Models
12. AI News Highlights

Introduction

In this article, we will discuss various AI-related tools and news that have been released or become popular over the past week. We will cover topics such as BingChat's new plugin feature, Stable Audio's new functionality, PaprCity Clinic's language models, GPT3.5 introduction, MyGPT for creating personalized chat GPTS, Hegi for instant avatar generation, Leonardo AI for real-time image generation, DEEPSEEKLLM for Japanese language processing, PO for AI chatbots with knowledge addition, and Visual Anagrams for image generation using diffusion models. Additionally, we will provide highlights of recent AI news.

BingChat's New Plugin Feature

Microsoft has recently implemented a new plugin feature in BingChat, a free AI chatbot powered by GPT4. This plugin, similar to the GPT plugin implemented in the paid chatbot GPTPLUS, allows users to ask specific questions and receive prompt responses. The BingChat plugin offers six different functionalities, including recipe recommendations, travel destination suggestions, price comparisons for online shopping, restaurant recommendations, brand/product searches, and shop searches enhancements. Currently, the plugin can only be accessed through Microsoft's web browser, ED, while BingChat itself can be used on other browsers as well. Users need to create a Microsoft account and log in to access the plugin feature.

Stable Audio's New Functionality

Stable Audio, the developer of AI music generation tool Stable Diffusion, has recently introduced new features to its tool. The updated version now includes an input audio feature, which allows users to input previously created music as a reference for generating new music. Additionally, it includes a prompt library feature that automatically provides Prompts from various genres, such as rock, pop, techno, and trance, to assist users in creating music. Users can also generate URL links for the music they create and customize settings such as step size, seed prompt, and strength. Stable Audio can be accessed by pressing the "Try IT OUT for free" button on the tool's page and logging in with a Google account. The free plan allows users to create up to 45 seconds of music per month.

PaprCity Clinic's New Language Models

PaprCity Clinic has announced two new large-Scale language models, ppx7bonl and ppx70B-Online. These models can be tested for free on the demo page of PaprCity Labs. By selecting either ppx7bonl or ppx70B-Online on the demo page's tabs and submitting a question in the form, users can receive the latest information related to their query. The models have demonstrated their accuracy by providing correct information about upcoming fighting cards for a combat sports event when tested. While existing AI chatbots often generate incorrect answers to queries about Current events, these new models from PaprCity Show promise in providing reliable outputs. The models are expected to be available through OpenAI's GPT Store in early 2024.

GPT3.5 Introduction

Introducing GPT3.5, a new AI model that allows users to create their own chat GPTs using the MyGPT tool. Developed by OpenAI, this tool enables users to generate AI chatbots customized to their preferences and share them with other GPTPlus users through a link. The MyGPT tool has gained popularity, and numerous GPTs have been shared on the internet daily. Notable examples include the Blog Writing Assistant, which assists users in generating blog ideas and structuring their content, and the Visual Weather Artist, which generates city images reflecting real-time weather conditions. The Assistant for Grants is another popular GPT that provides information on grants and answers queries related to financial support. OpenAI plans to launch a GPT Store in early 2024 to facilitate the distribution and access to GPTs.

MyGPT: Create Your Own Chat GPT

The MyGPT tool from OpenAI allows users to create their own chat GPTs that can be shared with others. By logging in with a Google account on the tool's page, users can access a user-friendly interface to create customized chatbots. After setting up the bot, users can engage in a chat-Based conversation with the bot and explore its capabilities. The tool does not currently support web browsing or file upload features. However, it provides a seamless chat experience for generating conversations and debugging code, making it an efficient platform for collaborative AI development. The tool is available for free with limitations, and a paid version offers additional features and benefits.

Hegi: Instant Avatar Generation

Hegi is an AI Tool that allows users to generate their own AI avatars in just five minutes through its new Instant Avatar feature. By clicking the "Instant Avatar" button on the top of the tool's page, users can access a free version of the feature. While a paid subscription is required to access all the tool's functionalities, even free users can generate one instant avatar. Hegi stands out among similar tools as it offers high-quality AI avatar generation functionality. Users can log in to the tool using their Google account and explore the possibilities of creating personalized avatars.

Leonardo AI: Real-time Image Generation

Leonardo AI, an AI tool developed for generating high-quality images from Texts, has introduced a new feature called Live Canvas. Live Canvas allows users to generate images in real-time based on their input text. Both web browser and iOS app versions of Leonardo AI are available, with the web browser version accessible by logging in with a Google account. After accessing the tool, users can input their desired image prompts and start creating images using the canvas on the left side of the screen. The generated images are displayed in real-time on the right side of the screen. Users can right-click on the generated images to save them. While the tool currently supports English, Spanish, French, and German, support for Japanese is expected in the near future.

DEEPSEEKLLM: Japanese Language Model

DEEPSEEKLLM is an open-source Japanese language model developed by using a 670SEC parameter configuration and a 2 trillion-token dataset that includes English and Chinese languages. The DEEPSEEKLLM model, recently released by a team of researchers from the University of Michigan, aims to reproduce the characteristics of the human brain using AI chips. The source code is available on GitHub, and a demo version can be accessed on the project's page. Users can input questions and queries in Japanese and receive Relevant outputs in real-time. Although the web browsing and file upload features are not implemented, DEEPSEEKLLM demonstrates excellent performance and efficiency in Japanese language processing.

PO: AI Chatbot with Knowledge Addition

PO is an AI tool that allows users to create AI chatbots using GPT3.5 and other language models. Recently, PO introduced a new feature that enables users to add knowledge to their AI chatbots by uploading PDF files and other resources. Users can create and customize their AI chatbots by logging in with a Google account on the tool's page. After logging in, users can access the bot creation section on the top left corner of the page. By clicking on the "Add Knowledge Source" button, users can upload PDF files or input text to enhance their chatbots' knowledge. The chatbots will then utilize the added knowledge to answer queries from users. While PO initially provided free access to GPT4, it now requires a paid subscription for usage.

Visual Anagrams: Image Generation with Diffusion Models

The research team at the University of Michigan has developed an open-source AI tool called Visual Anagrams, which can generate visual illusions and images that change depending on the viewer's perspective. This tool utilizes a diffusion model that reconstructs deteriorated images step by step to create the illusion of changes in the image. The source code of Visual Anagrams is available on GitHub, and the project page includes videos demonstrating the transformation of paintings and artist images through Puzzle-like arrangements, rotations, and color inversions. The tool employs a unique approach to generate visually intriguing results, and it is expected to garner significant interest among users.

AI News Highlights

1. OpAI to Independently Procure AI Chips: OpAI has reportedly signed a contract to independently procure AI chips from RayAI for approximately 7.5 billion yen. RayAI is developing chips that aim to replicate the characteristics of the human brain and is in close proximity to OpAI's headquarters. The availability of these chips could potentially transform the capabilities of chat GPTs and contribute to their future advancements.

2. Google's Gemini Model Release Delayed: Google's highly anticipated multimodal AI model, Gemini, had its release postponed to 2024 due to certain limitations in processing non-English queries. Gemini is expected to compete with OpenAI's GPT4 in terms of advanced language processing capabilities, and its delayed release has generated Curiosity about its potential features and pricing.

3. Top 10 Uses of Bird AI: Google has released a ranking of the top 10 use cases of Bird, an AI chatbot developed by researchers. The ranking highlights the diverse applications of Bird, including fact-finding research, consultations in specialized fields, translating single or multiple languages accurately, and providing dictionary-like searches based on specific words or explanations. The usage Patterns of the Japanese version of Bird differ slightly from its English counterpart, with a higher demand for translation services.

4. AI Atsushi Tamura's Virtual Avatar: Atsushi Tamura, a member of the comedy duo "London Boots 1 and 2," introduced an AI-based virtual avatar named AI Atsushi on his YouTube Channel. The Ai Avatar, which is designed to imitate Tamura's thoughts and opinions, interacted with users through a live stream, showcasing the potential of AI advancements in creating lifelike personalities. The avatar is still in the experimental stage, but further improvements are expected to blur the lines between real and virtual.

5. Learning AI Basics through Manga: Yahoo Japan's portal site, Yahoo! Kit, has released a comic guide called "Learn AI through Manga" to educate children about AI fundamentals. The comic guide covers a variety of topics, including AI concepts, working principles of AI models, and the future of AI in an engaging and easily understandable manner. The content aims to help children and even adults gain a foundation in AI knowledge while enjoying the visual storytelling provided by manga.

In this article, we have explored various AI Tools and news that have emerged in recent times. These tools offer users remarkable capabilities, such as generating personalized avatars, creating real-time images from text, and enhancing chatbots with additional knowledge. Meanwhile, news highlights have shed light on significant developments in the AI industry, including independent AI chip procurement, delayed model releases, and the exploration of AI in entertainment and education. The future of AI holds tremendous potential, and these tools and advancements promise to revolutionize various domains and facilitate profound innovation and creativity.

Highlights

• BingChat introduces a new plugin feature for enhanced AI Chat experience.
• Stable Audio now allows users to input previously created music for reference and offers prompt library functionality.
• PaprCity Clinic releases two new language models, ppx7bonl and ppx70B-Online.
• MyGPT enables users to create personalized chat GPTs and share them with others.
• Hegi introduces Instant Avatar feature for quick avatar generation.
• Leonardo AI implements Live Canvas for real-time image generation from text.
• DEEPSEEKLLM presents a large-scale Japanese language model with high performance.
• PO allows users to add knowledge to their AI chatbots by uploading PDF files.
• Visual Anagrams offers an open-source tool for generating visually intriguing images.
• OpAI independently procures AI chips from RayAI for further advancements in chat GPTs.
• Google's multimodal AI model, Gemini, experiences a release delay to address non-English processing limitations.
• The top 10 uses of Bird AI include fact-finding, specialized consultations, and accurate translations.
• Comedian Atsushi Tamura introduces an AI-based virtual avatar for live interactions with users.
• Yahoo! Kit releases "Learn AI through Manga" to educate children about AI fundamentals.

FAQs

Q: How can I access BingChat's new plugin feature? A: To access BingChat's new plugin feature, you need to create a Microsoft account and log in to BingChat using the Microsoft Edge web browser. Once logged in, you can find the plugin option in the top right corner of the screen.

Q: Can I use Stable Audio for free? A: Yes, you can use Stable Audio for free by pressing the "Try IT OUT for free" button on the tool's page and logging in with your Google account. The free version allows you to create up to 45 seconds of music per month.

Q: Does DEEPSEEKLLM support English and Chinese languages as well? A: Yes, DEEPSEEKLLM supports English and Chinese languages in addition to Japanese. It utilizes a 2 trillion-token dataset that includes these three languages for enhanced language processing capabilities.

Q: Can I create multiple chat GPTs using MyGPT? A: Yes, you can create multiple chat GPTs using the MyGPT tool. After customizing your chatbot, you will receive a link that can be shared with others. They can then use your customized chat GPT for free.

Q: Does Hegi offer free instant avatar generation? A: Yes, with Hegi, you can generate one instant avatar for free by clicking the "Instant Avatar" button on the tool's page. However, to access all the features and benefits of Hegi, a paid subscription is required.

Q: Does Leonardo AI support languages other than English? A: Yes, Leonardo AI currently supports English, Spanish, French, and German languages. However, Japanese language support is expected to be added in the near future.

Q: Can I add knowledge to an AI chatbot using PO? A: Yes, you can add knowledge to an AI chatbot created with PO by uploading PDF files or inputting text. This allows the chatbot to utilize the added knowledge to provide more accurate and informative responses.

Q: Is Visual Anagrams available for free? A: Yes, Visual Anagrams is an open-source AI tool, and its source code is available on GitHub. You can access and use the tool for free.

Q: How can I procure AI chips for my chat GPT? A: OpenAI is offering independent procurement of AI chips for chat GPTs. You can procure these chips from RayAI, the developing company, for approximately 7.5 billion yen.

Q: When will Google's Gemini model be released? A: The release of Google's Gemini model has been delayed to 2024 due to certain limitations in processing non-English queries. The exact release date has not been announced yet.

Q: How can I learn AI through manga? A: Yahoo! Kit has released a comic guide called "Learn AI through Manga" that offers an engaging and visually appealing way to learn AI fundamentals. You can access the guide for free on Yahoo! Kit's portal site.