It's another week of explosive news. The tech world has been buzzing with the release of several new tools and technologies. Cursor, an AI-powered coding assistant for developers, Adobe Firefly, an AI tool for designers, Google's ChatGPT "Bard," Bing Image Creator, and ChatGPT Plugins have all made headlines, leaving people overwhelmed with the constant stream of new developments.
In the previous week, there was a notable announcement regarding Cursor.so. You can find more information about it here: Twitter link.
On March 20th, Cursor.so was officially launched. It is an integrated platform that combines the power of GPT-4 with the excellent code generation capabilities of Github Copilot. This free tool supports multiple programming languages and can automatically generate code snippets based on your input and requirements.
One of its key features is that it is free to use. By incorporating GPT-4, the code quality surpasses that of Github Copilot (which is also integrated within Cursor.so). Additionally, it includes ChatGPT, allowing users to engage in Q&A directly within the interface. Unlike Github Copilot, which only supports VS Code, Cursor.so can be used with various mainstream IDEs.
On March 21st, Adobe announced the release of Firefly. Firefly is an AI-powered drawing tool developed by Adobe, primarily focused on image generation and artistic typography effects. It can create images based on text prompts and offers hundreds of styles to customize the results.
You can find more information about Firefly here: AI Art Generator - Adobe Firefly.
Another significant announcement on the same day was the official beta release of Google's Bard. However, it is currently only available for users with IP addresses from the United States and the United Kingdom. According to user reports, Bard performs slightly below ChatGPT in English language tasks but still holds its ground. The main drawback is that it does not support multiple languages and cannot write code.
On March 22nd, during GTC2023, NVIDIA announced the launch of cloud services for generative AI and introduced the NVIDIA H100 NVL, optimized for LLM (large language models). NVIDIA claims that the H100 NVL is ten times faster than the HGX A100, making it the only GPU capable of effectively handling ChatGPT.
On the same day, Bing Image Creator was released. It is an online AI drawing tool developed by Microsoft, leveraging AI technology to allow users to generate images based on textual descriptions. It is an advanced version based on OpenAI's DALL-E model and has been integrated into the latest versions of Bing and Edge browsers.
You can try Bing Image Creator here: Bing Image Creator.
March 23rd marked the official launch of Microsoft Loop. Microsoft Loop is a new collaborative canvas that enables teams to collaborate across Microsoft 365 applications. It is built on Microsoft's Fluid Framework open-source platform and offers components, pages, and workspaces for seamless collaboration.
On the same day, Canva introduced ten AI capabilities to compete with Adobe Firefly.
On March 24th, a major announcement was made regarding ChatGPT Plugins. These plugins are designed specifically for language models and allow ChatGPT to access the latest information, perform calculations, and utilize third-party services. This enables third-party developers to integrate their services directly into the ChatGPT interface. The initial set of plugins includes hotel and flight booking, food delivery services, online shopping, legal knowledge, professional Q&A, and text-to-speech generation.
OpenAI's official browsing plugin is particularly powerful. It utilizes the Bing search API to provide answers based on the latest information available on the internet, along with search steps and source links. This means that ChatGPT is no longer limited to knowledge from before 2021 but has access to the latest information on the internet.
For developers interested in creating plugins for ChatGPT, there is a development guide available here: Plugin for ChatGPT Development Guide.
On March 25th, Databricks open-sourced a model called Dolly, which has similar instruction-following capabilities to ChatGPT but with only 6 billion parameters. Surprisingly, Dolly is not a newly developed model but has been in existence for two years. It utilizes an existing open-source 6 billion parameter model from EleutherAI and fine-tunes it using Alpaca's data to enable instruction-following capabilities not present in the original model.
Databricks believes that models like Dolly will help popularize LLM and make it accessible and customizable for every company, rather than being limited to a few companies that can afford it. They believe that most ML users can benefit from owning and customizing their own models for long-term service.
You can find the repository for Dolly here: Dolly Repository.
If you found this translation helpful, please consider subscribing and sharing. You can also follow me on Twitter for more information about Web3, Layer2, AI, and related topics: Twitter link.