A quick roundup of the essential news stories in the AI world that made headlines in July.
Stack Overflow Announces AI for Developers
Stack Overflow, the developer Q&A platform, showcased its latest offering, OverflowAI, at the WeAreDevelopers World Congress in Berlin.
OverflowAI will bring new capabilities to both Stack Overflow’s public platform and its SaaS product for teams. For example, they will enable Stack Overflow for Teams users to quickly curate and build knowledge bases, while AI-powered search will allow them to efficiently find the most relevant answers. The Visual Studio Code extension and Slack integration will ensure a seamless workflow, keeping developers productive within their preferred environments.
OverflowAI will also extend to Stack Overflow’s public platform, providing instant and accurate problem resolution with AI-powered search and questioning. In addition, the platform will build a discussion focused on knowledge sharing for writing prompts for generative AI tools as well as debates on technical approaches and implementation strategies.
A major concern in AI adoption by the developer community has been the trust in the accuracy of AI-generated content. This fact, coupled with the influx of AI-generated answers on the platform, led Stack Overflow moderators to prohibit their contributors from using chatbots to write answers on the platform. Stack Overflow’s recent survey of 90,000 coders found that 77% of developers are open to AI tools, but only 42% trust the accuracy of those tools. OverflowAI aims to address this issue by providing trusted and accurate solutions to problems, engaging the developer community as an integral part of building the future of generative AI.
Meta Releases the New Version of its LLM, Llama 2
Meta has released Llama 2, the next generation of its open-source large language model.
With 7 to 70 billion parameters, Llama 2 models have been trained on 2 trillion tokens, which is 40% more training data than its predecessor, Llama 1, was trained on. In addition, Meta’s fine-tuned LLMs are also optimized for dialogue use cases and outperform other open-source models on most benchmarks tested.
Llama 2 is free and available for research and commercial use.
Clipdrop Unveils Stable Doodle, a Sketch-to-Image Tool
Clipdrop, an AR startup for creators acquired by Stability AI in March 2023, has launched Stable Doodle, a sketch-to-image tool that turns simple drawings into dynamic visuals in different styles.
By combining Stable Diffusion XL, a text-to-image algorithm, with the powerful T2I-Adapter, a condition control solution developed by Tencent ARC (license), the tool brings more control to AI image generation. While image generation technology creates images from text prompts, the latter algorithm provides additional guidance to understand the outlines of sketches. The result is an output image that is created based on both a text prompt and a sketch. Additionally, Stable Doodle allows users to customize the result by choosing from 14 different styles, ranging from pixel art or origami to cinematic or neon punk.
Everypixel Launches AI-powered Presentation Design Tool
Everypixel (the company behind the Everypixel Journal) has launched Wonderslide, an AI-powered tool that allows users to effortlessly create stunning presentations from basic .pptx drafts.
The AI designer seamlessly applies the user’s content to the design layout, selecting the right color schemes, images, and icons. It also provides the best font and layout combination to work well together and places all these visual blocks in the layout according to the principles of the golden ratio.
Since its launch, Wonderslide has been featured in the top products of the day and was named #1 Product of the Week in the ‘Design Tools’ category on Product Hunt.
