Ad
Short

GPT Researcher is an AI-based autonomous agent for conducting comprehensive online research on a variety of tasks. The tool, inspired by AutoGPT and the "Plan-and-Solve" prompting, seeks to improve on speed and determinism issues found in current language models, "offering a more stable performance and increased speed through parallelized agent work, as opposed to synchronous operations."

According to the team, GPT Researcher facilitates research by generating relevant research questions, aggregating data from over 20 web sources, and leveraging both GPT3.5-turbo-16k and GPT-4 to create comprehensive research reports. The project aims to address the time-consuming nature of manual research tasks and the biases that result from limited resource usage in alternative approaches such as ChatGPT + WebPlugin. It includes a web interface and provides an export function for research reports in various formats.

Video: https://github.com/assafelovic/gpt-researcher

Ad
Ad
Short

"There will be no programmers in five years," believes Emad Mostaque, founder and CEO of Stability AI. In an interview, Mostaque talked about the dominant role that generative AI systems like ChatGPT are already playing in programming. Forty-one percent of the code on GitHub is already AI-generated, he said.

Mostaque is committed to open source with his company and sees open AI as a "much better business model" than closed systems.

Ad
Ad
Short

Generative AI for videos is getting better and better: Runway has given Gen-2 an update that significantly improves the quality of the results. Gen-2 can generate short video clips from text descriptions, recently added a preview function, and has been available in the browser and smartphone app for iOS since June. The company also employs one of the AI researchers behind Stable Diffusion.

Short

Startup Inflection recently unveiled its first proprietary language model, Inflection-1, which is said to be on par with GPT 3.5, Chinchilla, and PaLM-540B.

The relatively unknown AI startup, whose CEO Mustafa Suleyman was a co-founder of DeepMind, is using the language model for Pi, its "personalized AI". According to the company, Inflection-1 has been trained on thousands of Nvidia H100 GPUs and reaches the level of GPT-3.5, LLaMA, Chinchilla or PaLM 540B in some benchmarks such as TriviaQA or MMLU. However, it cannot compete with OpenAI's GPT-4 or Google's PaLM 2-L. Pi can be tried out on Pi.ai/talk.

Ad
Ad
Short

Unity has announced two new AI products for accelerating content creation and providing real-time experiences: Unity Muse and Unity Sentis.

Muse is an AI platform that speeds up the creation of 3D applications using natural input like text prompts and sketches. Sentis enables embedding AI models in the Unity Runtime for games or applications, allowing models to run on any device. Both products are currently available via closed beta, with user feedback actively sought and integrated. Unity also launched a new marketplace for AI solutions.

Google News