AI research

Summary Researchers explored how human research patterns could be incorporated into AI models to better predict future scientific discoveries. AI models trained on published scientific findings are already being used to find new materials or targeted therapies. In a new paper, a team of researchers from the University of Chicago and the Santa Fe Institute …

Human-aware AI could help us move beyond current scientific frontier, researchers claim Read More »

Summary 3D-LLM integrates understanding of 3D environments into large language models, taking chatbots from the two-dimensional world to the three-dimensional world. Large language models and multimodal language models can handle speech and 2D images, examples include ChatGPT, GPT-4, and Flamingo. However, these models lack a true understanding of 3D environments and physical spaces. Researchers have …

3D LLM gives chatbots a deeper understanding of physical spaces. Read More »

Summary Google Deepmind introduces RT-2, a new model for robot control that learns from robot data and general web data, and uses these two sources of knowledge to generate precise robot instructions. Since the advent of large language models, robotics has tried to leverage the capabilities of LLMs trained on web data. Their general knowledge …

Google Deepmind’s latest AI model RT-2 “can speak robot” Read More »

Researchers have discovered that it is possible to automatically construct adversarial attacks that trick major language models (LLMs) such as ChatGPT, Bard, and Claude into serving unintended and potentially harmful content. Traditional jailbreaks require significant manual effort to develop and can usually be addressed by LLM vendors. However, these automated attacks can be created in …

Researchers uncover automated jailbreak attacks on LLMs like ChatGPT or Bard Read More »

Summary VideoControlNet uses diffusion models and ControlNet for more control over AI-generated video. There are several extensions to better control image synthesis for Stable Diffusion. One of the most important is ControlNet. It allows, for example, the poses of people or the structure of a room to be extracted from an input image and used …

VideoControlNet gives you more control over AI-generated video Read More »

Summary In one way, humans and large language models seem to exhibit similar behavior: they respond particularly well to information at the beginning or end of a piece of content. Information in the middle tends to get lost. Researchers at Stanford University, the University of California, Berkeley, and Samaya AI have discovered an effect in …

Large Language Models and the Lost Middle Phenomenon Read More »

Summary STEVE-1 is a generative AI model that can perform tasks in Minecraft using text instructions. AI models that can respond to natural language instructions have become incredibly popular, but creating models that can follow instructions for complex sequential tasks remains a challenge. Researchers have now introduced STEVE-1, an AI assistant that can follow a …

STEVE-1 is a chatbot that plays Minecraft Read More »

Summary Renowned AI researcher Geoffrey Hinton made headlines when he left Google to warn the world about AI threats such as mass fake news and autonomous weapons. He kept a low profile with another thesis. Hinton argues that human-like intelligence can only be achieved, and possibly surpassed, through deep learning – a view that has …

AI researcher Geoffrey Hinton thinks AI has or will have emotions Read More »

Summary There are doubts about the authenticity of the project. Indications include the non-existent address “500 Baudrillard Drive, San Francisco, CA 94127” and the team listed by Fable Simulation, whose profile pictures are AI-generated. Also, the name of the alleged CEO “Julian B. Adler” can be read as an anagram of Baudrillard. Jean Baudrillard was …

AI-generated South Park episode may be a hoax Read More »

Summary The creative capabilities of Stable Diffusion or GPT-4 are well known. However, they lack the consistency for complex stories. SHOW-1 aims to change that. The AI ​​company Fable Studio has combined several models into a new model called SHOW-1. It is capable of generating several coherent episodes of a series. They prove that their …

AI creates a complete South Park episode: It’s SHOW-1 time! Read More »