Welcome to Nextomoro -your go-to source for the top news and headlines of the week in AI. We'll dive into the top news and developments in AI, examining the most significant breakthroughs and their potential impact on industries and society. Join us as we explore the latest trends, exciting breakthroughs, and cutting-edge applications of AI taking the center stage every week.
Large language models are experiencing a similar moment to Stable Diffusion.
The emergence of Stable Diffusion as an open-source image generation model in August 2022 has sparked a new wave of interest in generative AI. The model allows people to generate images from the text on their own hardware, and developers to experiment with its inner workings. This has resulted in an explosion of innovation in the field, with ControlNet now surpassing other models such as Midjourney and DALL-E in terms of capabilities. This "Stable Diffusion moment" has kick-started the new wave of interest in generative AI, which was further accelerated by the release of ChatGPT at the end of November.
Other Large Language Models (LLMs), specifically Facebook's LLaMa, were recently released into the wild and are being rapidly innovated upon and improved by the open source community. After some ingenious hacking by a Bulgarian deverloper, it is now possible to run LLaMa on a MacBook Pro, a feat that was considered unrealistic just months ago. Imagine a world where anyone can freely run their own local version of ChatGPT without any censorship, control, or restrictions. This idea is fascinating and has significant implications for the future of humanity.
Grammarly introduces AI-powered assistant, GrammarlyGO
Grammarly has launched its new AI-powered assistant called GrammarlyGO, which uses generative AI to offer contextually aware communication assistance to individuals and businesses. The assistant can help users with writing from scratch or revising an existing piece, making suggestions to improve clarity and effectiveness. With its on-demand features, GrammarlyGO aims to make writing tasks more efficient and productive. The new AI-powered assistant is expected to enhance the user experience by providing more personalized and relevant suggestions.
Discord Levels Up with AI-Powered Chatbot - Ready to Revolutionize Conversations
Discord is set to release an advanced chatbot, powered by OpenAI technology, next week. The social media platform's current robot assistant, Clyde, will be updated with AI capabilities, allowing users to have extended conversations and ask questions. OpenAI is the company behind the popular ChatGPT AI language model. Clyde currently provides assistance on the platform, responding to basic slash commands and sending messages if an error has been detected.
Microsoft's Multi-Modal GPT-4 with Video Launching Soon
Microsoft's announcement of the upcoming GPT-4 model has caused a stir in the AI community as it will support multi-modal features including video, which marks a significant development in the technology's capabilities. The company's CTO and Lead Data & AI STU, Andreas Braun, made the announcement at the KI im Fokus event in Germany. While the specifics of the model are yet to be revealed, this news has raised hopes of exciting new applications for GPT-4 in various industries.
Azure's OpenAI Service Integrates ChatGPT.
Microsoft has announced the official integration of ChatGPT into Azure OpenAI Service, allowing its customers to apply advanced AI models to their business needs. The service will cost $0.002/1k tokens, and the billing for all ChatGPT usage will start from March 13. The service is currently limited and requires special access, but it joins other AI models, including Dall-E 2, GPT-3.5, and Codex. Microsoft believes that the addition of ChatGPT to the service will offer a wide range of possibilities for customers who want to integrate AI into their businesses.
AI Shows a Remarkable Ability to Accurately Interpret Human Thoughts
According to a recent study by Japanese researchers, AI has the potential to read human minds accurately. The researchers developed a new method that uses popular technology in generating images through linguistic prompts, like Stable Diffusion, which has gained popularity recently. This technology can create realistic images with just a few carefully selected words and can be used to produce static images or animations in popular styles such as anime. The study suggests that this technology could be used to decode human thoughts, offering possibilities for new applications in fields like healthcare, marketing, and more.
Wyoming Education Starts to Tackle Artificial Intelligence
The University of Wyoming has established an artificial intelligence working group to examine how the technology will impact higher education. The working group recently submitted recommendations to the university president, including prohibiting the "unpermitted use" of AI on campus. While the university is actively discussing the use of AI, the technology is only starting to gain attention in Wyoming's broader education system. School administrators and boards are beginning to learn about AI and its potential impact on student learning and teaching.
SoftBank and Other Japanese Companies Set Limits on ChatGPT Utilization.
Several Japanese companies, including SoftBank and Hitachi, have started to limit the use of interactive artificial intelligence (AI) services, such as ChatGPT, due to concerns about information leaks and other risks. In particular, SoftBank issued a warning to its employees last month, advising them not to enter confidential data or company-identifiable information while using ChatGPT and other business applications. Although guidelines for the use of cloud services and interactive AI were already in place, the rules were reiterated in response to the recent attention given to chatbots. SoftBank plans to establish more specific regulations on the use of AI, including which operations can utilize the technology and what applications can be used.
Should Teachers be afraid of ChatGPT?
In the midst of growing concerns about the use of ChatGPT in various aspects of society, including education, a group of experts proposes a solution: bring it into the classroom. According to them, teachers have little to fear from this technology, and it presents a unique opportunity for both educators and students to build knowledge together. Through the use of ChatGPT in the classroom, both the fascination with the technology and the desire to use it for nefarious purposes may be dampened. Instead, it can be harnessed to aid in the learning process and promote collaboration between students and educators.
Government to Establish UK as a Science Superpower with a £360m Bid
The UK government is set to invest £360 million in a bid to become a science 'superpower' by 2030. The investment will focus on three key areas: artificial intelligence, nuclear fusion, and genetics. The aim is to ensure that researchers have access to the best physical and digital infrastructure for research and development and to attract the best talent to the country.
The CEO of Stable Diffusion proposes that AI is observing alternative realities
"LLMS DON'T HALLUCINATE. THEY'RE JUST WINDOWS INTO ALTERNATE REALITIES IN THE LATENT SPACE." - Emad Mostaque, CEO of Stability.ai - Twitter
The statement "LLMs don't hallucinate" implies that language models do not generate completely random or fictitious outputs that are not based on any data or context. Rather, the outputs of an LLM are generated based on the patterns and relationships it has learned from the vast amounts of data it has been trained on. The concept of an "alternate reality in the latent space" refers to the idea that LLMs generate outputs based on patterns that exist in the data, but these patterns may not always align with human expectations or conventional reasoning.
