- VTEAM.AI Newsletter
- Posts
- 💠ChatGPT Customizes Responses, Llama 2 is going open-source👀
💠ChatGPT Customizes Responses, Llama 2 is going open-source👀
AND: SAP Investing in AI Startups
vTeam vMate
July 25, 2023
Hello vTeamers, and welcome to our AI fast-paced ride. Secure your seat belts as we explore:
📈Trend OpenAI's ChatGPT: Customize Responses as You Wish!
👩🔬Research Groundbreaking Research: Human Brain Cells and AI Merge with National Defense Funding!
💵Investment SAP Investing in AI Startups Cohere, Anthropic, and Aleph Alpha
💁Tips & Tricks Stable Diffusion is now in Discord and Some Great ChatGPT Cheat Sheets
👻AI Fun Weekly vTeam Memes
📖 Read time: 5 minutes
So let the journey begins. 🚀🚀🚀
For our index, we use https://arxiv.org, Google Trends, Crunchbase
TREND📈
Customizable AI Chatbot ChatGPT: Respond as You Wish!
OpenAI introduces "custom instructions" for ChatGPT, enabling personalized responses for Plus subscribers. Users can set parameters for formalness, references, and more to cater to specific needs. The feature will roll out broadly soon. However, concerns about sensitive data absorption remain due to the AI chatbot's large language model.
Llama 2 by Meta is going public
Meta has made its artificial intelligence model Llama 2 open-source, allowing startups, businesses, and developers to access and use it for creating AI chatbots. Llama 2's availability to the public aims to promote outside scrutiny and improve the safety of AI systems. This move potentially challenges the competitive advantage of tech giants like Google. However, concerns remain about self-regulation and misuse of open-source AI models. To address this, Meta has an "acceptable use" policy and a responsible use guide for developers.
Google Testing AI Tool for News Article Writing
Google is testing an AI tool called Genesis that can write news articles and has pitched it to several US news outlets as a support tool for journalists. The tool can generate news content by absorbing information on current events. It aims to assist journalists with tasks such as writing articles in unfamiliar coding languages or conducting debugging. While some view it as a useful aid, others find it unsettling and warn about potential misinformation. Google intends for the tool to be a complement to journalists' work rather than a replacement.
GitHub Introduces Copilot Chat IDE
GitHub introduces Copilot Chat, an AI-based context-aware conversational assistant for integrated development environments (IDEs) like Visual Studio and VS Code. The tool assists developers in tasks like writing code in unfamiliar languages, autocompleting boilerplate code, and debugging.
github.com
Copilot Chat's contextual awareness sets it apart from generic AI chat assistants, offering tailored guidance based on developers' specific environments. In a study, GitHub found developers achieved tasks 55% faster using Copilot, and it contributed to an average of 46% of code written across languages, increasing to 61% for Java developers.
RESEARCH 👩🔬
National defense funding secured for research on merging human brain cells with AI
Monash University-led research receives nearly $600,000 AUD in funding to grow human brain cells on silicon chips. The project aims to develop AI with continual learning capabilities, merging artificial intelligence and synthetic biology. The goal is to create programmable biological computing platforms that could outperform current silicon-based hardware. The research has potential implications in various fields, including planning, robotics, brain-machine interfaces, and drug discovery, offering Australia a strategic advantage.
New research demonstrates how Human-aware AI speeds up scientific discoveries
A recent study explores how AI can predict scientific discoveries and expand them. Researchers built models that predicted human inferences and created "alien" hypotheses not yet considered by humans, leading to advancements beyond the current scientific frontier. The "digital double" model simulated likely scientific developments, providing insights into the system's limits and the potential for radical augmented intelligence. By complementing human biases, AI can foster improved exploration and enhance collective knowledge.
Study: 59% of Organizations Lack Resources to Meet Generative AI Expectations
A recent study conducted by open-source AI solutions firm ClearML in partnership with the AI Infrastructure Alliance (AIIA) shed light on the adoption of generative AI among Fortune 1000 enterprises. The study revealed the economic impact and challenges faced by C-level executives in harnessing AI's potential within their organizations.
Budget constraints and limited resources emerged as critical barriers to successful AI adoption across enterprises. Despite challenges, most respondents believe AI is critical for creating business value, with 78% of enterprises planning to adopt generative AI as part of their AI transformation initiatives in fiscal year 2023.
INVESTMENT 💵
Runway secures $27M to enhance accessible and intelligent financial planning
San Francisco-based startup Runway raises $27.5 million in series A funding led by Initialized Capital. Runway's cloud-based platform streamlines financial planning with AI-generated insights and plans. Founded in 2020, the company aims to make financial data understandable and accessible to all team members, connecting people, data, and work.
SAP Makes Direct Investments in AI Startups Cohere, Anthropic, and Aleph Alpha
SAP (the third-largest software company in the world) has made direct investments in three AI startups: Cohere, Anthropic, and Aleph Alpha. While the company didn't disclose specific amounts, it signals SAP's commitment to offering valuable generative AI scenarios in its business applications. SAP, known for its enterprise resource planning software, is already piloting AI with customers.
TIPS & TRICKS 💁
Stable Diffusion is now in Discord. Join via the link.
Stability AI has released a new feature. Thanks to Stable Doodle, you can now generate cool art using simple sketches and drawings.
You can try it out on ClipDrop.
Malicious ChatGPT Clone WormGPT Used for Email Attacks. A hacker released WormGPT, a malicious version of OpenAI's ChatGPT, for email phishing attacks. Trained on malware data, it writes convincing human-like emails, aiding cyber attackers. SlashNext warns that AI tools like this make phishing easier for less skilled individuals.
Some great ChatGPT cheat sheets are discussed on Reddit.
AI FUN 👻
Reply