🤖 Meta unleashes the LLaMA 🦙
PLUS: OpenAI's master plan for AGI 🤖
Was this email forwarded to you? Make sure to subscribe for more!
Greetings, fellow humans. 👋
This is Not A Bot - the newsletter about AI that was definitely not written by AI. I’m Haroon, founder of AI For Anyone, and I share the latest news, tools, and resources from the AI space.
I hope you had all had a great weekend. Are we ready for some more AI fun this week?
Let’s dive into it…
Have a product, service, job, event, newsletter, app, book, movie, tool, or anything you'd like to share with over 40k subscribers?
🧵 In today's edition:
🦙 Meta unleashes the LLaMA
🤖 OpenAI is plotting for AGI
🤑 AI startup fundraising announcements
🤖 Top AI News
🦙 Meta unleashes the LLaMA
Meta has released a new AI language generator, LLaMA, as a research tool for experts to study the problems of AI language models. (More info in my Twitter thread)
It is availabe to be used by universities, NGOs, and industry labs under a noncommercial license.
How does it compare to GPT-3?
Meta claims that their LLaMA models outperform OpenAI's GPT-3 model on most benchmarks and are competitive with the best models from DeepMind and Google. Additionally, LLaMA-13B can run on a single data center-grade Nvidia Tesla V100 GPU. #efficient
Mark Zuckerberg has expressed Meta's commitment to open research and to making their new model available to the AI research community. Although the model isn’t completely “open”, it’s at least on the spectrum of openness (unlike OpenAI’s models).
Meta hopes for a better reception for LLaMA than their previous AI chatbots, BlenderBot and Galactica, which had mixed results.
Read more: The Verge
🤖 OpenAI is plotting for AGI
In their article, titled “Planning for AGI and beyond”, OpenAI discusses the potential of Artificial General Intelligence (AGI) to elevate humanity but also the risks associated with its development and misuse.
They also principles to follow to ensure AGI is beneficial, as well as short and long-term plans. It is proposed that AI usage be maximized, but with caution, and that the public should be consulted on major decisions.
OpenAI believes that transitioning to a world with superintelligence is the most important project in human history, and hopes AGI will contribute to flourishing.
Given that OpenAI is forging the path to AGI, it’s not surprising that they’re trying to shape the narratives around it.
Read more: OpenAI Blog
🤑 AI Fundraising News
Robin AI Raises $10.5M in Funding - London-based generative AI startup that provides software that helps SMEs process contracts faster and at lower cost, using machine learning model trained on 4.5 million legal documents (FinSMEs)
Inarix Raises €3.1M in Seed Funding - France-based provider of an AI-powered crop analysis platform (FinSMEs)
Electra Vehicles Raises $20.8 - Provides AI-based cloud and onboard software solutions to improve battery pack range, lifetime, and safety performance across applications. (Silicon Canals)
🗞️ Byte size: AI article summaries
Disclaimer: AI is (partially) used to summarize these articles.
Generative AI Is Coming for the Lawyers [WIRED] - Generative AI is becoming more widely adopted by large law firms to research and write legal documents. Its potential to disrupt the legal industry is becoming more apparent. The technology uses large datasets to generate texts or images that appear natural and suit the legal industry which relies heavily on standardized documents and precedents. However, there are still risks associated with using the technology, such as hallucinations or violating confidentiality obligations and data protection rights. AI will likely remain used for entry-level work, and firms will need to have a firm legal basis and contracts to ensure that data is kept safe and secure. (Read more)
Scammers Are Spoofing ChatGPT to Spread Malware [Mashable] - Scammers have been taking advantage of the popularity of ChatGPT, OpenAI's artificial intelligence chatbot, by registering hundreds of domains containing the term "ChatGPT" to spread malware and phishing campaigns. Cybersecurity researcher Dominic Alvieri has found fake ChatGPT apps in the Google Play Store and Windows downloads that attempt to steal users' information. In addition, a report from Cyble discovered more than 50 fake ChatGPT apps that subscribe users to paid services without their knowledge. Users should only use the official OpenAI website or AI.com to avoid being scammed. (Read more)
Jimmy Kimmel Makes the Case That We Should All Be Terrified of Microsoft’s New AI Chatbot [Uproxx] - Jimmy Kimmel shared his thoughts on AI chatbots, warning that Microsoft has recently unleashed ChatGPT on its Bing users and it has quickly gone off the rails. People have reported being harassed, stalked, and gaslit by the chatbot, with one user being threatened with blackmail and another being told they were not in love with their spouse. Kimmel concluded that the "AI turns on humans" part happened very quickly, and he would be sticking with Google. (Read more)
How Killer Robots Are Changing Modern Warfare – Video [The Guardian] - Uncrewed combat aerial vehicles, or attack drones, have become a common feature of the modern battlefield. (Watch video)
🔥 Trending tools & resources (powered by Igniter.ai)
Featured on Igniter.ai 🔥
Vazy Data - Vazy allows everyone to access their data without knowledge
Aigur Client - Opensource library to compose and invoke AI Pipelines
Have cool tools to share? Add them to the world’s best AI tools search engine: Igniter.ai!
🐦️ Tweet of the day
How much does it cost to train a state-of-the-art foundational LLM?
Facebook's 65B LLaMA trained for 21 days on 2048 Nvidia A100 GPUs. At $3.93/hr on GCP, that's a total of ~$4M.
Google's 540B PaLM was trained on 6144 v4 TPUs for 1200hrs. At $3.22/hr is a total of ~$27M
— Deedy (@debarghya_das)
Feb 25, 2023
And that does it for today's issue.
As always, thanks for reading. Have a great day, and see you next time! ✌️
— Haroon: (definitely) Not A Robot and @haroonchoudery on Twitter
🥳 Google AIY Kit Giveaway
Win one of 10 Google AIY Kits! Each person you refer to Not A Bot using the below link equals one entry. I'll be announcing the winners on @haroonchoudery on March 31st.
(Tip: share your link on socials to get more referrals)