• Work With AI
  • Posts
  • Meta is working on AI models that will rival OpenAI’s GPT-4

Meta is working on AI models that will rival OpenAI’s GPT-4

Sept 12, 2023

Made with Midjourney and Canva. Prompt: ‘a futuristic llama staring down a futuristic robot --ar 16:9 --v 5.2’

⚡️ Quick News Hits

📰 Top Stories This Week

(6 min read) (Source: The Wall Street Journal)

TLDR: Meta Platforms is developing an AI system that aims to rival OpenAI’s advanced models, such as ChatGPT. The new AI model aims to be exponentially more powerful than its last release, Llama 2.

  • The new system will assist other companies in creating services with advanced text generation and analytics. Meta's continues to focus on developing enterprise-level generative AI tools that mimic human-like expressions.

  • Meta plans to commence training of this latest LLM starting in early 2024. Unlike its partnership with Microsoft for Llama 2, Meta intends to use its own infrastructure for the new model. This is part of the company’s plan to become self-reliant in AI capabilities.

  • Mark Zuckerberg wants the new AI model to be open-sourced, thus providing free access for companies to develop AI-powered tools, a departure from other business models in the sector.

  • Meta’s ambitions may still fall short of rivals like OpenAI and Google. While Meta’s Llama 2 model was trained on 70 billion parameters, OpenAI’s GPT-4 is estimated to be around 1.5 trillion parameters, though experts suggest size isn't the only factor determining AI effectiveness.

(5 min read) (Source: Reuters)

TLDR: Tesla's stock rose significantly by nearly 10% following renewed interest in its AI capabilities, particularly focusing on its in-house supercomputing system, Dojo.

  • Dojo, Tesla's AI-driven computing system, is central to training its autonomous driving models. It's seen as a game-changer, potentially adding substantial value to Tesla, as it could spearhead innovations like robotaxis and advanced network services.

  • While Tesla previously relied on NVIDIA for its chips, the company is gearing towards independence by developing its own AI-centric chips. These chips would be optimized for autonomous driving tasks, promising better energy efficiency and reduced costs.

  • Beyond car manufacturing, Tesla's endeavors, such as its supercharging network and potential applications of its AI-driven chips, could diversify its revenue streams, reinforcing its market position and value proposition.

(7 min read) (Source: Business Insider)

TLDR: A recent study by researchers from Brown University and Chinese universities shows that AI bots, specifically versions of OpenAI's ChatGPT, can manage a software development process with minimal human intervention. The experiment involved a fictitious company, ChatDev, broken into stages like designing, coding, and testing.

  • In this framework, AI bots took on roles like CEO and CTO, making decisions about programming languages, identifying bugs, and moving through each stage of software development while communicating with each other.

  • On average, the ChatDev model could complete a software development task in under seven minutes at a cost of less than one dollar. The study reported an 86.66% flawless execution rate for the generated software systems.

  • The study also acknowledged limitations such as biases and errors in the AI language models, suggesting these could be potential pitfalls in fully automating the software creation process.

  • Workers across industries have used generative AI tools like ChatGPT to save time and improve productivity, including coders and Amazon employees who used the technology for software development tasks.

Big Picture: The increasing capabilities of AI in automating complex tasks like software development suggest a shift in the job market. As these technologies mature, businesses could prioritize employees skilled in managing and interpreting AI outputs over traditional software development skills, heralding a significant change in the nature of tech jobs.

👀 Interesting Reads

Deep Dives

Insightful Information

Analysis and Critiques

💰 Funding News

  1. Google invests $20M in AI Research and Policy through their new Digital Futures Project initiative aimed at supporting researchers, exploring public policy solutions, and encouraging the responsible development of AI.

    • The grants will be provided to think tanks, academic, and non-profit institutions around the world. Some of the first recipients included the Aspen Institute, Brookings Institution, Carnegie Endowment for International Peace, MIT Work of the Future, and SeedAI.

  2. Open Cosmos is a space tech company specializing in the build, launch, and operation of advanced satellites combined with AI-powered data analysis to simplify and democratize access to space and satellite data. They raised a $50M Series B led by ETF Partners, Trill Impact, and A&G.

    • Their OpenConstellation infrastructure monitors and provides data on greenhouse gasses, global temperatures, deforestation, natural disasters, polar ice caps, and sea levels amongst others.

    • Their end-to-end capabilities streamline mission complexity, reduce costs and timelines for orgs, and enable access to Earth Observation (EO) data, even for non-space entities.

  3. Two companies specializing in AI-powered real-time credit underwriting solutions for banks and other financial institutions raise huge rounds to fuel expansion. Perfios raised a $229M Series D with Zopa Bank raising $93M in a debt financing round.

  4. Pahdo Labs, a gaming company enabling players to create virtual worlds using AI tools for Anime RPG’s, raised $15M led by Andreessen Horowitz, w/ participation from Pear VC, BoxGroup, Long Journey Ventures, Neo, and Global Founders Capital.

  5. Mathpresso, an edtech company known for its LLM and AI-powered learning platform tailored for K-12 students, raised an additional $8M, bringing total funding to approximately $130M. Previous investors include Google, GGV Capital, Goodwater Capital, and SoftBank Ventures Asia.

  6. Mindtrip, an AI-native travel planning and booking platform that's simplifying travel using Large Language Models, raised a $7M Seed Round led by Costanoa Ventures.

  7. Scala Biodesign, which specializes in protein engineering through AI-driven computational methods, raised a $5.5M Seed Round led by TLV Partners.

    • Protein structure prediction is combined w/ clinical data and observations to improve the practicality of existing drugs and accelerate the engineering of therapeutic molecules, potentially revolutionizing drug development.

  8. DGTAL, a German SaaS company harnessing AI for claims audit and management, raised €3M led by 1750 Ventures.

  9. Takes, a sports experience app leveraging generative AI to curate real-time game feeds and foster sports-related conversations, raised a $1.6M Seed Round, led by Riccardo Silva, a co-owner of professional sport teams AC Milan and Miami FC.

That's all for today's email! If you want more please follow us at the social channels linked below, or check out our website!

How'd you like today's email?

Login or Subscribe to participate in polls.

Share our newsletter: If you like our work please share/forward this email with your friends, colleagues, and family. It's the best way to support us!

If this email was forwarded to you please sign up here to continue receiving them.

Want your content, product, jobs, or event featured in our newsletter? Reply to this email with the details, and our team will reach out to you.

Do you use AI for work? Tell us how, and you could be featured in our newsletter!

Check out our website for more resources, including a list of AI investors, products, events, and twitter follows.

For an archive of all our posts, click here.

We'd love to hear from you! You can always leave us comments or feedback by replying to this email!

Powered by AI. Curated and edited by Humans.