AGI, Agents and Regulations 😳

This was the most chaotic week in AI, if you couldn't keep up, no worries, i've got your back. We talk about Devin, AI Software Engineer; Google's Scalable Instructable Multiworld Agent (SIMA), Figure 01 by OpenAI, and finally EU’s AI Act that finally passed!

There’s a lot of talk about Agent, AGI and Regulations in this one. So hang tight.

Stuff you should know

Devin, the AI Software Engineer

Meet Devin, the world's first AI software engineer, developed by the US-based startup Cognition. Devin is designed to revolutionize the way software is developed, debugged, and deployed, offering a new level of efficiency and innovation in the tech industry. Unlike traditional coding assistants, Devin can autonomously manage the entire process of building and releasing software applications, making it a groundbreaking tool for developers and companies alike.

Stats and Bullets

• Developer: Cognition, a US-based startup.

• Capabilities: Autonomously writes, debugs, and deploys code; manages entire software development processes.

• Performance: Outperformed other AI models in the SWE-Bench coding benchmark, solving 13.86% of issues unassisted compared to 1.96% by its closest competitors.

• Real-World Application: Successfully passed engineering interviews and completed jobs on freelancing platforms like Upwork.

• Innovation Level: First of its kind to not just suggest code but also manage the creation and release of full software applications.

Business Use Cases

• Rapid Prototyping: Businesses can use Devin to quickly turn ideas into functional prototypes, significantly speeding up the innovation cycle.

• Efficiency in Development: By automating routine coding tasks, Devin allows human developers to focus on more complex and creative aspects of software development.

• Cost Reduction: Automating the coding process can significantly reduce the cost associated with software development, from initial coding to debugging and deployment.

• Scalability: Devin's ability to manage large-scale projects autonomously makes it an invaluable tool for businesses looking to scale their software solutions without proportionally increasing their development teams.

A lot of folks called this AGI, and devs went aboslute bananas, but facts are this is not AGI, and will not take your job, just yet.

SIMA by Google (Scalable Instructable Multiworld Agent)

Google DeepMind has recently introduced SIMA, the Scalable Instructable Multiworld Agent, a groundbreaking AI designed to operate within 3D virtual environments, particularly video games. Unlike traditional game AI, SIMA can follow natural language instructions to perform tasks in a variety of gaming worlds, learning from human gameplay to adapt and execute actions without needing access to game source codes or custom APIs.

Stats and Bullets

• SIMA is trained across nine different video games, including "No Man's Sky" and "Valheim."

• The AI agent has demonstrated proficiency in over 600 basic skills such as navigation, object interaction, and menu usage.

• SIMA outperforms specialized agents trained on individual games, showcasing its ability to generalize skills.

• The agent operates solely on visual input and language instructions, similar to human players.

Pretty soon, we’ll see players using this to hack in games. Which will be sad.

EU AI Act

The EU AI Act is a groundbreaking piece of legislation that aims to regulate the use of Artificial Intelligence (AI) within the European Union. It's designed to ensure that AI systems are safe, transparent, and respect fundamental rights. The Act categorizes AI applications based on their risk levels and imposes corresponding obligations on companies.

Stats and Bullets

• Unanimous endorsement by EU's 27 member states.

• Fines for non-compliance can reach up to €35 million or 7% of annual global turnover.

• High-risk AI systems face stringent regulatory requirements.

• General-purpose AI (GPAI) models have specific obligations, including transparency and risk management.

• The Act will likely be fully applicable 24 months after entry into force, with some parts effective sooner.

Notes

• Compliance: Businesses must ensure AI systems meet safety, transparency, and fundamental rights standards.

• Risk Management: Companies using high-risk AI must conduct thorough evaluations and implement robust risk management strategies.

• Innovation: The Act provides regulatory sandboxes for startups and SMEs to develop and test AI models.

• Transparency: Firms must disclose when content is AI-generated and provide summaries of training data.

Why It Matters

The EU AI Act is a significant step towards responsible AI development and use. It matters because it sets a precedent for AI regulation globally, balancing innovation with ethical considerations. The Act's comprehensive approach to AI governance will likely influence future AI policies worldwide, making it essential for businesses to understand and prepare for its implications.

Figure 01, Robots have entered the chat

Imagine a world where robots walk among us, not just as novelties but as an integral part of our daily lives and workforce. That's the vision behind Figure AI's Figure 01, the world's first commercially-viable autonomous humanoid robot. Valued at $2.6 billion and backed by tech giants like Bezos, OpenAI, and Nvidia, Figure AI is on the brink of revolutionizing how we interact with machines. Figure 01, standing at 5'6" and capable of carrying a 20kg payload, is designed to work alongside humans in various sectors, including manufacturing, logistics, and retail, promising to address labor shortages and improve workplace safety.

Stats and Bullets

• Valuation: $2.6 billion

• Height: 5'6"

• Payload Capacity: 20kg

• Weight: 60kg

• Runtime: 5 hours

• Speed: 1.2M/S

• Key Backers: Jeff Bezos, OpenAI, Nvidia

• Capabilities: Visual reasoning, language understanding, and dexterous manipulation

Business Use Cases

• Manufacturing: Automating repetitive tasks, reducing the need for human labor in dangerous environments.

• Logistics and Warehousing: Streamlining inventory management, packing, and shipping processes.

• Retail: Assisting in customer service, stocking shelves, and managing inventory.

• Healthcare and Elderly Care: Providing assistance to healthcare professionals and offering companionship and basic care to the elderly.

Why It Matters 

The development of Figure 01 marks a significant milestone in the field of robotics and artificial intelligence. By combining the dexterity and versatility of the human form with cutting-edge AI, Figure AI is not just creating a robot; it's paving the way for a future where humans and humanoid robots collaborate to achieve more. This innovation has the potential to transform industries by making them safer, more efficient, and less reliant on human labor for high-risk or mundane tasks. Moreover, the involvement of OpenAI ensures that Figure 01 can interact with humans in a natural and intuitive way, further blurring the lines between human and machine capabilities.

Top Products launched this week

• PitchBob.io - PitchBob.io: Your AI-powered entrepreneurial assistant. From idea validation to pitching and networking, it helps wantrepreneurs become entrepreneurs seamlessly.

• Tavus for Developers Beta -  Breakthrough replica and text-to-video model, now available via APIs. Users create their own replica with 2 minutes of footage and generate authentic videos in 25+ languages in minutes. No more recording or retakes. 

• Cycle 3.0 - Cycle is the fastest way for your team to capture product feedback and share customer insights - without the busywork.

• Katalist AI Storytelling Studio- Empowering everyone to craft compelling visual stories effortlessly. Enjoy full creative control without needing AI expertise. Automatic consistency ensures seamless storytelling.

• Airtrain.ai LLM Playground - A no-code LLM playground to vibe-check and compare quality, performance, and cost at once across a wide selection of open-source and proprietary LLMs: Claude, Gemini, Mistral AI models, Open AI models, Llama 2, Phi-2

• Story.com - Automate Video Editing & Eye-Catching Captions for Your Shorts in Just a Few Clicks. Let Our AI-Powered Platform Do the Rest.

• Fine - Fine's AI agents: Your tireless software developers. From analyzing business needs to generating and testing code, they handle it all, ensuring unmatched efficiency in achieving your goals.

• AI Agents by B2B Rocket-  Unlimited Digital Workforce For You 24/7 Our Intelligent AI Agents Find & Engage 1000s of Ready-To-Buy Leads on Your Behalf Outperforming humans by 5x and saving up to 90% on costs Experience a surge in sales with minimal effort!

Handpicked video

Easiest masterclass to teach you everything about AI Agents.

A Question 🙋🏻‍♂️

Working on a complete course on AI Agents.

This week in AI

• Anthropic's Claude 3 models: Haiku, Sonnet, Opus. Opus and Sonnet on claude.ai globally. Opus excels in comprehension, Haiku in speed, Sonnet in double speed. Enhanced vision, accuracy, and 200K context window. Opus twofold improvement in factual question answering.

• Google is restricting its Gemini AI chatbot from answering election-related questions in countries where voting is taking place this year, limiting users from receiving information about candidates, political parties and other elements of politics.

• Puzzle Game Innovation: Pixaverse introduces a unique puzzle game that allows players to customize their experience by switching between drawing styles and play modes, making it an inclusive and engaging option for all types of gamers

• Partnership Between Le Monde and OpenAI  : The partnership between Le Monde and OpenAI involves an agreement that allows OpenAI to leverage Le Monde's corpus to enhance the reliability of ChatGPT's answers.

Hope this was a helpful issue. That’s going to be all.