OpenAI's "Leaked" AGI Roadmap

Welcome to this week's issue, where we explore the latest breakthroughs in artificial intelligence that are shaping our technological future. From OpenAI's ambitious AGI plans to significant regulatory developments in the EU, the AI landscape is evolving rapidly. Let's dive into the most exciting developments.

5 Stages that need to achieve for AGI Stage 1 Chatbots ( AI with Conversational language )

Stage 2 Reasoners ( Human Level Problem Solving )

Stage 3 Agents ( System that take actions )

Stage 4 Innovators ( AI that can aid in inventions )

Stage 5 Organizations ( AI that can do work of an organizations )  To gain comprehensive knowledge about AGI and its key concepts, research frontiers, ethical challenges, and future impact for both tech and non-tech learners, check out our AGI course on Udemy.So lets explore into OpneAI’s AGI roadmap.

OpenAI's Roadmap to the Future of AI

OpenAI has introduced a five-tier classification system to track its progress towards Artificial General Intelligence (AGI). This framework spans from basic conversational AI to AI capable of performing tasks equivalent to entire organizations. The current level is conversational AI, with the next stage, "Reasoners," expected soon. The higher levels envision AI systems capable of independent task execution, generating original ideas, and managing entire organizational workflows. This structured approach aims to enhance transparency and attract investment, while also raising concerns about the potential societal impacts of achieving AGI

Step 1: Develop Large Language Models

• Create AI models that can understand and generate human-like text

• Examples include OpenAI's GPT models like GPT-4

Step 2: Build Models with Commonsense Reasoning

• Develop AI systems that can reason about the world, understand cause and effect, and make logical inferences

• Go beyond just language understanding to stronger commonsense reasoning abilities

Step 3: Create Goal-Oriented AI Agents

• Build AI agents that can plan, set goals, and take actions to achieve those goals in complex environments

• Require reinforcement learning algorithms that allow AI to learn optimal behaviors through trial-and-error

Step 4: Develop Flexible, General Learning Capabilities

• Create AI systems that can learn and acquire knowledge in a more general, flexible way, similar to human learning

• Explore meta-learning techniques that allow AI to learn how to learn

Step 5: Achieve Genuine Autonomy and Self-Improvement

• Develop AI systems with genuine autonomy, self-awareness, and the ability to recursively improve themselves

• These are the hallmarks of true Artificial General Intelligence (AGI)

OpenAI's "Strawberry" Project

OpenAI, the company behind ChatGPT, is developing "Strawberry," a new AI project to significantly enhance reasoning capabilities and advance towards Artificial General Intelligence (AGI). Strawberry involves refining pre-trained models to enable autonomous web navigation and "deep research," a capability lacking in current AI. OpenAI aims to dramatically improve reasoning for long-horizon tasks requiring extensive planning and execution. Models are tested on a "deep-research" dataset assessing web browsing and information gathering with a "computer-using agent." Strawberry also evaluates AI's ability to perform software and machine learning engineering tasks. Evolved from the earlier Q* breakthrough, Strawberry signals OpenAI's progress in overcoming AI limitations and represents a crucial roadmap component for improving reasoning and achieving AGI. As this project advances, it will be crucial to monitor its implications for the future of AI in scientific discovery, software engineering, and complex problem-solving.

EU's AI Act Takes Effect

The European Union has officially published the EU's AI Act in the Official Journal, starting the clock on legal deadlines for AI developers. The law will take effect on August 1 and will be fully enforceable by mid-2026. The regulation is based on risk and assigns different responsibilities to AI developers depending on the applications and perceived risks involved. Key points include a risk-based framework categorizing AI into low-risk, high-risk, and a third tier with lighter transparency obligations for tools like AI chatbots. High-risk applications, such as biometric AI usage, law enforcement AI, and AI in employment and education, require developers to adhere to specific requirements on data quality and anti-bias measures. General-purpose AI models like OpenAI's GPT, which powers ChatGPT, have transparency mandates, and the most advanced GPAIs may be required to conduct systemic risk assessments. The law includes a phased approach with different deadlines for compliance, such as prohibiting certain AI uses six months after the law takes effect and implementing transparency requirements for GPAIs after 12 months. The EU AI Act aims to ensure AI applications are developed responsibly and safely, addressing concerns about potential risks and biases, with the EU's AI Office providing guidelines and codes of practice for developers.

Hand Picked Video

In this video, we delve into OpenAI's recently unveiled AGI roadmap and their claim of nearing level two out of five on the path to Artificial General Intelligence (AGI)

Top AI Products from this week 

• SuperGPT - SuperGPT integrates GPT with Google Sheets, boosting productivity by 10x through automating repetitive tasks. Ideal for founders, marketers, and analysts.

• Telescope - Telescope is an AI tool for anyone needing to quickly generate a targeted list of B2B leads. Trusted by thousands of companies, it is invaluable for acquiring new clients, conducting market research, and scheduling customer development calls.

• Studio by Gan.AI Recorder - Studio by Gan.AI makes it easy to create and share high-quality, personalized videos at scale. Record unlimited videos, personalise subtitles, backgrounds, thumbnails and share them instantly on AI-built, personalised landing pages that feature your branding.

• Slides.bot: AI Presentation Maker - Convert text to professional presentations in seconds with the AI presentation maker. Ideal for business, education, and creative projects, this AI PPT generator ensures high-quality and efficient presentation creation.

• Phaie AI by Creatr - Phaie AI is an open-source tool you can use to generate and fix design systems. Using our Figma plugin, you can also work in existing files to create styles using AI to detect and change different colors, typographies, strokes, border radii, and much more.

• AI image extender - Click and drag to expand beyond the edges of your image, to enlarge the background, adjust the aspect ratio, and more.

• PngMaker.io: Convert Text to PNG Quickly - PngMaker.io is a free online tool that converts text to PNG images with transparent backgrounds in seconds. Perfect for web design, marketing, and digital content.

This week in AI

• Whistleblowers Accuse OpenAI of Illegal NDAs - According to the TechCrunch article, several whistleblowers have accused OpenAI of imposing illegally restrictive non-disclosure agreements (NDAs) on its employees.The whistleblowers claim that these NDAs prevent them from reporting potential legal violations or other misconduct at the company. This raises concerns about OpenAI's transparency and accountability to the public.

• ElevenLabs Introduces AI-Generated Voices with Avatars - ElevenLabs, a text-to-speech and AI voice generation platform, has launched a new feature that allows users to generate AI voices with accompanying avatars. This feature, called "Voices with Bodies", enables users to create more lifelike and engaging AI-generated content by pairing the voice with a visual representation. ElevenLabs aims to make AI voice generation more accessible and versatile for various applications, such as video production, podcasting, and virtual assistants.

• Fine-tuning Claude 3 Haiku in Amazon Bedrock - Anthropic now allows customers to fine-tune Claude 3 Haiku, their fastest and most cost-effective model, in Amazon Bedrock. This enables customizing the model's knowledge and capabilities for specific business needs, improving performance on specialized tasks

• Synchron Integrates OpenAI's AI into BCI for Chat Feature - Synchron, a brain-computer interface (BCI) company, has integrated OpenAI's AI technology into its BCI system to create a new chat feature. The feature aims to enable users with severe paralysis to interact with the digital world hands-free using thought alone. By combining OpenAI's generative AI capabilities with Synchron's endovascular BCI approach, the company hopes to offer simplicity in learning and use, automated prompts for texting and chatting, and more natural, real-time human-AI interactions