^{Welcome, AI enthusiasts}

OpenAI has just made waves in the AI community with the release of ‘o1’, a new model that "thinks" before responding — even surpassing human performance on PhD-level science questions.

But how does it compare to GPT-4o, what real-world innovations could it lead to, and how can you get access? Let’s dive in...what’s this really mean? Let’s dive in!

^{Today’s Insights}

• OpenAI unveils o1: A new 'reasoning' AI model

• Frontier: AI’s energy crisis

• White House launches AI datacenter task force

• Google’s robots master shoelace tying

Read time: under 5 minutes

^{_{OpenAI’s STRAWBERRY}}

🍓OpenAI launches new series of AI models with 'reasoning' abilities

Disclosing AI: OpenAI has officially launched 'o1' (previously known internally as Project Strawberry/Q*), its first AI model featuring advanced reasoning abilities, now available to ChatGPT Premium and Teams users.

The Details: 

• The o1 model utilizes reinforcement learning and chain-of-thought processing, allowing it to "think" before answering, similar to human problem-solving.

• It surpasses human experts on PhD-level science questions and ranks in the 89th percentile for competitive programming challenges.

• Impressively, it solved 83% of the International Mathematics Olympiad qualifying exam problems, a stark contrast to GPT-4o’s 13%.

• There are two versions: o1-preview and o1-mini, both of which have been made available to all ChatGPT Premium and Teams users at the time of this newsletter.

• However, API access comes at a much higher cost than GPT-4o, priced at $15 per 1 million input tokens and $60 per 1 million output tokens.

Why it Matters: OpenAI’s o1 model surpasses expert humans on PhD-level science questions. With its advanced reasoning ability, where it "thinks" before responding, it promises not only more accurate AI answers but also paves the way for a whole new range of real-world applications in fields like science, coding, math, and beyond.

^{_{SAM ALTMAN}}

🏛️Sam Altman talks AI energy at the White House

Disclosing AI: The International Energy Agency predicts AI energy consumption will increase tenfold from 2023 to 2026, raising concerns that current infrastructure may fall short. In response, top AI industry leaders — including OpenAI CEO Sam Altman, Nvidia CEO Jensen Huang, and Anthropic CEO Dario Amodei — recently visited the White House to emphasize the urgency of addressing this challenge as a priority for the US government.

Why is the purpose: According to CNN, representatives from Microsoft, Alphabet, and other major tech companies are urging the US government to address the looming energy strain by constructing more data centers, semiconductor plants, and power generation facilities. They also proposed exploring sustainable alternatives, such as solar power, to lessen AI’s environmental impact.

^{_{THE WHITE HOUSE}}

🏛️ White House launches AI datacenter task force

Disclosing AI: The White House is forming a new AI datacenter infrastructure task force, spearheaded by the National Security Council, National Economic Council, and the Deputy Chief of Staff's office, to maintain U.S. leadership in AI development.

The Details:

• Tech leaders from Nvidia, OpenAI, Anthropic, Google, Microsoft, and Amazon met with government officials to discuss key issues such as AI energy consumption, datacenter capacity, job creation, and strategic site selection.

• The newly established task force will focus on aligning datacenter development policies with economic, national security, and environmental objectives.

• The administration is also simplifying the permitting process for datacenters and utilizing Department of Energy resources to support AI infrastructure expansion.

• In addition, tech companies reaffirmed their commitments to achieving net-zero carbon emissions and procuring clean energy for their operations.

Why It Matters: This task force marks a significant shift in the U.S. AI strategy, moving from a focus on safety testing to actively building the infrastructure needed to secure America's AI leadership. The announcement comes on the heels of news that OpenAI and Anthropic will allow the U.S. AI Safety Institute to test new models before they are publicly released.

^{_{GOOGLE DEEPMIND}}

🤖 Google’s robots master shoelace tying

Disclosing AI: Google DeepMind has introduced two new AI systems that enhance robot dexterity, allowing machines to carry out intricate tasks requiring precise movements, such as tying shoelaces and hanging shirts.

The Details:

• The first system, ALOHA Unleashed, enables robots to execute five intricate bi-arm manipulation tasks, such as tying shoelaces, hanging shirts, and repairing other robots.

• The second system, DemoStart, employs reinforcement learning to train robots in dexterous behaviors within simulations, needing 100 times fewer demonstrations compared to traditional techniques.

• Both systems utilize diffusion methods akin to those used in image generation models to forecast robot actions from random noise.

• In simulations, robots achieved a 98% success rate and up to 97% in real-world tasks, such as cube reorientation.

Why It Matters : Although these advancements are not yet at human-level dexterity, they represent a significant step toward developing more practical robots for everyday tasks. Additionally, the application of image generation techniques to robotics demonstrates how breakthroughs in one area of AI can lead to progress across other areas of the field.

^_FEEDBACK

How would you rate today's newsletter?

Vote below to help us improve the newsletter for you.

⭐️⭐️⭐️⭐️⭐️ Nailed it

⭐️⭐️⭐️ Average

⭐️ Epic Fail

Login or Subscribe to participate in Polls.