Welcome to this week's edition of Jumble, your go-to source for the latest in AI. This issue covers the new DeepSeek model trying to takeover OpenAI. It’s a packed week in AI and here’s what you need to know ⬇️

Introducing DeepSeek-R1🔥

Chinese AI lab DeepSeek, who brought us DeepSeek-V3, has released another powerful model: DeepSeek-R1. This reasoning-focused large language model is as good as OpenAI’s o1 in math, coding, and general knowledge, but 90-95% cheaper, so advanced AI is now available to the masses, and it won’t break the bank. 

What Is DeepSeek-R1?
DeepSeek-R1 is a state-of-the-art reasoning model for problem solving and analytical tasks. It comes in two versions: DeepSeek-R1-Zero, trained entirely through reinforcement learning (RL), and DeepSeek-R1, which adds a cold-start phase and multi-stage RL for better reasoning.

DeepSeek-R1 Vs OpenAI-O1 – Who Wins? 

 DeepSeek-R1 Shines Across Benchmarks:

• Mathematics: 79.8% (Pass@1) in AIME 2024 and 93% in MATH-500.

• Coding: Ranked in the 96.3rd percentile on Codeforces.

• General Knowledge: Scored 90.8% on MMLU and 71.5% on GPQA Diamond.

• Writing: Achieved 87.6% on AlpacaEval 2.0 for question answering.

These results showcase its competitive performance alongside industry leaders like OpenAI and Meta. While it’s not entirely beating o1 across the board, it does get very close and even surpasses it in several categories. 

How DeepSeek-R1 Can Make Life a Touch Easier
DeepSeek-R1 is great for:

• Education: Advanced tutoring for complex reasoning and math.

• Software Development: Code generation and debugging.

• Research: Long-context understanding and data analysis.

Game-Changing AI That’s Actually Free
In terms of open-source models, DeepSeek-R1 sets a new bar for performance, versatility, and price. It’s a big deal, making AI available for innovators, researchers, and educators.

Want to sponsor Jumble?

Trump Redefines AI Policy: More Speed, Less Safety 🫨

On his first day in office, President Donald Trump repealed Biden’s 2023 executive order on AI safety. This order had required AI developers to report safety test results for systems posing national security, economic, public health, or safety risks. Trump's repeal reflects his administration's focus on fostering AI innovation.

The nullification of Biden's order has sparked concerns among experts about potential regulatory uncertainties and the risks of reduced oversight. However, supporters argue that easing these constraints will boost technological advancement and maintain America's competitive edge in the global AI race. The move has also highlighted ongoing debates about balancing innovation with security and public safety in the rapidly evolving AI landscape.

This Week’s Scoop 🍦

🎮 AI initiatives spark unease among video game creators

🏂 X Games introduces AI for more accurate halfpipe scoring

👩‍⚖️ Germany’s GEMA sues gen-AI company SUNO for copyright infringement

💻 Is this the year coders are replaced with AI?

🌽 Iowa's Landus Cooperative cuts workforce by 10% to embrace AI technology

🤖 New AI at Goldman Sachs aims to replicate human interaction

🏆 Oscar hopes dim for 'The Brutalist' amid AI usage controversy

AI Challenge of the Week: To Be or Not To Be 📜

#1: 

"Dost thou not see, fair soul, how time’s cruel scythe doth carve the world in endless change? Each hour fades as a petal from the rose, and yet, within decay, springs beauty anew. Forsooth, tis not the end we fear, but the silence where dreams do falter."

#2:

"Be not afraid of greatness: some are born great, some achieve greatness, and some have greatness thrust upon them. Thy fates open their arms to those who seize bold opportunity, though fortune’s fickle hand may oft play the jester. Fear not the shadows, for they are but the heralds of dawn."

Respond to this email with your answer! 💌

Thank you for being a valued reader of Jumble! See you next week for more updates on the latest trends and developments in AI.

Stay informed, stay curious, and stay ahead with Jumble!

Zoe from Jumble