AI discovers latest math algorithms

Good morning, AI enthusiasts. The race to attain AI that makes real scientific breakthroughs just hit a milestone — with DeepMind’s AlphaEvolve discovering latest math solutions which have eluded humans for the reason that Nineteen Sixties.

By harnessing Gemini’s language capabilities inside an evolutionary framework, this AI coding agent is not just theoretically impressive — it’s already optimizing Google’s data centers and accelerating the very systems that power it.

In today’s AI rundown:

• Google’s AlphaEvolve discovers math breakthroughs

• Anthropic set to launch latest Sonnet, Opus models

• Transform text into polished PDFs immediately

• OpenAI’s latest Safety Evaluations dashboard

• 4 latest AI tools & 4 job opportunities

LATEST DEVELOPMENTS

GOOGLE

🔬 Google’s AlphaEvolve discovers math breakthroughs

The Rundown: Google just debuted AlphaEvolve, a coding agent that harnesses Gemini and evolutionary strategies to craft algorithms for scientific and computational challenges — driving efficiency inside Google and solving historic math problems.

The small print:

• AlphaEvolve uses a mixture of Gemini models (Flash for idea generation, Pro for evaluation) to create code, which is tested by evaluators and evolved iteratively.

• The system has already made several mathematical discoveries, including finding the primary improvement on Strassen’s algorithm from 1969.

• It’s also boosting efficiency for Google, optimizing data center scheduling, improving AI training (including its own), and helping with chip design.

• When tested on 50+ open math problems, it matched SOTA solutions in 75% and discovered entirely latest, improved solutions in one other 20%.

Why it matters: Yesterday, we had OpenAI’s Jakub Pachocki saying AI has shown “significant evidence” of being able to novel insights, and today Google has taken that a step further. Math plays a task in nearly every aspect of life, and AI’s pattern and algorithmic strengths look able to uncover an entire latest world of scientific discovery.

TOGETHER WITH ENCORD

📶 World models unleash multimodal AI’s true potential

The Rundown: Encord consolidates multimodal AI data management, curation, and annotation pipelines to 1 single platform — helping teams speed up model iteration cycles through the use of an agentic AI data workflow system to organize balanced, accurately labeled datasets 10x faster.

Join the Encord ML team on May 22 for a demo-focused webinar where you’ll learn to:

• Use world models to construct agents that adapt and reason across multimodal contexts

• Discover and supervise edge-case behavior inside petabyte-scale real-world sensor data

• Create high-quality datasets powering VLAs for robotics, ADAS, and more

Register here.

ANTHROPIC

🚀 Anthropic set to launch latest Sonnet, Opus models

The Rundown: Anthropic is reportedly preparing to launch advanced versions of Claude’s Sonnet and Opus models within the “upcoming weeks,” featuring hybrid pondering and expanded tool use capabilities.

The small print:

• The models are reportedly able to alternating between reasoning and power use, and might self-correct by stepping back to look at what went fallacious.

• For coding, the models can test their generated code, ID errors, troubleshoot with reasoning, and make corrections without requiring human intervention.

• An Anthropic model, codenamed Neptune, is undergoing safety testing, with some believing the name hints at a 3.8 (eighth planet from the sun) release.

• The news coincides with Anthropic launching a brand new bug bounty program focused on testing Claude’s principles on safety measures.

Why it matters: While Anthropic has been in the combination with Google and OpenAI for the highest model within the industry, the corporate has been much slower to bring latest ones to market — with 3.7 Sonnet in February marking its only release in 2025. With each other rivals also likely releasing upgrades soon, we could possibly be in for a wild few months.

AI TRAINING

📄 Transform text into polished PDFs immediately

The Rundown: On this tutorial, you’ll learn learn how to use Grok’s latest PDF rendering feature to create professional-looking documents directly from prompts — with fast previews and editing capabilities.

Step-by-step:

1. Visit Grok out of your computer browser to access the fundamental chat.

2. Write an in depth prompt describing the document you would like (resume, literature review for a research paper, or invoices).

3. Review the preview and refine your document using follow-up prompts or by editing the LaTeX code directly through the Code button.

4. Download your finalized PDF using the download button.

Pro tip: For LaTeX research papers, remember to save lots of each the PDF and source code for future editing or journal submissions that require the unique LaTeX files!

PRESENTED BY HACKERRANK

⚙️ Custom training data for the SDLC

The Rundown: Struggling to source high-quality data on your AI models? HackerRank now delivers custom datasets designed by the experts who test thousands and thousands of human developers yearly.

With HackerRank, you’ll be able to:

• Curate a custom dataset on specific software development skills

• Access a workforce of development experts for data labelling and annotation

• Request an evaluation dataset to check your model’s performance

Request a custom dataset today.

OPENAI

🔍 OpenAI’s latest Safety Evaluations dashboard

The Rundown: OpenAI launched a brand new Safety Evaluations Hub that can publicly and repeatedly display test results for its AI models, showing how they perform on metrics like harmful content generation, hallucination rates, and jailbreak attempts.

The small print:

• The hub shows comparative performance data across OAI models, including metrics for refusing harmful content and accuracy on factual questions.

• The dashboard currently focuses on 4 categories: harmful content, jailbreak vulnerability, hallucination rates, and adherence to instruction hierarchy.

• OpenAI guarantees to update the page “periodically” as a part of what it calls a company-wide effort to speak more proactively about AI safety.

• The discharge comes after critiques that the corporate just isn’t transparent with safety testing, and following issues with a recent rollout of a GPT 4o update.

Why it matters: With labs racing to push out models to maintain pace with rivals, many consider safety has been taking a backseat to hurry. That is an important step towards more transparency, but it’s going to be counting on OpenAI to self-report and continually update the information — which likely won’t completely satisfy those calling for stricter safety measures.

QUICK HITS

🛠️ Trending AI Tools

• 🔌 Gemini Advanced – Connect Google’s advanced assistant to GitHub repos

• 🤖 GPT 4.1 – OpenAI’s advanced coding model, now available in ChatGPT

• 🤳 TikTok AI Alive – Turn static images into dynamic videos for TikTok Stories

• 🐰 CodeRabbit – AI code reviews directly in Cursor, Windsurf, and VSCode

💼 AI Job Opportunities

• 🎨 The Rundown – Designer (Brand & Platform)

• 🧪 Author – AI Researcher

• ⚙️ OpenAI – Software Engineer, Inference

• 💻 Siena – Senior Fullstack Engineer

📰 All the pieces else in AI today

OpenAI added GPT 4.1 and GPT 4.1-mini coding-focused models to ChatGPT, now available to each free and paid users.

Stability AI open-sourced Stable Audio Open Small, a text-to-audio model for generating music samples, able to running on consumer devices with no web.

Perplexity and PayPal announced a brand new partnership, allowing users to examine out with each PayPal and Venmo when making purchases on the AI platform.

Meta’s released science research, including the Open Molecules 2025 dataset, the Universal Model for Atoms, and a study on language development and AI training.

NVIDIA is securing AI chip deals within the Middle East, supplying Saudi Arabia’s Humain and the UAE after meetings with the Trump admin and other regional leaders.

Nous research launched Psyche, a brand new open, decentralized AI infrastructure that permits individuals to pool compute to coach models without massive investment costs.

Klarna CEO Sebastian Siemiatkowski revealed the fintech giant cut 40% of its workforce as a consequence of AI, but now plans to rent human agents after a success on work quality.

COMMUNITY

🎥 Join our next live workshop

Join our next workshop this Friday, May sixteenth, at 4 PM EST with Dr. Alvaro Cintas, The Rundown’s AI professor. By the tip of the workshop, you’ll confidently understand learn how to design, construct, and deploy your individual AI systems using OpenAI’s Agents SDK.

RSVP here. Not a member? Join The Rundown University on a 14-day free trial.

🤝 Share The Rundown, get rewards

We’ll at all times keep this text 100% free. To support our work, consider sharing The Rundown with your folks, and we’ll send you more free goodies.

See you soon,

Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team