/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

← → days · ↑ ↓ browse · Enter similar · o open

DeepMind's AlphaDev, an AI for discovering better algorithms, devised a C++ sorting algorithm that is up to 70% faster; ISO has since added the algorithm to C++

MIT Technology Review Will Douglas Heaven

Context & Ripple Effects

This closes a three-step arc in DeepMind's applied-reasoning line. In 2022, AlphaCode showed an AI writing competitive-level programs, and months later AlphaTensor broke a 50-year-old record on matrix multiplication. AlphaDev extends the pattern from solving isolated problems to producing a routine useful enough that ISO has folded it into the official C++ standard.

That standards-body ratification is what separates this from prior demos: the algorithm isn't sitting in a paper, it ships inside the toolchain every C++ developer compiles against.

First-order effects

  • C++ developers get sorting routines up to 70% faster with zero migration effort, because the improvement arrives through the standard library rather than a new library they must adopt.
  • ISO becomes the first standards body in this coverage to ratify an AI-discovered algorithm, making DeepMind a direct contributor to core language infrastructure.

Second-order effects

  • Stewards of other languages and standard libraries now face pressure to evaluate AI-discovered implementations, or explain why their defaults are slower than what machine search can find.
  • Rival AI labs' coding agents are benchmarked against a higher bar: matching human-written code (the AlphaCode framing) is no longer the frontier — beating decades-old library code and getting it standardized is.

Third-order effects

  • The pipeline from research demo (AlphaCode) to record-breaker (AlphaTensor) to ratified infrastructure (AlphaDev) suggests AI systems moving upstream into foundational software layers, where improvements propagate through everything built on top.
  • If later efforts like AlphaEvolve finding improved solutions on roughly 20 of 67 math problems keep the streak alive, standards bodies may need formal processes for vetting and accepting machine-discovered algorithms — an institutional question no coverage here yet answers.

The trend: DeepMind's Alpha series is climbing from playing games to writing programs to discovering algorithms that standards bodies ratify into the base layer of computing.

Discussion

  • @plinz Joscha Bach on x
    I hear on average twice a week that current (or even all) AI models will never produce a solution to a complex problem for which they have not found human solutions in the training data https://twitter.com/...
  • @s_scardapane Simone Scardapane on x
    Personal take: the generated compiler code here had to be double checked and understood before being deployed because there is no correctness guarantee. Also, this is a pretty broad definition of “algorithm”. These press takes are very bad for the field. https://twitter.com/...
  • @deepmind @deepmind on x
    As society becomes more digital, it's critical to improve the code powering the world's computing. Today in @Nature, we present AlphaDev, an AI system using reinforcement learning to discover enhanced computer science algorithms. How does it work? 🧵 https://dpmd.ai/... [video]
  • @demishassabis Demis Hassabis on x
    Our latest work in @Nature today: #AlphaDev discovered a new faster sorting algorithm that we open sourced to the main C++ library for all developers to use. This is just the beginning of AI being used to find many more efficiencies in code in future https://dpmd.ai/...
  • @dj_mankowitz Daniel J. Mankowitz on x
    ✨When I joined @DeepMind, I dreamed of using AI to impact the lives of billions of people. Today in @Nature I am proud to present our system AlphaDev that has discovered enhanced fundamental algorithms in computer science! 🚀 A quick thread. 🧵
  • @andrewcurran_ Andrew Curran on x
    Deepmind on how they trained AlphaDev. AlphaDev was so good at this game that they announced in Nature today that some of the new algorithms it discovered are - in some ranges - up to 70% faster than existing ones! https://twitter.com/... [image]
  • @natureportfolio @natureportfolio on x
    A study in @Nature shows that AlphaDev, an artificial intelligence agent from @DeepMind, discovered new and improved computer sorting algorithms widely used in C++ (a common computer programming language) libraries. https://go.nature.com/3CjfIPe [image]
  • @0x00b1 Allen Goodman on x
    bully for you [image]
  • @0x00b1 Allen Goodman on x
    I'm shocked Nature published this. https://www.nature.com/...
  • @gynvael @gynvael on x
    I don't like the overhyped PR about this. The title claim is about “faster sorting algorithms discovered”, which suggests something along the lines of a new quick sort/merge sort. Instead this is about assembly-level micro optimizations. Useful? Sure. Groundbreaking? Not sure. ht…
  • @kelmoujahid Kemal El Moujahid on x
    Sorting is the first thing developers study when learning algorithms, and one of the most optimized fields of CS. AI finding better sorting algorithms is truly groundbreaking. https://twitter.com/...
  • @jasonacox Jason A Cox on x
    AI used to create a new sort algorithm that is more efficient than LLVM libc++ sort (up to 70% faster)! The model is thinking in machine language. https://www.nature.com/...
  • @oriolvinyalsml Oriol Vinyals on x
    It took a lot of Deep RL finesse (& some Transformers ❤️) to get the first change in over a decade accepted into the LLVM standard C++ library. AlphaDev, published @Nature by my brilliant colleagues! https://www.deepmind.com/...
  • @mistertechblog Lup Yuen Lee on x
    “AlphaDev uncovered new Sorting Algorithms that led to improvements in the LLVM libc++ sorting library that were up to 70% faster for shorter sequences and about 1.7% faster for sequences exceeding 250,000 elements” https://www.deepmind.com/...
  • @sundarpichai Sundar Pichai on x
    Great progress with #Alphadev from Google Deepmind! https://twitter.com/...
  • @omarsar0 Elvis on x
    Just In! DeepMind presents AlphaDev, a deep reinforcement learning agent which discovers faster sorting algorithms from scratch. The algorithms outperform previously known human benchmarks and have been integrated into the LLVM C++ library. https://www.nature.com/... [image]
  • @dj_mankowitz Daniel J. Mankowitz on x
    Really good article in @WIRED about #AlphaDev! https://www.wired.com/...
  • @gadgetlab @gadgetlab on x
    AlphaDev has made small but significant improvements to decades-old C++ algorithms. Its builders say that's just the start. https://www.wired.com/...
  • @hubertusbecker Dr. Hubertus Becker on x
    DeepMind's AI system, AlphaDev, is redefining code writing, making notable improvements to fundamental C++ algorithms. A noteworthy step towards AI mastering coding tasks across a variety of disciplines. #AI #DeepMind #AlphaDev #CodingInnovation https://www.wired.com/... [image]
  • @techreview @techreview on x
    DeepMind's run of discoveries in fundamental computer science continues. https://www.technologyreview.com/ ...