Add Series Four: Gods of AI (50 cards) - #15
Merged
Merged
Conversation
The architects of machine learning: the people who worked out backpropagation, attention and reinforcement learning, and the ones who wrote the frameworks, kernels and quantisers the field actually runs on. Because this is an open source card series, the set is weighted toward people who gave the work away, and the stat block says so out loud: research, systems, openness, influence. Karpathy scores 99 on openness; Dario Amodei scores 74, and his card says why. That is the editorial line of the whole deck. Nobody appears in more than one series. Likeness rule, same as Series Three ----------------------------------- Portraits are generated image-to-image from an identified photograph. A card with no verified photograph gets a plate saying so — never an invented face. 37 of 50 have one. GitHub profile photographs are a strong source for this set specifically: the account is the person's own and the API reports their real name, which is checked against the roster before anything is accepted. Not freely licensed, so recorded as `supplied` with the profile as the source. Three failures caught by eye, not by code ----------------------------------------- Filename matching on Commons accepted a blue wolf statue outside a brewery for "Thomas Wolf" and a Bulgarian basketball player for "Georgi Gerganov". Both were replaced from their own GitHub accounts. Worse, Wikidata's photo for François Chollet is a body shot with his head cropped out of frame, and his GitHub avatar is a felt sculpture — conditioning on either produced a confident, completely invented face. He now has no portrait, which is the correct outcome. Every rejection is recorded in scripts/gods-refs.mjs with its reason so a rerun cannot silently undo it. Still faceless (13): Andrew Barto, Ashish Vaswani, Noam Shazeer, Alex Krizhevsky, Albert Gu, Oriol Vinyals, John Jumper, Rajat Monga, François Chollet, Guillaume Lample, Robin Rombach, Chris Olah, Paul Christiano. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_014htrUgbnoKQt9iT4G7hUpy
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| <> | ||
| <script | ||
| type="application/ld+json" | ||
| dangerouslySetInnerHTML={{ __html: JSON.stringify(jsonLd) }} |
ThreatCrush Security Scan5 finding(s) MEDIUM: 5
Snippets are redacted; ThreatCrush never prints matched credential material. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A fourth collection at
/gods-of-ai: the architects of machine learning — the people who worked out backpropagation, attention and reinforcement learning, and the ones who wrote the frameworks, kernels and quantisers the field actually runs on.The editorial line
Because this is an open source card series, the set is weighted toward people who gave the work away, and the stat block says so out loud:
research,systems,openness,influence. Andrej Karpathy scores 99 on openness; Dario Amodei scores 74, and his card says why. A merely good open contribution outranks a brilliant closed one.Roughly: 8 foundations (Hinton, LeCun, Bengio, Schmidhuber, Sutton, Barto, Pearl, Fei-Fei Li), the Transformer authors, the DeepMind line, and a deliberately large block of open-source infrastructure — PyTorch, TensorFlow, Keras, XGBoost/TVM, Caffe/ONNX, transformers, llama.cpp, bitsandbytes, llamafile, EleutherAI, LAION, Mistral, Stable Diffusion.
Nobody appears in more than one series.
Likeness rule, same as Series Three
Portraits are generated image-to-image from an identified photograph. A card with no verified photograph gets a plate saying so, never an invented face. 37 of 50 have one.
GitHub profile photographs turn out to be a strong source for this set specifically: the account is the person's own and the API reports their real name, which is checked against the roster before anything is accepted. Not freely licensed, so recorded as
suppliedwith the profile as the source.Three failures caught by eye, not by code
The first two were replaced from their own GitHub accounts. The third is worse and more instructive: conditioning on a reference with no visible face produced a confident, completely invented portrait — the exact failure this pipeline exists to prevent. He now has no face, which is the correct outcome.
Every rejection is recorded in
scripts/gods-refs.mjswith its reason, so a rerun cannot silently undo the judgement.Verified
tsc --noEmitandpnpm buildclean; 50 card fronts, 50 card pages, 50 OG imagespnpm gods*scripts wiredStill faceless (13)
Andrew Barto, Ashish Vaswani, Noam Shazeer, Alex Krizhevsky, Albert Gu, Oriol Vinyals, John Jumper, Rajat Monga, François Chollet, Guillaume Lample, Robin Rombach, Chris Olah, Paul Christiano.
🤖 Generated with Claude Code
https://claude.ai/code/session_014htrUgbnoKQt9iT4G7hUpy