Articles by randomwalker
106

What will be left for us to work on? (normaltech.ai)

1

Up the Stack: How AI's Escape from the Commodity Trap Risks Enterprise Lock-In (normaltech.ai)

4

Did Google's AI agents build an operating system for $916? (normaltech.ai)

1

Open-world evaluations for measuring frontier AI capabilities [pdf] (cruxevals.com)

1

Towards a science of AI agent reliability (normaltech.ai)

1

When AI Builds AI – Findings from a Workshop on Automation of AI R&D [pdf] (georgetown.edu)

1

The Longitudinal Expert AI Panel: Understanding Expert Views on AI [pdf] (static1.squarespace.com)

1

Holistic Agent Leaderboard: The Missing Infrastructure for AI Agent Evaluation (arxiv.org)

5

America's AI Action Plan [pdf] (whitehouse.gov)

1

Could AI slow science? Confronting the production-progress paradox (aisnakeoil.com)

36

AI as Normal Technology (knightcolumbia.org)

1

Why an overreliance on AI-driven modelling is bad for science (nature.com)

3

Is AI progress slowing down? (aisnakeoil.com)

3

We Looked at 78 Election Deepfakes. Political Misinformation Isn't an AI Problem (knightcolumbia.org)

1

Inference Scaling FLaws: The Limits of LLM Resampling with Imperfect Verifiers (arxiv.org)

12

Is the UK's liver transplant matching algorithm biased against younger patients? (aisnakeoil.com)

1

Core-Bench: Computational Reproducibility Agent Benchmark (arxiv.org)

92

AI companies are pivoting from creating gods to building products (aisnakeoil.com)

28

AI Agents That Matter (aisnakeoil.com)

1

AI Agents That Matter (arxiv.org)

85

Scientists should use AI as a tool, not an oracle (aisnakeoil.com)

2

AI safety is not a model property (aisnakeoil.com)

1

AI safety is not a model property (aisnakeoil.com)

1

On the Societal Impact of Open Foundation Models [pdf] (stanford.edu)

1

Will AI transform law? The hype is not supported by current evidence (aisnakeoil.com)

4

Generative AI's end-run around copyright won't be resolved by the courts (aisnakeoil.com)

1

Model alignment protects against accidental harms, not intentional ones (aisnakeoil.com)

2

What the executive order means for openness in AI (aisnakeoil.com)

8

The Foundation Model Transparency Index (stanford.edu)

2

Evaluating LLMs Is a Minefield (princeton.edu)

3

Does ChatGPT have a liberal bias? (aisnakeoil.com)

2

The REFORMS checklist for ML-based science (aisnakeoil.com)

20

ML is useful for many things, but not for predicting scientific replicability (aisnakeoil.com)

3

Is GPT-4 getting worse over time? (aisnakeoil.com)

1

Generative AI companies must publish transparency reports (knightcolumbia.org)

58

Twitter showed us its algorithm – what does it tell us? (knightcolumbia.org)

4

A misleading open letter about sci-fi AI dangers ignores the real risks (aisnakeoil.substack.com)

210

OpenAI’s policies hinder reproducible research on language models (aisnakeoil.substack.com)

2

GPT-4 and professional benchmarks: the wrong answer to the wrong question (aisnakeoil.substack.com)

1

URL to bibtex converter: GPT-4 code generation, GPT-3 back end (twitter.com/random_walker)

6

Understanding Social Media Recommendation Algorithms (knightcolumbia.org)

3

Artists can now opt out of generative AI. It’s not enough (aisnakeoil.substack.com)

1

The LLaMA is out of the bag. Should we expect a tidal wave of disinformation? (aisnakeoil.substack.com)

32

Against predictive optimization (princeton.edu)

4

People keep anthropomorphizing AI. Here’s why (aisnakeoil.substack.com)

2

ChatGPT is a bullshit generator. But it can still be amazingly useful (aisnakeoil.substack.com)

3

The bait and switch behind AI risk prediction tools (aisnakeoil.substack.com)

285

The availability of text generators will force positive changes to education (aisnakeoil.substack.com)

2

Sloppy Use of Machine Learning Is Causing a ‘Reproducibility Crisis’ in Science