Human Benchmark - discu.eu

Not only is AI development accelerating, the rate it is surpassing human-level performance on each new benchmark measurement model is accelerating too. https://contextual.ai/plotting-progress-in-ai/ 30 comments 13/8/2023 futurology

AI now surpasses humans in almost all performance benchmarks https://newatlas.com/technology/ai-index-report-global-impact/ 191 comments 22/4/2024 artificial

AI now surpasses humans in almost all performance benchmarks https://newatlas.com/technology/ai-index-report-global-impact/ 444 comments 20/4/2024 futurology

Human Benchmark - Quanto siete medi? https://www.reddit.com/r/italy/comments/hankii/human_benchmark_quanto_siete_medi/ 90 comments 17/6/2020 italy

[R] Agent57: Outperforming the Atari Human Benchmark https://deepmind.com/blog/article/Agent57-Outperforming-the-human-Atari-benchmark 6 comments 31/3/2020 reinforcementlearning

Detroit: Become Human Benchmark - Wine vs Windows https://www.youtube.com/watch?v=iH81xrJLTdM 62 comments 1/1/2020 linux_gaming

Human and machine aspects of typing latency and text editor benchmarks https://pavelfatin.com/typing-with-pleasure 5 comments 1/1/2016 programming

Once Human Handheld Performance Benchmark Review - Steam Deck and ROG Ally Tested https://www.techpowerup.com/review/once-human-steam-deck-fps-performance-benchmark/ 13 comments 23/8/2024 amd

Waypoint - The official Waymo blog: Waymo significantly outperforms comparable human benchmarks over 7+ million miles of rider-only driving https://waymo-blog.blogspot.com/2023/12/waymo-significantly-outperforms.html?m=1 8 comments 27/12/2023 futurology

Google AI, DeepMind And The University of Toronto Introduce DreamerV2, The First Reinforcement Learning (RL) Agent That Outperforms Humans on The Atari Benchmark https://arxiv.org/abs/2010.02193 3 comments 23/2/2021 artificial

Tabula Sapiens is a benchmark, first-draft human cell atlas of nearly 500,000 cells from 24 organs of 15 normal human subjects. https://cellxgene.cziscience.com/collections/e5f58829-1a66-40b5-a624-9046778e74f5 3 comments 10/10/2022 science

Scientists have calculated that the average person can recognize 5000 faces, giving a benchmark for comparing AI facial recognition to human performance https://www.smithsonianmag.com/smart-news/average-person-can-recognize-5000-faces-180970527/ 3 comments 1/11/2018 artificial

Google DeepMind has trained a reinforcement learning agent called AlphaDev to find better sorting routines. It has discovered small sorting algorithms from scratch that outperform previously known human benchmarks and have now been integrated into the LLVM standard C++ sort library. https://www.deepmind.com/blog/alphadev-discovers-faster-sorting-algorithms 109 comments 8/6/2023 science

LLMs saturate another hacking benchmark: "Frontier LLMs are better at cybersecurity than previously thought ... advanced LLMs could hack real-world systems at speeds far exceeding human capabilities." https://x.com/PalisadeAI/status/1866116594968973444 8 comments 9/12/2024 artificial

Humans may have emerged as a distinct species as early as 350,000 years ago, according to a new study. DNA from the skeleton of a boy who lived in what's now South Africa may be the best benchmark so far for gauging when Homo sapiens originated in Africa. https://www.sciencenews.org/article/ancient-boys-dna-pushes-back-date-earliest-humans 126 comments 28/9/2017 science