Bb5a31f81ClaimThe technology company Transfer gathers laboratory sensor and video data of human researchers to train robotics systems and smart-glasses guidance overlays for wet-lab proceduresAgreeDisagree
Bb5a31f81ClaimMathematicians utilized Harmonic's Aristotle artificial intelligence model to solve unsolved Erdős problemsAgreeDisagree
Bb5a31f81ClaimOpenAI developed the Lifeval benchmark to evaluate artificial intelligence agents on practical everyday tasks such as finding childcare, navigating government services, and budgetingAgreeDisagree
Bb5a31f81ClaimOn OpenAI's Lifeval benchmark, task success rates rose from 49.3% on GPT-5.2 to 80.6% on GPT-6 AstraAgreeDisagree
Bb5a31f81ClaimThe SemiAnalysis GDPX benchmark evaluates artificial intelligence models on economically valuable tasks such as chip design, finance, and biologyAgreeDisagree
Bb5a31f81ClaimOn the SemiAnalysis GDPX benchmark, Gemini 4 Carbon achieved a 10.9% score at an operational cost of $16 per taskAgreeDisagree
Bb5a31f81ClaimOn the SemiAnalysis GDPX benchmark, Opus 5.5 achieved a 12% score at an operational cost of $51 per taskAgreeDisagree
Bb5a31f81ClaimRichard Socher's concept of a Eureka Machine envisions AI systems that autonomously generate ideas, run experiments, and validate discoveries across scientific disciplinesAgreeDisagree
Bb5a31f81ClaimParallel Bio uses artificial intelligence computer vision to track individual cell behaviors as experimental drugs interact with human organoidsAgreeDisagree
Bb5a31f81ClaimRecursive applied its automated research system to Andrej Karpathy's NanoChat benchmarkAgreeDisagree
Bb5a31f81ClaimThe nanoGPT Speedrun benchmark measures how quickly an artificial intelligence training process reaches a fixed validation loss targetAgreeDisagree
Bb5a31f81ClaimNvidia developed the Sol-Exec Bench benchmark to evaluate automated CUDA kernel optimizationAgreeDisagree
Bb5a31f81ClaimThe NanoChat benchmark evaluates an artificial intelligence model's ability to achieve minimal bits per byte within a five-minute hardware budgetAgreeDisagree
Bb5a31f81ClaimArca guides artificial intelligence model selection using proprietary internal benchmarks rather than relying on external public benchmarksAgreeDisagree
Bb5a31f81ClaimGPT-6 showed little performance improvement on existing financial benchmarks due to saturated evaluation tasks, ambiguous prompts, and erroneous answer keysAgreeDisagree
Bb5a31f81ClaimOpenAI discontinued tracking the GDP eval benchmark because the benchmark became saturated and flawedAgreeDisagree
Bb5a31f81ClaimArtificial intelligence models struggle more with subjective human taste and formatting nuance than with objective factual accuracyAgreeDisagree
Bb5a31f81ClaimArtificial intelligence models exhibit lower rates of factual hallucination than observed during 2022 and 2023AgreeDisagree
Bb5a31f81ClaimArtificial Analysis removed long-context reasoning benchmark AA-LCR from the Intelligence Index Version 5AgreeDisagree
Bb5a31f81ClaimGoogle's Gemini 4 Carbon gained 8 points on Terminal Bench 4.0 and 6 points on Frontier Suite compared to Gemini 4 ArgonAgreeDisagree