human-ml   20

Basic Error Rates
This page describes human error rates in a variety of contexts.

Most of the error rates are for mechanical errors. A good general figure for mechanical error rates appears to be about 0.5%.

Of course the denominator differs across studies. However only fairly simple actions are used in the denominator.

The Klemmer and Snyder study shows that much lower error rates are possible--in this case for people whose job consisted almost entirely of data entry.

The error rate for more complex logic errors is about 5%, based primarily on data on other pages, especially the program development page.
org:junk  list  links  objektbuch  data  database  error  accuracy  human-ml  machine-learning  ai  pro-rata  metrics  automation  benchmarks  marginal  nlp  language  density  writing  dataviz  meta:reading  speedometer 
7 weeks ago by nhaliday
The Bridge: 数字化 – 网络化 – 智能化: China’s Quest for an AI Revolution in Warfare
The PLA’s organizational tendencies could render it more inclined to take full advantage of the disruptive potential of artificial intelligence, without constraints due to concerns about keeping humans ‘in the loop.’ In its command culture, the PLA has tended to consolidate and centralize authorities at higher levels, remaining reluctant to delegate decision-making downward. The introduction of information technology has exacerbated the tendency of PLA commanders to micromanage subordinates through a practice known as “skip-echelon command” (越级指挥) that enables the circumvention of command bureaucracy to influence units and weapons systems at even a tactical level.[xxviii] This practice can be symptomatic of a culture of distrust and bureaucratic immaturity. The PLA has confronted and started to progress in mitigating its underlying human resource challenges, recruiting increasingly educated officers and enlisted personnel, while seeking to modernize and enhance political and ideological work aimed to ensure loyalty to the Chinese Communist Party. However, the employment of artificial intelligence could appeal to the PLA as a way to circumvent and work around those persistent issues. In the long term, the intersection of the PLA’s focus on ‘scientific’ approaches to warfare with the preference to consolidate and centralize decision-making could cause the PLA’s leadership to rely more upon artificial intelligence, rather than human judgment.
news  org:mag  org:foreign  trends  china  asia  sinosphere  war  meta:war  military  defense  strategy  current-events  ai  automation  technology  foreign-policy  realpolitik  expansionism  innovation  individualism-collectivism  values  prediction  deepgoog  games  n-factor  human-ml  alien-character  risk  ai-control 
june 2017 by nhaliday

related tags

accuracy  acemoglu  acm  acmtariat  adversarial  ai-control  ai  akrasia  algorithms  alien-character  alignment  asia  attention  automation  bare-hands  bayesian  benchmarks  best-practices  biases  bits  bonferroni  books  business  c(pp)  chart  china  clever-rats  cog-psych  commentary  complement-substitute  current-events  data-science  data  database  dataviz  deep-learning  deepgoog  defense  definite-planning  density  differential-privacy  dirty-hands  engineering  error  ethics  exocortex  expansionism  expert-experience  expert  explanation  exploratory  exposition  features  flux-stasis  foreign-policy  frequentist  futurism  games  gelman  generalization  google  gradient-descent  hard-tech  hmm  hn  homo-hetero  hsu  humanity  hypothesis-testing  individualism-collectivism  inequality  info-dynamics  information-theory  innovation  iteration-recursion  labor  language  latent-variables  learning-theory  len:short  libraries  liner-notes  links  list  local-global  lower-bounds  machine-learning  map-territory  marginal  matching  math.ds  matrix-factorization  measurement  meta:reading  meta:science  meta:war  metameta  methodology  metrics  military  miri-cfar  model-class  models  mrtz  multi  n-factor  neuro  neurons  news  nibble  nlp  no-go  objektbuch  online-learning  openai  optimization  org:bleg  org:com  org:foreign  org:junk  org:mag  org:mat  org:med  organization  oss  overflow  papers  parametric  pdf  perturbation  pinboard  poll  popsci  positivity  prediction  preprint  presentation  pro-rata  project  psychology  q-n-a  qra  quora  ranking  ratty  realpolitik  recommendations  regularizer  reinforcement  repo  research-program  research  responsibility  risk  robotics  robust  safety  scaling-tech  science  scitariat  sensitivity  signal-noise  signum  singularity  sinosphere  skunkworks  slides  software  speculation  speedometer  ssc  stagnation  stats  strategy  summary  talks  tech  technology  techtariat  the-bones  thesis  thinking  todo  tradeoffs  transportation  trends  values  video  volo-avolo  war  winner-take-all  writing  yvain  🔬 

Copy this bookmark: