Inter
Measuring the predictability of life outcomes with a scientific mass collaboration https://t.co/oCR5RovCfL We do not understand the social phenomenon, ML algorithms are not enough, the theories do not explain well, and the prediction is no good way of und
RT @msalganik: If hundreds of scientists created predictive algorithms with high-quality data, how well would the best predict life outcome…
RT @drjenndowd: PS: I wish I'd heard the great talk from @msalganik @NuffieldCollege a week earlier, as "dark matter" is a great term for a…
RT @BSADigitalSoc: Despite their hype, endorsement & use, predictive models that use machine learning methods, even with robust social sci…
@roieki @Plinz Not AI but maybe of interest https://t.co/pXzkVMZxa8
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @AmyRuckes: Measuring the predictability of life outcomes with a scientific mass collaboration Salganik et al. #MachineLearning #BigDa…
Measuring the predictability of life outcomes with a scientific mass collaboration Salganik et al. #MachineLearning #BigData https://t.co/cJHX7zX2KQ
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @BSADigitalSoc: Despite their hype, endorsement & use, predictive models that use machine learning methods, even with robust social sci…
Despite their hype, endorsement & use, predictive models that use machine learning methods, even with robust social scientific data, can't predict individual life outcomes. They therefore need to be kept out of welfare & criminal justice systems.
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
People saying this shows limits of ML. Looks to me like it points towards the limits of social science, w. trickle effects for ML. After all, 750 journal articles, largely soc sci, had used same dataset; question not why ML failed but why soc scientists th
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @jrpybus: Woah! https://t.co/LoeV3K12PV
Woah!
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @NoraNiLoideain: ‘For policymakers considering using predictive models in settings such as criminal justice and child-protective service…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
Drops this on the policy making table and walks away, checking to make sure someone picked it up and read it. https://t.co/ukE771PUL3
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @NoraNiLoideain: ‘For policymakers considering using predictive models in settings such as criminal justice and child-protective service…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
.@Corydoctorow is very good on the question of whether the AI/ML project is largely an elaborate sales pitch https://t.co/ZQtStJD2vE
RT @tante: This paper is important. Experts in machine learning given great data still cannot predict life outcomes. Prediction with "AI"…
RT @henryfarrell: This interview provides useful context, especially the interesting fact that it involved a common task framework https://…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
ML has introduced new ways to walk the bias variance tradeoff which are better than many conventional methods but in reality measurement is really the only thing that matters and that's always going to be what limits predictive capabilities
RT @javisamo: El aprendizaje máquina/estadística puede predecir quién es un "mal parado", quién necesita ayuda social, qué mujer puede ser…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
We don't know how to predict life trajectories like grade point average, evictions, layoffs, etc.
“complex machine-learning methods & thousands of predictor variables were only somewhat better than a simple benchmark model that used linear or logistic regression with four predictor variables selected by a domain expert.” ML: it’s barely even stati
Important news about how difficult it is to predict human outcomes, even with high-quality data & enormous effort by 100s of data scientists. This is a strong model for how to evaluate (lack of) predictive power in real outcomes for large numbers of ac
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @henryfarrell: This interview provides useful context, especially the interesting fact that it involved a common task framework https://…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
In other words, beware of snake oil sellers claiming magic computer systems "solutions" for complex social issues. https://t.co/JfN71MnKMr
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
ai modellers *shocked* they totally unable to predict when people lose their jobs or when they are evicted etc from data about the person involved https://t.co/RrhZzHMDPp this is obviously mainly because those things have nothing to do with the person invo
The world is not isomorphic to data. Human responses to, and understanding of, the world is semantic and symbolic, not extensional or statistical
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @msalganik: If hundreds of scientists created predictive algorithms with high-quality data, how well would the best predict life outcome…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @msalganik: If hundreds of scientists created predictive algorithms with high-quality data, how well would the best predict life outcome…
RT @heresathought: This study should be required reading for anyone who develops, uses or approves the use of predictive algorithms about p…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
This study should be required reading for anyone who develops, uses or approves the use of predictive algorithms about people
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
Reading for #HUMN1001 @bernardopnunes
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
LA IA no sirve para predecir la evolución vital de personas y familias. Se temía la aparición de sesgos, se obtiene la incapacidad de predicción. https://t.co/4uMMMWDL39
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
Measuring the predictability of life outcomes with a scientific mass collaboration https://t.co/pEmkjZvkC8 via @katecrawford
RT @NoraNiLoideain: ‘For policymakers considering using predictive models in settings such as criminal justice and child-protective service…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
"- predictions were not very accurate and were only slightly better than those from a simple benchmark model....", BUT!, due to how talk about this they're lent an air of being MUCH more accurate, which is what I've been criticizing for years...
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @NoraNiLoideain: ‘For policymakers considering using predictive models in settings such as criminal justice and child-protective service…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
‘For policymakers considering using predictive models in settings such as criminal justice and child-protective services, these results raise a number of concerns.’
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
Thought this was interesting re ML limits @gbernardes_chem
aka AI system engineers stumbling on the Problem of Induction again
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…
RT @RDBinns: The one weird study the AI companies don't want you to know! Perhaps the most important ML paper from 2020 showing fundamenta…