This pseudoscience is rife with ideological bias and orthodoxy! It amounts to unnecessary propaganda and demagoguery! It is scaremongering for no good reason!
If AI algorithms and models work to the benefit of at least 80-90% of all patients to which this new technology is applied, then this new, major innovation in health care is doing a great job! Pseudoscientists usually omit this inconvenient fact!
It is well known, that medical datasets or most datasets used to train AI algorithms or models are deficient or incomplete in some respects. They are being constantly improved on! Most researchers strive to use the best datasets available to train their models. This has very little to with bias!
Only pseudoscientists have been busying themselves claiming that e.g. a lack of diversity is causing bias etc.
"... Researchers [pseudoscientists] at the University of Chicago found that pervasive algorithmic bias is infecting countless daily decisions about how patients are treated by hospitals, insurers, and other businesses. Their report points to a gaping hole in oversight that is allowing deeply flawed products to seep into care with little or no vetting, in some cases perpetuating inequitable treatment for more than a decade before being discovered. ..."
From the playbook below:
" ... —success stories—that demonstrate how bias can be mitigated, transforming flawed algorithms into tools that fight injustice. ...
But before we ever touch the data, we need to articulate the ideal target for the algorithm. That ideal target embodies our value system: what do we want the algorithm to learn? ..."
"... All of these proxy variables are distorted, biased versions of the ideal target, and similar problems—and solutions—apply. ..." This is well known for several decades, but the benefits exceed and corrections make this much less a problem!
From the abstract of the research paper below: "The disproportionate burden of COVID-19 among communities of color, together with a necessary renewed attention to racial inequalities, have lent new urgency to concerns that algorithmic decision-making can lead to unintentional discrimination against members of historically marginalized groups. ..." Sounds strongly like ideological bias has crept in!
‘Nobody is catching it’: Algorithms used in health care nationwide are rife with bias
This is the first linked research paper in the playbook above published only as a preprint in the Social Science Research Network in August 2020:
Algorithmic Discrimination and Input Accountability under the Civil Rights Acts
No comments:
Post a Comment