The contrast between template-driven adjudication and actual statutory analysis is pretty damning. If decision-makers were genuinely following the law's requirement for "just cause" analysis, thered be way more variance in outcomes instead of the systematic denials. The fact that Atrium templates enabled reverse-engineering outcomes basically means the system was designed for predetermined results rather tahn individualized assessment, which seems fundamentally incompatible with fairness. Wild how much automated workflows can undermine legal accountability.
re. Government of Canada [TBS] Policy on Automated Decision Making (Auto-DM):
For those wanting something deeper to look into, I present our GC Policy on 'Auto-DM'.
At some point, we will need a coordinated series of thorough ATIPs, cross-Agency, followed by a carefully-constructed legal challenge to this entire system.
We're not going to tip our hand too much [in this public post], but there are a lot of hidden gems to discuss buried in this 'cover story' pretextual & binding TBS Policy... (Note where Auto-DM is required in §8.3...)
Government of Canada [TBS] – Directive on 'Automated Decision-Making':
Relatively speaking, we are just in the early stages of this crowd-sourced investigation, but there is a similar tool called 'Chinook' used in Immigration Canada. (Which is publicly acknowledged, unlike Atrium.)
If its use [really abuse] is provable here, it makes one wonder: where else is it [not] being used?
This might be the beginning of uncovering 'algorithmic unfairness' across the country in numerous use cases...
We already have all the Supreme Court precedent we need to fix this abuse of Natural Justice – "will we have Courageous-enough Judges?" is our primary question – Justices willing to describe & dismantle this 'engineered' system in a formal Court Order‽
P.S. When reading the Chinook advisory, certain questions come to mind:
Why split the advisory into two sections? What provoked them [each/both]? Why use overtly legal language in them? Why all the repetition? And several others...
The contrast between template-driven adjudication and actual statutory analysis is pretty damning. If decision-makers were genuinely following the law's requirement for "just cause" analysis, thered be way more variance in outcomes instead of the systematic denials. The fact that Atrium templates enabled reverse-engineering outcomes basically means the system was designed for predetermined results rather tahn individualized assessment, which seems fundamentally incompatible with fairness. Wild how much automated workflows can undermine legal accountability.
re. Government of Canada [TBS] Policy on Automated Decision Making (Auto-DM):
For those wanting something deeper to look into, I present our GC Policy on 'Auto-DM'.
At some point, we will need a coordinated series of thorough ATIPs, cross-Agency, followed by a carefully-constructed legal challenge to this entire system.
We're not going to tip our hand too much [in this public post], but there are a lot of hidden gems to discuss buried in this 'cover story' pretextual & binding TBS Policy... (Note where Auto-DM is required in §8.3...)
Government of Canada [TBS] – Directive on 'Automated Decision-Making':
https://tbs-sct.canada.ca/pol/doc-eng.aspx?id=32592§ion=html
TBS – Guide to Peer Review of [AI] Automated Decision Systems:
https://canada.ca/en/government/system/digital-government/digital-government-innovations/responsible-use-ai/guide-peer-review-automated-decision-systems.html
Legal Analysis:
Federal Government’s Directive on Automated Decision-Making: Considerations and Recommendations
https://lexology.com/library/detail.aspx?g=4c96398f-0f16-4f35-90a0-ae280838a2e1
DLA Piper: How AI will change Administrative Law (GC Directive on Auto-DM)
https://dlapiper.com/en-ca/insights/publications/2023/05/how-artificial-intelligence-will-change-administrative-law-in-canada
Relatively speaking, we are just in the early stages of this crowd-sourced investigation, but there is a similar tool called 'Chinook' used in Immigration Canada. (Which is publicly acknowledged, unlike Atrium.)
If its use [really abuse] is provable here, it makes one wonder: where else is it [not] being used?
This might be the beginning of uncovering 'algorithmic unfairness' across the country in numerous use cases...
We already have all the Supreme Court precedent we need to fix this abuse of Natural Justice – "will we have Courageous-enough Judges?" is our primary question – Justices willing to describe & dismantle this 'engineered' system in a formal Court Order‽
https://canada.ca/en/immigration-refugees-citizenship/corporate/transparency/committees/cimm-feb-15-17-2022/chinook-development-implementation-decision-making.html
P.S. When reading the Chinook advisory, certain questions come to mind:
Why split the advisory into two sections? What provoked them [each/both]? Why use overtly legal language in them? Why all the repetition? And several others...