I am 100% definitely, absolutely NOT going to live tweet my exasperated reading of another document on gov.uk *but* the phrase "The UK will be unable to embrace the opportunities presented by AI” might just have been crafted to yank my chain of disagreeability.
As if “AI" is out there, doing its own thing, making all these opportunities out of the goodness of its heart.
In case you’re interested, it’s the second sentence of this: gov.uk/government/pub…
I’m specifically interested in:
- opportunities for whom?
- in which specific contexts is embracing or not embracing AI an ethical issue?
- what is the definition of “embracing an opportunity”?
Here end my 4 tweets about the first half of the 2nd sentence of that report.
- opportunities for whom?
- in which specific contexts is embracing or not embracing AI an ethical issue?
- what is the definition of “embracing an opportunity”?
Here end my 4 tweets about the first half of the 2nd sentence of that report.
It’s of particular interest because - in section 2 - none of the hypothetical ways CDEI imagine public interest data-sharing involve any kind of machine learning. 
Having skimmed the rest of the report, here’s a new exec summary: “Public interest data sharing is limited by low levels of public trust in govt’s competency and intentions; technical and perceived legal barriers are surmountable if trust can be built and useful use cases shown"
And for the 2nd sentence: “Most cases for public interest data sharing do not involve machine learning.” However, sthg not covered at all is the likely deepening of systematic inequalities likely to be caused by data-driven decisions, and the resulting further erosion of trust.
Also (promise, LAST TWEET), there is some mention of standards, PETs and attribute exchange, but no mention I can see of any genuine alternatives to data sharing.
