Totally agree with the problem statements. However, if we are to address e.g. transparency in AI, we cap some of the benefits in terms of high performance. My take would be that transparency in _data_ (i.e. understanding data set composition, balance, bias, etc. as well as its representativeness to the real world) solves most of the challenges mentioned, but without necessarily inhibiting the power of high-performing AI-systems:)

--

--

Thanks Daniel - awesome post! DG has really suffered from bureaucratic connotations (councils, box-ticking, etc.), but it's seriously valuable to manage data efficiently and effectively!

One of our raisons-d'être at Validio is enabling tracking and proactive upholding of [whichever metric is relevant to you] and in practice, implementation really has to start with a problem.

People tend not to care too much about adding additional checks and balances before lack thereof has hurt them a bit - be it suffering customer quality, bad decisions, or penalties for non-compliance...

--

--

Ingrid af Sandeberg

Ingrid af Sandeberg

Incessantly curious polynerd. Head of Marketing and Commercialisation at Validio: good quality data will save the world!