DeepMind needs to think about the broader implications of its tech – report

DeepMind doesn’t fully understand the complexity of the problems it is trying to address and needs to think through the broader implications of its work, a panel in the UK has said. Google’s healthcare arm last year appointed an independent panel to assess its work, and the team of nine - including former GDS man Mike Bracken …

[DagD]

Forty Two...

once you know what the question actually is, you'll know what the answer means.

[Anonymous Coward]

Re: Forty Two...

I thought the name was derived from DeepMining, or maybe DeeplyMined.

[Anonymous Coward]

Thank you once again CSC et al for your decade-long plundering of the NHS, leaving its IT fucked for generations.

[John Smith 19]

Google want personal information the way a junkie wants heroin.

Try talking ethics with someone like Ewan McGregors character in Trainspotting.

[Daggerchild]

Re: Google want personal information the way a junkie wants heroin.

That's too simple a take - Google are absolutely addicted to finding things out, but they're not like Facebook who deliberately reverse engineer your social circle. They pretty obviously have a different core drive: Shiny Problems.

e.g. If I lie to Google, they believe me. They don't want *me*. They want to know what I *want*.

[Daggerchild]

Cause of death: Exposure.

I had no idea this latest PR wound was inflicted by the watchdog they themselves hired. Hah!

"DeepMind had failed to carry out enough public engagement" - Oh I'd say the public are pretty engaged right now.. :)

"not all the 1.6 million patients were aware their identifiable data was being used" - Eh? I thought this was anonymised data? It's almost impossible to get a grasp on what this is all about when everyone's saying it's different things.

This stands a good chance of becoming a good solid template for all the others (desperately avoiding the torchlight) to use, assuming it survives the liquid fire being poured on it, and the lynchmobs.

[Doctor Syntax]

Clinical trials

If a new drug or anything else to do with medical treatment is introduced it has to go through limited trials so that it's results can be compared with controls.

The results of this data mining are being fed back to recommend treatments but so far I haven't seen the phrase "clinical trials" being used in any articles I've read about this. isn't there a formal protocol to make a statistical comparison between the outcomes of Google's diagnosis and those of existing diagnostic methods?

If there were then surely such protocols would include issues such as informed consent.

[Destroy All Monsters]

Re: Clinical trials

isn't there a formal protocol to make a statistical comparison between the outcomes of Google's diagnosis and those of existing diagnostic methods?

I thought this basically is the statistical comparison?

[Anonymous Coward]

No

Google and Deep MInd need to be sued out of existence and their controlling directors locked up to send a message.

The application of the Law (like Tax) should not be optional.

[MrChristoph]

DeepMind doesn't understand the complexity of what they are trying to achieve? The DeepMind folks must be laughing their asses off at this. If there's anyone who knows what complexity is and how to deal with it, it's deep mind.