Smarter and more transparent food ecosystem.

[gsymon@mstdn.social]

@ki @witchescauldron @EUCommission

That's not how it's going to work. It's not based on statistics. It's based on 'following the food':

"By monitoring the agri-food supply chain in real-time, it will:

🚫 Spot risks with improved screening accuracy
🦠 Track shipments in seconds to stop foodborne disease outbreaks
🛒 Remove non-compliant products from the market faster"

[ki@chaos.social]

@gsymon @witchescauldron @EUCommission

> "not based on statistics"

what do you think "AI-powered" implies?

[gsymon@mstdn.social]

@ki @witchescauldron @EUCommission

Good question.

I imagine that it will enable dealing with huge amounts of information.

Perhaps it will apply 'probability' in terms of where something is originating and where it's going and how it's getting there?

[soulsource@mastodon.gamedev.place]

@thibaultmol @EUCommission I wonder though how AI comes into play here. From the video it looks like a problem that would allow for an exact solution.
Is the search space so big that an analytic solution is no longer feasible?

[statsguy@mas.to]

@EUCommission AI powered? Seriously? For *safety* purposes?

Forgive me if I don't feel safer.

[katzenberger@tldr.nettime.org]

@janantos

From the article you have linked:

»At the same time, the new technology brings false alarms and added responsibilities and runs up against the country's e-state data challenges.«

Hallucinations in "AI"-powered transcripts are putting the health of patients at risk.

Also, there's already a AI Mental Health Project that helps with "AI"-induced, substantial psychological problems.

I wish you good health and good luck, should "AI" have any say in the treatments of yours.

@RachelThornSub @EUCommission

[janantos@f.cz]

@katzenberger @RachelThornSub @EUCommission yeah, but also positives exists and they are also appreciated by staff ;-)

[janantos@f.cz]

@katzenberger @RachelThornSub @EUCommission have you actualy read the article where doctors say how it is saving lives of patients with strokes and heart attacks, i have seen my x-ray where it pointed where it is problem and helped doctor to move quicker. Same my x-ray lungs image during first or second covid wave

[katzenberger@tldr.nettime.org]

@janantos

Yup, saw that. I'm glad it was helpful for you.

And just today, I read a post here by a woman who saw an AI assistant cancel her hormone therapy via an incorrect transcription, despite her MD said and prescribed the exact opposite. She was struggling to revert that, because she had waited for her appointment for a long time and was lucky her MD resolved that emergency problem in a timely fashion.

@RachelThornSub @EUCommission

[janantos@f.cz]

@katzenberger @RachelThornSub @EUCommission and thats the thing in Estonian setup, it does not make decisions, but make hints to medical staff to make decision. In Estonia AI does not make decisions. doctors do. So the case you told me, is process setup problem, not the AI. And if doctor taken this advice as granted, then we shall argue about doctors quality. I remember there has been plan in Estonia to AI have access to peoples medical records and based on its findings preemptively recommending medical examinations, that was quite interesting idea, but I think it did not went to some implementation.

[katzenberger@tldr.nettime.org]

@janantos

Well, in that case, the doctor made the exact opposite prescription, but the transcription was altered by the transcribing assistant.

If there are experienced reviewers (not just in medicine) who can invest considerable time and focus into cross-checking outputs, the worst can be prevented. But the effort that goes into such reviews was the cost factor that they wanted to eliminate, in the first place.

Its nothing short of amazing that in a domain where even the tiniest of devices is regulated to death, to prevent harm, an try-and-error approach is followed with respect to the introduction of "AI".

Safeguarding this via higher demands on already stressed MDs, and holding them accountable for catching all errors they didn't even make themselves, is not sustainable, IMHO. For me, that's way too risky.

@RachelThornSub @EUCommission