I'm very interested to see how they are going to differenciate an if-clause from an actualy AI algorythm...
On the other hand I do believe that there should be limits for healthcare AI. We can not discriminate against the few to protect the many here. These are in fact, live and death situations we could be talking about.
They don't need to differentiate, this is an extension of current national liability rules which handles instances where software could cause harm but had a blind spot for AI-type applications.
Until now companies could shield themselves from that liability by converting that set of if-else:s to a neural net. With this proposal the loophole should be closed.
Would that not require that even small companies get audited and thus their code inspected?
I can totally see it for the european giants but I am wondering how this thing can even be detected and also not be denied by saying "we use provider X what and thus their TOS" until the problem is eventually deported.
> For example, job seekers who can prove that an AI system for screening résumés discriminated against them can ask a court to force the AI company to grant them access to information about the system so they can identify those responsible and find out what went wrong. Armed with this information, they can sue <
I wonder whether legislation like this might have the downstream effect of finally making resumes obsolete, by essentially making them unusable. A resume is a proxy for capability, and all the information contained within only be used in a way which can be argued to be discriminatory
i.e
- company hiring for smart people
- education level as a proxy for intelligence
- train algorithm to screen out non-Ivy League applicants
- unsuccessful candidate sues for proven harms, provider loses / goes bust
5 comments
[ 3.2 ms ] story [ 22.9 ms ] threadOn the other hand I do believe that there should be limits for healthcare AI. We can not discriminate against the few to protect the many here. These are in fact, live and death situations we could be talking about.
Until now companies could shield themselves from that liability by converting that set of if-else:s to a neural net. With this proposal the loophole should be closed.
Would that not require that even small companies get audited and thus their code inspected?
I can totally see it for the european giants but I am wondering how this thing can even be detected and also not be denied by saying "we use provider X what and thus their TOS" until the problem is eventually deported.
I wonder whether legislation like this might have the downstream effect of finally making resumes obsolete, by essentially making them unusable. A resume is a proxy for capability, and all the information contained within only be used in a way which can be argued to be discriminatory
i.e
- company hiring for smart people - education level as a proxy for intelligence - train algorithm to screen out non-Ivy League applicants - unsuccessful candidate sues for proven harms, provider loses / goes bust
Sure but that's already been happening for decades, hence the "HR scrubbing personal information before presenting CVs", etc.