An EU draft regulation for Artificial Intelligence (AI) leaked today. After a quick scan here are my first few impressions:
- In trying to regulate AI is the EU overshooting? The definitions are set so wide and so vast that it may affect ANY type of software or SaaS.
After all: what modern software system won't include some level of "search" or "optimization"?
Will that imply that just about any software or SaaS will have to be verified against EU Compliance and regulation before it can be deployed?
How is the government going to enforce this? An army of software checkers?
Which software system does NOT contain search, optimization or statistics these days?
Screenshot from Annex 1 of the AI regulation draft document of the European Union
- No more AI in marketing: (Article 4.1).
"AI systems designed or used in a manner that manipulates human behaviour, opinions or decisions through choice architectures or other elements of user interfaces, causing a person to behave, form an opinion or take a decision to their detriment.
Also (4.2): can't use prediction to target an individual's vulnerabilities or special circumstances causing them to behave or decide to their detriment.
What will happen to Facebook and Google Ads and Adword bidding?
- In addition Article 41.1: No more Human Resources Analytics for job interviews:
"any natural person whose personal data is being processed by an emotion recognition system or a categorization system shall be notified that they are exposed to such a system'"
Most if not all HR Analytics rely on emotion expressions and interpretation. And obviously HR can't discriminate against those who refuse such a system.
And while it's not OK to create an AI marketing system... Is it OK to do predictive policing and build a killer robots 🤔??? Because 👇
- While companies can't monitor user behaviour for marketing purposes.... "This prohibition shall not apply when such practices are authorized by law and are carried out [by public authorities or on behalf of public authorities] in order to safeguard public security". (Article 4.2)
- And article 2.4 states: "This Regulation does not apply to AI systems exclusively used for the operation of weapons or other military purposes."