NL AIC Feedback to the European Commission's AI Legislative Proposal (The Netherlands)

Dear all, it is my pleasure to share feedback to the EC's artificial intelligence (AI) regulatory initiative, on behalf of the Dutch AI Coalition (NL AIC). Here's a link to the public consultation: https://ec.europa.eu/info/law/better-regulation/have-your-say/initiatives/12527-Artificial-intelligence-ethical-and-legal-requirements

NL AIC welcomes the opportunity to provide feedback to the European Commission’s Inception Impact Assessment on the “Artificial intelligence – ethical and legal requirements” legislative proposal.

1. We support the Commission’s mission to foster the development and uptake of safe and lawful AI that offers legal certainty, a favourable investment climate and an innovation optimum across the Digital Single Market, while respecting fundamental rights, ensuring inclusive societal outcomes, protecting citizen’s wellbeing and safeguarding our common Humanist moral values.

2. We believe that the EU should step up and take the lead to set global norms and standards that will shape the international Law of AI & Data system.

3. The EU should use interoperability in combination with data portability as a policy lever. AI & data driven products and services created within the EU or elsewhere in the world should abide by EU benchmarks, together with associated IEC, ISO and NEN standards, before they can obtain a CE-marking and enter the European markets.

4. AI’s dynamic and elusive nature asks for agile, flexible governance solutions. Designing a system that can quickly adapt to changing circumstances should be a key starting point.

5. In our view Option 4, ‘a combination of any of the options above taking into account the different levels of risk...’, would best serve the EC’s objectives. Since both innovation incentive & reward mechanisms, as well as safety/security risks vary per industry and per technology, policy makers should differentiate more explicitly between economic sectors when they design their digital governance solutions. We suggest a differentiated risk-based approach that contains industry specific boundary setting requirements and sector-specific AI regimes.

6. We prefer a broad definition of AI (subject matter) that includes synergies with other disruptive tech such as DLT and quantum computing. A broader scope (Option 3c) means more impact (though perhaps more initial costs/investments) and increased long term benefits.

7. To make AI and machine learning thrive, we have to re-examine the applicability and scope of (intellectual) property rights to data, and construct territorially applicable antitrust laws. Forum shopping should be avoided. There should be a right to process (e.g. access, share, analyse, re-use) data for machine learning purposes. We advise against introducing new layers of innovation stifling exclusive rights. A robust public domain, that includes open, democratized data should be promoted in general.

8. We think that the EU must also provide incentives to build and augment datasets, algorithms and inference systems, by layering traditional and alternative innovation incentive & allocation options such as prizes, subsidies, fines, benchmarks and competitions.

9. In our view, guidance is an important part of the implementation and enforcement phase of the Law of AI, as explaining its requirements encourages trust, legal certainty and freedom to operate in the data-driven economy.

10. Adjacent to regulation we can see an important role for harmonized AI Impact Assessments such as the Dutch AIIA & Code of Conduct that combines technical, legal and ethical standards, HLEG’s ALTAI and CoE’s Recommendations. Self-regulation alone should never be enough: industries simply do not have the same incentives to promote public good as governments do.

11. Synchronous to a coordinated, differentiated industry-specific approach regarding incentives and risks, the EU should actively shape technology for good and embed norms, standards, principles and values into the architecture of our technology, by means of Trustworthy AI by Design.

12. Lastly, NL AIC believes it is crucial for the EU to work together with countries that share our European digital DNA, based on common interests and mutual values. It is essential to incentivise systematic transatlantic cooperation. Sovereignty will ensure strong partnerships amongst equals.

Comments

Submitted by Juan LLORET on Sat, 12/09/2020 - 13:50

Hi, Mauritz, please:

 

There is something that seems very worrying: in addition to the law we must (together) promote the birth of electronics associated with the laws... Otherwise we can fall into the trap of, for example, regulating the use of weapons and not having the ability to manufacture them to defend ourselves if necessary.

In other words, we must promote, in addition to laws, realities that allow chips to pick up the laws. (This explained before, must necessarily knock on the door of the Alliance for Artificial Intelligence or we run the risk of being left in unnecessary expenses for European citizens and losing an important opportunity).

Thank you for sharing the information, it is important that everyone can read it and know the free and interest-free work that all actors in Europe are doing, I think.

Personally, I will read this report out of pure citizen responsibility, first. Thank for you time.

Regards,