There are several initiatives currently running at the UvA in the field of AI [link], both in the development of artificial intelligence itself and in humanities and social science research into AI.
In the context of this, a recent opinion piece in Het Parool discusses not only the Dreamslab collaboration between the UvA, the Vrije Universiteit Amsterdam (VU) and Huawei Finland, but also the UvA’s Research Priority Area ‘Human(e) AI’. It is important to emphasise that the researchers at Human(e) AI have no affiliation with the Huawei Finland-funded lab. They do not work with Huawei, do not receive any money from them, and were in no way involved in the creation or implementation of that cooperation.
The RPA Human(e) AI stimulates and combines research into the legal, ethical and social consequences of AI, looking at how can we encourage the development of AI that is consistent with cultural and socio-economic equality and fundamental civil rights.
The collaboration between the VU, the UvA and Huawei Finland focuses on search technology in European languages. As with all UvA research, it is scientifically valuable and is fully and openly shared in accordance with the standards of the Netherlands Code of Conduct for Academic Integrity. Such matters are laid down in the contract between the parties to guarantee academic freedom and integrity. Moreover, it came about after consultations with the relevant national governmental authorities.
We understand that a partnership with a business whose parent company operates in a country that is not free raises many questions. Such situations are being playing out around the world. As AI research continues to gain momentum, it is becoming entangled in the problems of big-tech companies and other geopolitical developments.
Choices like those made by the VU and the UvA about the partnership with Huawei Finland are not simply black and white. Academic freedom and the importance of openness to collaboration with researchers regardless of their nationality must be weighed against the interests of companies and governments, and geopolitical and social considerations. The UvA is not the only one looking for answers in this area; the issue is being discussed and legislated for at national and European levels.
The UvA is of course also outraged about the reports about ethnic profiling by Huawei. It must be clear that all forms of ethnic profiling and use of ethnic profiling are completely unacceptable. The search technology aimed at European languages that our Dreamslab researchers are working on cannot be used for these types of purposes. It is understandable and right that this type of story – which came to light after the collaboration was entered into – leads to further discussion. In the future, we will offer more clarity on how these decisions on collaboration are reached.
The UvA is continuously working on improving decision-making processes and checks and balances in research and research collaborations. For example, research is subjected to ethical and legal review, and there are (nationally coordinated) protocols and guidelines for dealing with intellectual property. Dutch universities as a bloc have asked the national government to consider how it can provide clearer frameworks for cooperation with parties affiliated with unfree countries. And we are also working on assessment frameworks to help our scientists as much as possible when it comes to dealing with these kinds of dilemmas in future collaboration initiatives.