Contrasting Perspectives: Google and the EU's Visions for the Future of AI.
In contrast to the rapid deployment of new AI products by tech giants like Google, the European Union is actively developing legislation aimed at limiting their impact.
The race to deploy artificial intelligence is unfolding in parallel with efforts to regulate its use, as exemplified by two significant developments this week.
On May 10th, Google unveiled its plans to integrate new large language models, powered by machine learning techniques, into its existing product line-up. During a press conference, Sundar Pichai, CEO of Google's parent company Alphabet, highlighted the company's vision to reimagine core products, including search. This strategic move is widely interpreted as a response to Microsoft's recent inclusion of similar capabilities in its search engine, Bing.
Just a day later, European Union politicians reached a consensus on new regulations that outline the parameters for the deployment of AI. The AI Act, a legislative framework that has been in the works for years, swiftly adapted to keep pace with technological advancements. In the past month alone, lawmakers drafted and approved regulations governing the use of generative AIs, which have gained tremendous popularity in recent months. Notably, these rules mandate the disclosure of any copyrighted material employed in the training of such AIs. The draft text is set to undergo a vote in the European Parliament in June.
However, it seems that Google, along with Microsoft and other tech giants, is largely overlooking what could soon become the most influential AI legislation worldwide. Although EU laws are only binding within member countries, the extensive reach of the bloc means that companies often end up complying with its regulations on a global scale, similar to the impact of the EU's General Data Protection Regulation (GDPR).
So, how can we reconcile this apparent contradiction? Carissa Véliz, a scholar at the University of Oxford, expresses her concern: "Although I may be mistaken, it seems that these companies' disregard for copyright issues is a deliberate strategic maneuver. They are wagering that their products are so enticing that governments will need to adapt to them, rather than these companies conforming their products to the rule of law."
While certain AI companies have established licensing agreements for copyrighted material, others seem to be adopting an approach of seeking forgiveness rather than seeking permission. The EU's AI Act may eventually compel companies to formalize their utilization of copyrighted content, but the precise implications remain uncertain.
According to Michael Veale from University College London, it is likely that companies such as Google will create their own versions of a Content ID system, similar to the one used on YouTube. This system would enable rights holders to identify and manage their content by either removing it or monetizing it. Veale speculates that AI firms may be exploring similar models that allow them to comply with regulations while maintaining control over pricing. Despite a request for comment, Google did not provide a response.
