Prime Minister Justin Trudeau met earlier this week with Jean-Francois Gagné, the CEO of Element AI, the Montreal-based applied artificial intelligence lab. Trudeau tweeted that the two men “talked about what Canadians are doing in AI in Montreal & across the country, and how we can keep the industry thriving.”
Well, since the Prime Minister asked, Element AI recently appeared before the Standing Committee on Industry, Science and Technology as part of the copyright review with a simple ask: create a new copyright exception for information analysis. Element AI was joined by several other technology groups, including ITAC and the Business Software Alliance, who all raised the same issue.
I highlighted the link between copyright and AI in a column last year, calling for a text and data mining exception. The column noted:
Making machines smart – whether engaging in automated translation, big data analytics, or new search capabilities – is dependent upon the data being fed into the system. Machines learn by scanning, reading, listening or viewing human created works. The better the inputs, the better the output and the reduced likelihood that results may be biased or inaccurate.
Copyright law crops up because restrictive rules may limit the data sets that can used for machine learning purposes, resulting in fewer pictures to scan, videos to watch or text to analyze. Given the absence of a clear rule to permit machine learning in Canadian copyright law (often called a text and data mining exception), our legal framework trails behind other countries that have reduced risks associated with using data sets in AI activities.
Element AI describes information analysis as “encompassing the use of processing techniques to obtain and process text, images, sound, video, and other forms of data in order to generate new facts, discover patterns, and analyse relationships.” It notes that current uncertainties in the Canadian Copyright Act limit the ability for Canadian companies to “access a basic, necessary resource to train their algorithms.” The solution? A fair dealing exception for informational analysis:
Recognizing a limited informational analysis fair dealing exception will echo and support research to grow Canada’s IP culture. This exemption can help all players of Canada’s AI ecosystem further test, train and research new innovative techniques, thereby connecting with the clear policy objectives of fostering a culture in Canada of innovation and intellectual property.
The Element AI position is echoed in similar submissions to the committee from the BSA (which includes members such as Apple, Adobe, and Oracle) and Microsoft, both strong advocates for copyright. The BSA warns:
As currently enacted, Canada’s Copyright Act creates uncertainty about the legal implications of key analytical techniques, such as text and data mining and machine learning, that are foundational to the development of AI. Accordingly, to help realize Canada’s strategy for becoming a global leader in AI and to facilitate the many societal benefits of AI, we urge the Committee to recommend the adoption of an express exception to ensure that copying a lawfully accessed work for the purpose of “information analysis” is not infringing.
Microsoft’s brief to the committee calls for a machine learning exception, noting that many other countries have developed similar exceptions:
Outside of Canada, countries, including Japan, the United States and China, have extended and are expanding legal protections for broad machine learning techniques including text and data mining. Japan recently implemented changes to its copyright laws that significantly expanded and clarified an already forward-looking machine learning exception. The UK permits text and data mining for certain purposes, and is currently exploring broadening its exception as it increasingly recognizes the benefits of machine learning for all users. China, Singapore and Thailand are focused on similar digital copyright reform and have similarly proposed broad and unrestricted machine learning exceptions. Unsurprisingly, these countries are also at the forefront of research involving data analytics and artificial intelligence, including machine learning.
As I argued last year, there are two ways to overcome the AI barrier:
First, Canada could emulate the U.S. fair use model by making the current list of fair dealing purposes illustrative rather than exhaustive. The U.S. exception is open to any purpose, as striking a fair balance depends upon the use of the work, not the purpose of the copying. Since machine learning does not harm the primary purposes of the original work, most text and data mining will qualify as fair use.
Second, other countries have tried to address the issue by creating a specific exception for text and data mining or computer informational analysis. For example, Britain’s exception allows copies of works to be made without permission of the copyright owner for the purposes of automated analytical techniques to analyze text and data for patterns, trends, and other information. The law does not allow contracts to restrict data mining activities, but the exception is limited to non-commercial research.
Canada’s significant investment in AI needs a legal framework that ensure Canadian businesses and researchers are not placed at a global disadvantage. Whether by way of fair use provision or a more targeted informational analysis fair dealing exception, the government’s hopes for Canadian AI leadership is linked to AI-focused copyright reforms.