Language selection

Search

Processing Artificial Intelligence: Analysis from a Canadian Perspective

From: Canadian Intellectual Property Office


Analysis from a Canadian Perspective

This section provides a snapshot of Canadian involvement in AI based on researchers who patented inventions between 1998 and 2017. Canadian participation in this field is likely much larger based on those who use non-formal forms of IP, such as trade secrets, and involvement in scientific publications, among other activities. Based on the information extracted from the patent data, this report provides a partial indication of the overall activity level in AI using the detailed information that is captured from this source.

On this page

Canadian researchers

As organizations turn their focus towards the ever-evolving world of AI, researchers, academics and experts are coming into higher demand. This has led to a greater push by academic institutions to fund AI-related disciplines, such as computer science, electrical and electronic engineering, mathematics and statistics, and neuroscience to name a few.Footnote 18 In turn, retaining those trained in a country while simultaneously attracting those trained in other countries has become an initiative many governments have taken on. In their annual Global AI Talent Report,Footnote 19 Element AI has found that, ". . . the AI talent pool is highly mobile, with about one-third of researchers working for an employer based in a country that was different from the country they received their PhD." Of this talent pool, Canada is one of the leaders in high-impact research. Element AI defines high-impact research as a country where a ". . . higher-than-average percentage of the local talent pool is making high-impact contributions to the field" of AI. This attributes to Canada’s label as a platform country: a country which sees both an inflow and an outflow of talent. Our research in this section validates the aforementioned statement.

Patent filing trend

Patent filing activity by Canadian researchers in AI has experienced waves of increased activity over the past 20 years, with a significant increase over the past five years. The 31% average annual growth over this short period is similar to the growth rate experienced at the global level. As seen in Figure 11, inventions patented by Canadian researchers increased gradually between 1998 and 2001, before experiencing a significant surge in 2002. Canada experienced a slightly higher average annual growth rate of 24% between 2005 and 2010 compared to the global 20% growth rate observed during this timeframe.

Figure 11: Patent activity by Canadian researchers in AI between 1998 and 2017

Description of figure 11

Figure 11 consists of a line chart placed on top of a stacked bar chart. Each chart uses a different axis. The stacked bar chart shows the number of patented inventions by Canadian researchers filed at CIPO, USPTO, WIPO (PCT) and others from 1998 to 2017. This chart uses the axis on the left. The line chart shows the annual growth rate in the same time period and uses the axis on the right.

Figure 11: Patent activity by Canadian researchers in AI between 1998 and 2017
Publication year CIPO USPTO WIPO (PCT) Other Annual growth rate
1998 4 4 4 1  
1999 8 2 2 0 -7.69%
2000 9 6 5 0 66.67%
2001 4 20 5 0 45.00%
2002 17 31 10 0 100.00%
2003 14 29 5 1 -15.52%
2004 12 17 10 1 -18.37%
2005 13 15 5 1 -15.00%
2006 12 28 9 4 55.88%
2007 10 45 12 1 28.30%
2008 16 40 20 3 16.18%
2009 31 52 14 0 22.78%
2010 14 64 18 2 1.03%
2011 22 45 11 0 -20.41%
2012 16 40 18 1 -3.85%
2013 11 40 16 4 -5.33%
2014 21 43 24 5 30.99%
2015 16 66 23 6 19.35%
2016 20 65 36 13 18.92%
2017 25 138 34 14 58.33%

Relative specialization of Canadian researchers

In order to gain a better understanding of a country’s performance in terms of AI patent activity, we use the Relative Specialization Index (RSI) (additional details in Annex D). This measure uses patenting intensity to allow for technology sectors to be compared between countries of different sizes. The index provides a measure of each country’s share of patented inventions within the AI field as a share of the country’s total patented inventions produced within a given timeframe. For countries where the value is greater than zero, they are seen as being relatively specialized compared to the rest of the world. Conversely, countries with an RSI value of less than zero are deemed to not have a specialization. In Figure 12, we incorporate a time dimension to present the change in degree of specialization by splitting the dataset in two 10-year periods. It is interesting to see that the index scores for Canada, along with Germany, Japan, and the U.S., have decreased over the index scores for the first decade.

Figure 12: Relative Specialization Index by researcher’s country of origin in AI

Description of figure 12

Figure 12 shows a horizontal bar chart illustrating the revealed specialization index for seven countries, including Canada, for researchers. Each country has two bars with the green bar representing data from 2008 to 2017 on top of the blue bar representing data from 1998 to 2007.

Figure 12: Relative Specialization Index by researcher’s country of origin in AI
Publication year 1998-2007 2008-2017
Republic of Korea -0.56 -0.44
Germany -0.06 -0.27
Japan -0.07 -0.18
United Kingdom -0.14 -0.03
Australia 0.12 0.22
Canada 0.34 0.30
United states 0.40 0.35

The index can be further broken down at an AI Applications grouping level to determine areas in which Canadian researchers are specialized. As observed in Figure 13, Canadian researchers are highly specialized in Knowledge Representation and Reasoning, and NLP. It is interesting to compare RSI values between Canada and the U.S., as Canada appears to hold its own against a country that is a world leader in terms of patent activity in AI more broadly. The question then arises: how can Canadian researchers’ talent be leveraged to advance innovation in this field so as to allow this specialization to increase further?

Figure 13: Relative Specialization Indices by AI Applications for American and Canadian researchers

Description of figure 13

Figure 13 shows a horizontal bar chart illustrating the revealed specialization index for different AI applications for Canadian and American researchers. Each AI Application has two bars with the red bar representing Canada and blue bar representing the United States.

Figure 13: Relative Specialization Indices by AI Applications for American and Canadian researchers
AI application RSI Canada RSI United States
Distributed artificial intelligence -0.55 -0.65
Planning and scheduling -0.17 -0.28
Control methods -0.02 0.12
Speech processing 0.09 -0.06
Predictive analytics 0.08 0.32
Robotics 0.42 0.34
Knowledge representation and reasoning 0.48 0.29
Computer vision 0.43 1.72
Natural language processing 1.32 2.41

From the perspective of absolute patented invention counts, it is reassuring to observe that Canadian researchers are prolific filers in areas in which they are deemed specialized. This is not always the case, as seen with Robotics, where Canadian researchers do not hold many patented inventions but are considered specialized relative to researchers from other countries who file a proportionately lower number of patented inventions.

Figure 14: Growth in patent activity for Canadian researchers in various AI Applications

Description of figure 14

Figure 14 is a heatmap showing the growth experienced by Canadian researchers by AI Applications (sorted from most activity to least: Computer Vision, Natural Language Processing, Knowledge Representation and Reasoning, Control Methods, Speech Processing, Planning and Scheduling, Predictive Analytics, Robotics, and Distributed Artificial Intelligence) from 1998 to 2017. All AI Applications have been represented with shades of blue.

Figure 14: Growth in patent activity for Canadian researchers in various AI Applications
AI Applications 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Computer vision 1 1 0 1 6 6 2 6 4 3 5 16 8 6 5 14 6 11 11 38
Control methods 2 0 2 0 3 0 2 2 1 0 3 7 2 1 0 0 3 2 6 6
Distributed artificial intelligence 1 0 0 0 0 1 0 1 0 0 0 0 1 0 0 1 0 4 1 2
Knowledge representation and reasoning 1 0 0 0 3 0 1 2 2 3 2 5 7 3 9 10 5 2 15 24
Natural language processing 1 0 2 1 0 1 5 1 0 1 6 12 5 5 4 6 8 11 19 16
Planning and scheduling 0 0 0 0 2 1 0 1 1 1 1 2 4 3 2 1 3 0 2 4
Predictive analytics 0 0 0 0 0 0 1 0 1 0 1 0 1 1 0 1 2 1 2 7
Robotics 0 0 0 0 0 1 0 0 0 0 1 2 2 0 1 0 1 1 0 4
Speech processing 0 0 1 2 3 1 0 2 1 2 0 1 2 0 3 0 3 2 2 4

The leading Canadian researchers patenting in AI are presented in Figure 15 below. The institution associated with a particular researcher and the researcher’s primary area of expertise are shown in blue and red respectively in the figure below. Interestingly, 93% of Canadian institutions include at least one Canadian researcher on patented inventions included in this dataset. However, only 51% of Canadian researchers in the AI field are associated with Canadian institutions or a Canadian subsidiary of a multi-national company. Focusing on the top ten researchers, only two are associated with international institutions. Canada’s top patenting researchers have a significant presence in Canada, working for established companies, such as Primal AI and D-Wave Systems.

Figure 15: Top Canadian researchers along with their associated institutions (in blue) and their main area of expertise (in red)

Description of figure 15

Figure 15 is a horizontal bar chart illustrating the top Canadian researchers with their associated institutions (in blue, below the inventor name label) along with their associated main area of expertise (in red, beside each respective bar on the right).

Figure 15: Top Canadian researchers along with their associated institutions (in blue) and their main area of expertise (in red)
Researcher name Patented inventions Name of institution AI technique
Boyer, John M. 5 IBM, US Natural language processing
Hardjasa, Amelia 5 Pulse Energy, CA Predictive analytics
Ka-Hing, Lin 5 Rockwell Automation Technologies Inc, CA Control methods
Macready, William G. 6 D-Wave Systems Inc, CA Knowledge Representation and Reasoning
Francis, Ilyas Ihab 9 Primal AI, CA Knowledge Representation and Reasoning
Bertrand, Benoit Patrick 10 Accenture LLP, CA Knowledge Representation and Reasoning
Hatami-Hanza, Hamid 10 Science Counter Inc, CA Knowledge Representation and Reasoning
Natural Language Processing
Edson, Tirelli 10 Red Hat, US Other
Joseph, Sweeney Peter 15 Primal AI, CA Knowledge Representation and Reasoning
Voon, Gerard 27 Ecco Projects Inc, CA Computer vision

Beyond using patented invention data to identify trends and top filers, it is useful to further explore the types of technologies being created. A patent landscape map is presented in Figure 16, which was generated using an algorithm that relies on the keywords from patent documentation to cluster patented inventions according to shared language. The patented inventions are organized based on common themes and grouped as "contours" on the map to identify areas of high and low patent activity. The white peaks represent the highest concentrations of patented inventions, and each peak is labelled with key terms that tie the common themes together. The distance between keywords helps to illustrate the relationship between peaks, where shorter distances indicate that the patented inventions they represent share more commonalities relative to those that are further apart. Words located close together may be part of similar systems or technologies, whereas keywords that are further apart likely have less of a relationship.

Superimposing the names of the leading Canadian researchers is a useful way to highlight who is working in what space. For example, we see that Peter Joseph Sweeney and Ihab Francis Ilyas—both associated with Primal AI and patenting in the Knowledge Representation and Reasoning application field—are both captured in the middle-right side of the map and are associated with patented invention characterised by keywords such as "Atomic knowledge representation model", "Complex knowledge representation", "Concept", "Structure" and "Program". Perhaps as expected, we see that Amelia Hardjasa, who is the only leading researcher patenting in the Predictive Analytics application field, has patented inventions on the map that are located at the bottom centre on an island separated from the mainland on the map. This type of landscape map can also be useful to institutions in order to target talent. For example, in the case where a researcher’s patented invention portfolio overlaps with that of another researcher, it can indicate that the two are patenting in similar areas, as in the case of Gerard Voon and Patrick Benoit Bernard in the upper right quadrant of the map.

Figure 16: Landscape Map of Canadian researcher's patent activity highlighting top filers

Description of figure 16

Figure 16 is a patent landscape map that provides a visual representation of Canadian researcher’s patent activity in data set in this report. Derwent Innovation's ThemeScape mapping tool was utilized to produce this visualization, using term frequency (keywords from a patents title and abstract) and other algorithms to cluster documents based on shared language. The result is a patent landscape map, a map very much resembling that of a topographic map, where there are sections of turquoise and white. Sections are comprised of peaks, some of which have bright white peaks, representing the highest concentration of patented inventions and are labelled with key terms that tie common themes together. Turquoise is used to separate terms where there is no commonality between them. Eight peaks are highlighted in yellow on the patent landscape map corresponding to each of the top inventors identified below. The legend provides more details on the top inventors and key words as described in the report.

Top inventors include:

  • Gerard Voon
  • Patrick Benoit Bertrand
  • Lin Ka-Hing
  • William Macready
  • Peter Joseph Sweeny & Ihab Francis Ilyas
  • Hamid Hatami-Hanza
  • Edson Tirelli
  • Amelia Hardjasa

Broad themes written in all capital letters include:

  • RNA
  • PIXEL
  • VIRTUAL ENVIRONMENT
  • WORLD
  • RECOMMENDER SYSTEM
  • RECOMMENDER
  • QUANTUM
  • COUPLER
  • NEUROLOGICAL
  • PCRN
  • POWER SYSTEM
  • RULE NODE
  • HANDHELD ELECTRONIC DEVICE

Common key words in this graph include:

  • Sequence
  • Function
  • Select
  • Image
  • Pixel
  • Feature
  • Object
  • Recognition
  • Sensor
  • Camera
  • Video
  • Intelligence
  • Artificial
  • Artificial Intelligence
  • Customer
  • Derive
  • Location
  • Output
  • Programmable
  • Circuit
  • Vehicle
  • Operation
  • Classification
  • Classify
  • Class
  • Overflow
  • Round
  • Multiplier
  • Train
  • Semantic
  • Language
  • Design
  • Optimization
  • Solution
  • Optimization problem
  • Document
  • Search
  • Query
  • Atomic knowledge representation model
  • Complex knowledge representation
  • Knowledge representation model
  • Computer
  • Structure
  • Program
  • Knowledge
  • Human
  • Service
  • Cost
  • Patient
  • Medical
  • Monitor
  • Wearable
  • Signal
  • Analysis
  • Parameter
  • Fuzzy
  • Logic
  • Engine
  • Rule
  • Storage
  • Ontological subject
  • Ontological
  • Participation
  • Communication
  • Application
  • Server
  • Business
  • Personal
  • Assistant
  • Electronic
  • Singer-word term
  • Plurality of business category
  • Train data business name
  • Speech
  • Speech recognition system
  • Speech recognition
  • Neural
  • Layer
  • Question
  • Answer

Figure 17 shows the distribution of the patent activity by Canadian researchers by province across the country. Each province is shaded in orange, with darker shades representing a higher number of patented inventions. The number of researchers residing in prominent Census Metropolitan Areas (CMA) with prolific patenting populations is highlighted on the map using red location pointers. It should be noted that patented invention volumes were calculated using the fractional counting approach and have been normalized by population size in this figure. The provinces that have a higher number of patented inventions, such as Ontario, British Columbia and Quebec, remain the leading provinces after being adjusted. The provinces with more patent activity tend to be associated with clusters of researchers around large cities.

Figure 17: Geographical clusters of inventive activity by Canadian researchers

Description of figure 17

Figure 17 is a choropleth map depicting the geographical clusters of inventive activity by Canadian researchers. The map is shaded in blue with the darkest shade representing the highest number of patented inventions per population. Two types of labels are on the graph. A dot represents a census metropolitan area (CMA) with less than 20 researchers while a waypoint with a number labeled represents a CMA with at least 20 researchers. The number inside the waypoint represents the number of researchers in that specific CMA.

Number of patented inventions per population
Province Patented inventions / population
Northwest Territories 0
Nunavut 0
Alberta 1.48735E-05
Yukon 0
Saskatchewan 8.16868E-06
Newfoundland and Labrador 3.80515E-06
Ontario 4.14448E-05
British Columbia 3.86343E-05
Manitoba 1.04921E-05
Quebec 1.60857E-05
New Brunswick 9.08005E-06
Nova Scotia 5.94059E-06
Prince Edward Island 0
Geographical clusters of inventive activity by Canadian researchers
Census metropolitan area (CMA) Number of researchers
Abbotsford-Mission 1
Barrie 7
Belleville 1
Calgary 44
Edmonton 41
Greater Sudbury 1
Guelph 16
Halifax 8
Hamilton 18
Kelowna 1
Kingston 12
Kitchener-Cambridge-Waterloo 162
Lethbridge 3
London 21
Moncton 1
Montréal 171
Oshawa 11
Ottawa-Gatineau 273
Peterborough 2
Québec 29
Regina 8
Saint John 3
Saskatoon 9
Sherbrooke 4
St. Catharines-Niagara 9
St. John's 2
Thunder Bay 1
Toronto 496
Trois-Rivières 4
Vancouver 238
Victoria 19
Windsor 8
Winnipeg 22

Gender analysis: Female participation in AI patent activity

In 2014, 59% of graduates aged 25 to 34 in Canadian science and technology programs were female. However, only 23% of engineering graduates of that same age group were female.Footnote 20 Understanding that female representation in science, technology, engineering, and mathematics (STEM) fields is lower than desired, this observation is expected to also be reflected in female participation in patenting. To measure female participation in AI, we leverage WIPO’s comprehensive name dictionary to assign genders to the names of researchers listed on patented inventions.Footnote 21

At the international scale, there was one female identified for every three males involved in AI patenting. By comparison, for patented inventions containing at least one Canadian researcher, that ratio decreases to one female for every six male researchers.

Figure 18: Gender representation in AI, both globally (left) and in Canada (right)

Description of figure 18

Figure 18 illustrates the gender representation in AI. On the left is an image of an icon of a woman and three red icons of a man, representing the gender ratio on the international stage (1 woman for every 3 men). On the right is an icon of a woman and six icons of a man, representing the gender ratio in Canada (1 woman for every 6 men). The icons for woman are in green and ones for man are in violet. There is a large “vs” between the images in bold.

To better understand female participation in this technology field, Figure 19 effectively shows the evolution of females involved in AI over the twenty-year period. The following figure shows new researchers by gender entering the field based on their involvement as captured by the first patent application published. Unlike the trend observed for women in Canada, which is relatively steady over the twenty-year period, the international trend is much different, with female representation as a share of the total number of researchers decreasing over time.

Figure 19: Trend in gender distribution by new entrants in AI, both globally (top) and in Canada (bottom)

Description of figure 19

Figure 19 shows two 100% stacked column charts. The one on top illustrates the number of male, female and not assigned inventors on the international stage from 1998 to 2017. The graph on the bottom shows the same information for Canada. Blue bar represents the number of male inventors, green bar number of female and gray bar number of not assigned.

International
Publication year Male Female Not assigned Total

1998

1,067 1,037 210 2,314

1999

943 1,024 203 2,170

2000

1,087 900 210 2,197

2001

1,226 1,102 198 2,526

2002

1,936 1,265 310 3,511

2003

2,052 1,300 295 3,647

2004

2,106 1,347 304 3,757

2005

2,312 1,342 419 4,073

2006

2,767 1,492 466 4,725

2007

3,333 1,353 491 5,177

2008

4,173 1,636 682 6,491

2009

5,224 1,967 731 7,922

2010

5,716 2,032 889 8,637

2011

5,312 1,954 890 8,156

2012

6,474 2,106 1,072 9,652

2013

8,488 2,544 1,403 12,435

2014

10,197 2,833 1,724 14,754

2015

12,281 3,219 2,028 17,528

2016

14,422 3,639 2,593 20,654

2017

27,031 80,52 3,883 38,966
Canada
Publication year Male Female Not assigned Total

1998

14 4 0 18

1999

19 2 1 22

2000

16 2 1 19

2001

91 13 6 110

2002

128 11 7 146

2003

116 15 5 136

2004

102 19 5 126

2005

73 10 4 87

2006

75 9 5 89

2007

105 19 6 130

2008

118 18 8 144

2009

109 21 5 135

2010

128 14 13 155

2011

117 21 10 148

2012

101 23 7 131

2013

82 17 5 104

2014

138 25 19 182

2015

170 32 18 220

2016

199 36 23 258

2017

269 62 29 360

Non-traditional sectors, such as mining, forestry, electricity and the skilled trades among others, tend to experience a drop in women’s participation five to ten years after graduation.Footnote 22 This lack of retention could possibly be due to the systemic barriers faced by women who already work in the field, a theory more closely examined in a report CIPO had published titled "Women’s Participation in Patenting: An Analysis of PCT Applications Originating in Canada".Footnote iii Increasing this retention rate is one of the values set forth by the Department for Women and Gender Equality. A result of this would be an expected increase in women’s representation in patented inventions across STEM fields, which would include AI. In 2012, Germany’s Institute of Labour Economics published an article that looked into the causes for the lack of participation by women in patenting, and the results coincide with the observations made for Canada.Footnote 23

To continue growing in AI, Canada must work to retain our best talent and also attract new talent into the country. In Element AI’s 2019 Global AI Talent Report, Canada cements itself as a platform country.Footnote 24 Interestingly, the U.S., Germany and China are anchored countries: able to retain more talent but unable to attract talent into the country. Australia, however, finds itself as an inviting country; meaning it is able to draw in a large portion of international talent without losing much of its domestic talent. Canada is among the top countries in the world consisting of high-impact researchers, joined by the U.S., China, the U.K. and Australia.

Canadian institutions

After gaining a better understanding of the AI patent landscape from the perspective of Canadian researchers, this section examines AI from the perspective of Canadian institutions. A Canadian institution for the purpose of this report includes a corporate firm, an academic institution, or a government establishment. Overall, corporate firms were responsible for 82% of the patented inventions in this section, whereas academic institutions and government departments accounted for the remaining 15% and 3% of the patent activity, respectively. To better understand innovation in the field by Canadian institutions and where their specialization lies, this section examines the associated patent activity in Canada and abroad. To limit the overall data cleaning effort in order to capture patented inventions by institutions, CIPO only focused its attention to cleaning the institution data for select countries that were used to benchmark against Canada. These countries include Australia, Germany, the U.K., Japan, Republic of Korea, and the U.S., and were selected based on the fact that they are among the leaders in terms of rankings by the WIPO Technology Trends report and their inclusion in the UKIPO’s AI report.

Innovation Superclusters Initiative

The Innovation Superclusters Initiative, introduced in 2017, is an investment of $950 million, matched dollar for dollar by the private sector, to support "superclusters" across Canada, which bring together "small, medium-sized and large companies, academic institutions and not-for-profit organizations" to "transform regional innovation ecosystems".Footnote 25 While the Scale. AI supercluster is largely focussed on AI, each of the superclusters plan to support projects that will promote the use of AI in their respective industries. The superclusters include the Digital Technology Supercluster, the Protein Industries Supercluster, the Next Generation Manufacturing Supercluster, SCALE.AI Supercluster, and the Ocean Supercluster.Footnote 26 The funding is allocated to the superclusters to help support advances in each sector and to help build and support emerging start-ups.Footnote 27

The second investment is a joint venture between Canada and the U.K. known as the Canada-UK AI Initiative. This is a funding opportunity that requires interdisciplinary research between three major domains: social sciences and humanities, health and biomedical sciences, and natural sciences and engineering. The initiative looks to support the responsible development of AI while creating partnerships between researchers in Canada and the U.K.Footnote 28

Finally, in 2017, the Government of Canada appointed the Canadian Institute for Advanced Research (CIFAR) to develop and lead the Pan-Canadian Artificial Intelligence Strategy.Footnote 29 This strategy is a $125 million initiative in partnership with the Alberta Machine Intelligence Institute (Amii), the Montréal Institute for Learning Algorithms (Mila) and the Vector Institute.Footnote 30 This funding initiative looks to do several things:

  1. Increase the number of AI researchers and skilled graduates in Canada
  2. Connect three of the major centres for AI in Canada
  3. Invite conversation surrounding economic, ethical, policy and legal implications of advances in AI
  4. Support the national research community

These initiatives work to further AI development in Canada ethically and fairly. Many of these initiatives focus on improving the lives of Canadians by funding and supporting researchers and developers as they explore the world of AI.

Relative specialization of Canadian institutions

Prior to exploring the AI Canadian institution data more closely, it is useful to begin by establishing where Canadian institutions rank in terms of relative specialization. In comparison to the six other countries presented in Figure 20, Canada is seen to be specialized because it holds a positive index score. When breaking down the institution dataset over the two decades covered, it is interesting to observe that the specialization for all seven countries under consideration has decreased during the second decade. Later in this section, we will explore the AI fields in which Canadian institutions are predominantly patenting and their respective RSIs.

Figure 20: Relative Specialization Index by institution’s country of origin in AI

Description of figure 20

Figure 20 shows a horizontal bar chart illustrating the revealed specialization index for seven countries, including Canada, for institutions. Each country has two bars with the green bar representing data from 2008 to 2017 on top of the blue bar representing data from 1998 to 2007. The green and blue bars for Canada are highlighted in gray. Only bars for Canadian data are highlighted.

Figure 20: Relative Specialization Index by institution’s country of origin in AI
Publication year 1998-2007 2008-2017
Republic of Korea -0.21 -0.26
Germany 0.00 -0.24
United Kingdom -0.08 -0.13
Japan 0.62 0.11
Canada 0.26 0.22
Australia 0.31 0.22
United states 0.38 0.34

Patent filing trend

Figure 21 shows the general trend in AI patent filing activity by Canadian institutions. The annual growth rate of 8% between 1998 and 2010 for Canadian institutions is similar to the growth rate experienced at the global level during this timeframe. Although the 21% growth rate observed in patent activity by Canadian institutions over the 2011 and 2017 period is significantly higher, this rate is lower than the 31% growth rate observed at the international scale. Even though Canada has assumed a leadership role in defining a policy framework for AI,Footnote 31 their overall influence is limited considering Canadian institutions account for less than 1% of the total number of inventions patented by institutions globally.

Figure 21: Patent Activity by Canadian institutions in AI between 1998 and 2017

Description of figure 21

Figure 21 consists of a line chart placed on top of a stacked bar chart. Each chart uses a different axis. The stacked bar chart shows the number of patented inventions by Canadian institutions filed at CIPO, USPTO, WIPO (PCT) and others from 1998 to 2017. This chart uses the axis on the left. The line chart shows the annual growth rate in the same time period and uses the axis on the right.

Figure 21: Patent Activity by Canadian institutions in AI between 1998 and 2017
PUBLICATION YEAR CIPO USPTO WIPO (PCT) OTHERS ANNUAL GROWTH RATE
1998 3 2 6 1  
1999 3 1 1 1 -50%
2000 1 1 0 0 -67%
2001 5 4 3 0 500%
2002 15 2 7 0 100%
2003 10 2 3 0 -38%
2004 8 4 7 0 27%
2005 7 0 2 0 -53%
2006 7 1 4 4 78%
2007 11 12 7 1 94%
2008 13 3 15 2 6%
2009 12 10 13 0 6%
2010 6 12 10 2 -14%
2011 12 14 9 1 20%
2012 11 13 13 2 8%
2013 7 14 7 1 -26%
2014 13 12 12 3 38%
2015 13 18 17 2 25%
2016 19 19 30 1 38%
2017 20 53 33 7 61%

Understanding in which countries Canadian institutions are seeking protection for their inventions internationally provides an indication of which markets they are strategically targeting. Not surprisingly, Canadian institutions, apart from filing at CIPO, file predominantly in the U.S. because of its large market size, and have priority filings in that country for each year over the 20year timespan. The USPTO and CIPO administered 64% of all inventions patented by Canadian institutions. Inventions patented via the Patent Cooperation Treaty (PCT) system account for 32% of the filings. Other IPOs targeted by Canadian institutions, but to a significantly lower degree, include the IP Australia (IPA), China National Intellectual Property Administration (CNIPA), EPO, Japan Patent Office (JPO) and the French National Institute of Industrial Property (INPI).

Distribution of patented inventions

Since the early 1990s, IP has emerged as an important asset class to the corporate sector, be it in terms of protecting the value of its inventions resulting from significant investments in R&D or, alternatively, if there is interest in acquiring a targeted firm, or even simply if institutions are looking to exit the market and reap the benefits of their efforts.Footnote 32 Large acquisitions, resulting in significant transfers of IP, can create a monopoly-like condition by realizing economies of scale and driving out smaller players from the market. In this section, the IPCI is used to understand the distribution of patented inventions held by Canadian institutions actively patenting in AI and benchmark it against institutions from other leading countries. An index value closer to 0 in this section would indicate a country having a higher number of less-active institutions patenting in AI, whereas an index value closer to 1 would indicate a country has a few dominant players which patent extensively in AI. In reference to Figure 22, Canada has one of the lowest index values, thereby indicating that Canadian institutions operate in a highly competitive patenting AI environment.

Figure 22: IP Concentration Index for institutions from leading countries in AI

Description of figure 22

Figure 22 shows a horizontal bar chart illustrating the Intellectual Property Concentration Index for seven countries, including Canada, for institutions. The number of institutions for each country are included as data labels.

Figure 22: IP Concentration Index for institutions from leading countries in AI
Country of origin IP concentration index Number of institutions
Germany 0.1877 479 institutions
United Kingdom 0.0577 166 institutions
Republic of Korea 0.0539 613 institutions
Japan 0.0430 541 institutions
United States 0.0279 4,136 institutions
Canada 0.0152 284 institutions
Australia 0.0111 144 institutions
International 0.0039 14,277 institutions

Figure 23 shows the distribution of inventions patented by Canadian institutions and their overall representation in the Canadian AI institution dataset. The size of each pie depicts the proportion of Canadian institutions responsible for the inventions in each of the five groups, whereas the angle prescribed by each pie in the figure represents the proportion of patented inventions in each group. As expected, a vast majority of the Canadian institutions hold between one and four patented inventions. On the other hand, institutions holding five or more patented inventions account for 43% of the Canadian institution AI dataset. Nevertheless, simply holding more patented inventions does not reflect the importance of such inventions.

Figure 23: Distribution of Canadian institutions by patented inventions and their overall representation

Description of figure 23

Figure 23 is a radial bar chart depicting the distribution of Canadian institutions by patented inventions and their overall representation. The chart looks like a pie chart with pies of different radii. There are five pies representing the five groups created based on number of patented inventions (1 patented invention, 2-4 patented inventions, 5-9 patented inventions, 10-19 patented inventions and 20+ patented inventions). The angle of each pie represents the percentage of inventions that fall under each group while the size of each pie represents the proportion of Canadian institutions responsible for the inventions in each group.

Figure 23: Distribution of Canadian institutions by patented inventions and their overall representation
Patented invention group Number of Canadian institutions Total number of patented inventions
1 199 199
2-4 62 156
5-9 13 70
10-19 6 86
20+ 4 117

Total

284

628

Figure 24 shows the breakdown of the data by application fields in order to gauge the industrial applications of AI inventions patented by international and Canadian institutions, respectively. As observed in Figure 24, institutions globally have been consistently filing for patented inventions pertaining to Life and Medical Sciences, and Physical Sciences and Engineering between 1998 and 2017, whereas patented inventions pertaining to Transportation gained prominence post 2011. On the other hand, Canadian institutions seem to specialize in the field of Telecommunications, apart from Life and Medical Sciences, and Physical Sciences and Engineering. However, there is a need for Canadian institutions to be more specialized in the field of Transportation. This fact is corroborated by Figure 25, in which Canadian institutions seem to have a low specialization index in this field.

Figure 24: Growth in patent activity for international (top) and Canadian (bottom) institutions in various AI Fields

International institutions

Canadian institutions

Description of figure 24

Figure 24 shows two heatmaps. The heatmap on top shows the growth experienced by international institution in various AI Fields. All AI Fields are shaded in various shades of blue except for Life and Medical Sciences, Transportation, and Physical Sciences and Engineering, which are shaded red. The heatmap on the bottom shows the same information for Canadian institutions. All AI Fields are shaded blue on the graph for Canada.

Data for International institutions
AI field 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Agriculture 6 5 6 2 8 4 10 7 11 6 11 8 21 14 22 53 50 56 72 166
Arts and Humanities 4 15 10 6 12 13 16 24 26 42 31 30 37 32 30 36 43 48 63 118
Banking and Finance 31 22 35 30 33 38 25 36 40 38 54 86 95 103 138 143 198 239 289 556
Business 7 4 19 24 27 19 15 44 27 39 59 101 136 134 129 172 205 254 296 557
Cartography 4   6 3 7 4 3 8 9 14 12 19 24 19 39 49 41 44 72 134
Computing in Government 59 61 56 52 81 71 63 77 82 93 132 155 184 186 211 329 424 508 685 1,141
Document Management and Text Processing 5 5 5 16 16 17 10 30 28 29 21 51 43 33 33 53 48 63 87 185
Education 15 14 27 23 26 22 24 36 31 27 16 35 42 26 48 61 72 85 143 257
Energy management 21 29 11 11 30 27 28 18 24 16 35 42 43 71 105 158 220 235 340 506
Entertainment 12 10 12 19 29 21 20 20 12 29 21 23 25 16 16 32 32 44 53 101
Industry and Manufacturing 50 31 31 41 48 54 53 36 45 49 60 103 97 75 130 187 226 315 431 996
Law, Social and Behavioral Sciences     1 1 1         1 1 1 5   4 4 1 11 9 8
Life and Medical Sciences 94 129 129 152 185 189 199 224 215 276 304 349 452 418 566 639 802 1,002 1,150 1,899
Military 2 3 4 2 1 3 2 6 6 7 13 7 15 14 13 18 22 30 25 37
Networks 64 85 67 56 52 40 52 70 89 77 98 110 113 135 176 234 319 491 686 1,617
Personal Computers and PC Applications 40 37 54 46 66 45 46 68 78 77 111 120 125 110 134 167 220 252 291 551
Physical Sciences and Engineering 155 139 145 137 168 158 167 177 145 156 200 250 245 230 297 368 467 561 782 1,264
Publishing   2 3 2 2 2 2 3 6 3 3 4 3 5 7 7 11 11 13 27
Security 13 16 21 29 32 24 35 49 53 52 68 89 87 74 107 137 140 198 250 394
Telecommunications 76 63 75 87 97 99 90 87 114 122 161 159 201 203 225 311 359 542 651 1,130
Transportation 65 67 93 75 76 87 108 138 140 149 221 236 239 228 347 505 659 1,021 1,523 3,108
Data for Canadian institutions
AI FIELD 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Agriculture 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1
Arts and Humanities 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
Banking and Finance 0 1 1 1 2 2 0 1 0 0 1 0 0 1 0 0 1 0 0 1
Business 1 0 0 1 1 0 0 0 0 1 4 2 2 2 0 1 0 0 3 3
Cartography 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 2 4
Computing in Government 0 0 0 1 4 2 0 0 3 0 1 4 1 2 0 2 1 3 4 7
Document Management and Text Processing 0 0 0 1 1 1 0 0 0 1 1 1 0 0 0 0 0 0 0 0
Education 0 0 0 0 1 0 0 0 2 0 0 0 0 0 1 0 0 0 4 2
Energy Management 0 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 1 0 0 8
Entertainment 0 0 0 0 0 0 0 2 1 1 2 0 0 0 1 0 1 2 2 0
Industry and Manufacturing 0 0 0 1 1 0 0 0 1 0 1 0 2 2 0 1 1 3 1 6
Life and Medical Sciences 2 1 1 2 6 6 8 3 5 7 5 3 8 7 11 5 9 14 24 23
Military 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 1
Networks 1 0 0 0 0 1 0 1 0 1 0 1 0 1 2 2 0 4 4 12
Personal Computers and PC Applications 2 1 0 0 4 0 1 1 0 2 1 2 6 1 3 1 2 2 1 4
Physical Sciences and Engineering 2 1 1 3 4 2 5 3 1 5 3 7 6 7 4 1 12 9 15 9
Security 0 0 0 0 1 1 1 0 1 2 2 3 1 1 1 0 2 0 6 4
Telecommunications 6 2 1 5 2 3 2 3 0 4 3 5 4 4 2 3 3 4 1 6
Transportation 0 0 0 0 1 1 1 0 0 1 1 0 3 0 1 2 4 2 4 8

In reference to Figure 25, it is important to note that there are certain AI Fields, such as Banking and Finance, Military, Law, social and behavioural sciences, etc., that have not been included in the figure because they have not attracted considerable amounts of patent activity globally. As a result, the specialization indices for such AI fields may be biased.

Figure 25: Relative Specialization Index by AI Fields for American and Canadian institutions

Description of figure 25

Figure 25 shows a horizontal bar chart illustrating the revealed specialization index for different AI Field for Canadian and American institutions. Each AI Field has two bars with the blue bar representing Canada and red bar representing the United States.

Figure 25: Relative Specialization Index by AI Fields for American and Canadian institutions
AI field Canada United States
Transportation -0.36 -0.14
Networks -0.13 -0.12
Energy management -0.11 -0.51
Document management and text processing -0.08 0.11
Industry and manufacturing -0.04 -0.18
Computing in government -0.03 -0.13
Banking and finance -0.01 -2.11
Education 0.02 0.13
Business 0.04 0.02
Personal computers and PC applications 0.17 0.10
Telecommunications 0.21 -0.01
Security 0.22 0.06
Life and medical sciences 0.24 0.02
Physical sciences and engineering 0.25 0.03

Canada’s leading institutions patenting in the AI field over the 1998 to 2017 period are presented in Figure 26. Many of the institutions listed are prominent patenting entities that patent in a variety of technology fields while others, such as D-Wave and Primal AI, specialize in this field. It is also interesting to note that Nortel Networks, a now defunct business entity, is among the leading filers. The fact that owners of patented inventions are not required to update the information held in the patent database reflects a challenge when using patented invention data to identify owners of such rights.

Figure 26: Top Canadian institutions and their associated patented invention counts

Description of figure 26

Figure 26 is a horizontal bar chart showing the top 10 Canadian institutions and their associated invention counts.

Figure 26: Top Canadian institutions and their associated patented invention counts
Institution name Patented inventions
University of British Columbia 11
1QB Information Technologies Inc. 12
Alcatel-Lucent Canada Inc. 13
IBM Canada Ltd. 13
Maluuba Inc. 14
Primal AI 15
Nortel Networks Corporation 19
D-Wave Systems Inc. 25
Schlumberger Canada Ltd. 25
BlackBerry Ltd. 40

Presenting the patent landscape map for Canadian institutions is another way to visualize patented invention data. In Figure 27, we superimpose the names of the top ten leading Canadian institutions to highlight who is working in which space. Unlike the landscape map for Canadian researchers, there is not much overlap in areas where multiple institutions are patenting, the exception being D-Wave Technologies and 1QB, which are both operating in the quantum computing technology field. One thing that is striking is that Primal AI’s patented inventions are captured in a peak on a peninsula secluded from the rest in the bottom-centre of the map. This peak is characterized by keywords such as "Knowledge Representation", "Concept", and "Data Structure". This placement on the map may suggest the company is operating in a niche area, far different than technologies related to those captured in the larger AI technology space.

Figure 27: Landscape map of Canadian institutions in AI patent activity highlighting top filers

Description of figure 27

Figure 27 is a patent landscape map that provides a visual representation of the patent activity by Canadian institutions. Derwent Innovation's ThemeScape mapping tool was utilized to produce this visualization, using term frequency (keywords from a patents title and abstract) and other algorithms to cluster documents based on shared language. The result is a patent landscape map, a map very much resembling that of a topographic map, where there are sections of turquoise and white. Sections are comprised of peaks, some of which have bright white peaks, representing the highest concentration of patented inventions and are labelled with key terms that tie common themes together. Turquoise is used to separate terms where there is no commonality between them. Ten peaks have been highlighted in yellow on the patent landscape map corresponding to the legend below. 

Top institutions include: 

  • Nortel
  • Alcatel-Lucent
  • Blackberry
  • University of British Columbia
  • Schlumberger
  • Primal.ai
  • IBM Canada
  • Maluuba
  • 1QB
  • D-Wave

Broad themes written in all capital letters include: 

  • SUBSCRIBER
  • PACER
  • CANCER
  • Heart Rate
  • SUBTERRANEAN
  • KNOWLEDGE REPRESENTATION
  • HUMIDITY
  • TWITTER
  • REDUCER
  • OPTIMIZATION PROBLEM
  • QUANTUM

Common key words in this graph include: 

  • Management
  • Readable
  • Schedule
  • Service
  • Associative
  • Presence
  • Exemplary embodiment
  • Exemplary
  • Charge
  • Application
  • Resource
  • Database
  • Personal
  • Digital
  • Assistant
  • Disease
  • Patient
  • Sample
  • Tissue
  • Region
  • Sensor
  • Sense
  • Vehicle
  • Image
  • Object
  • Segmentation
  • Drill
  • Wellbore
  • Oil
  • Cluster
  • Protein
  • Structure
  • Signal
  • Speech
  • Output
  • Knowledge representation
  • Concept
  • Data structure
  • Portable compute device
  • Simulator
  • Process unit
  • Business category
  • Business name
  • Correct business category
  • Neuron
  • Neural network
  • Neural
  • Classification
  • Train
  • Enable
  • Design
  • Circuit
  • Optimization
  • Quantum
  • Processor
  • Quantum processor

Geographical clusters

In Figure 28, Canada’s CMAs associated with more than 10 patented inventions by AI institutions are presented in a geographic map. It should be noted that patented invention volumes were calculated using the fractional counting approach and have been normalized by GDP in this figure. In 2018, 650 startups were created "across all cluster cities". Many of these start-ups found support in local investors who helped give credibility to the ecosystem, leading to interest from international investors. The attention received by foreign investors has raised AI related deals by 41% and acquisition rates by 50% each year on average.Footnote 33 There are many benefits for institutions to cluster together, including increased productivity, faster innovation through collaborative research, and the creation of small institutions to cater to the niche needs of the industry.

Figure 28: Geographical clusters of inventive activity by Canadian institutions

Description of figure 28

Figure 28 is a choropleth map depicting the geographical clusters of inventive activity by Canadian institutions. The map is shaded in blue with the darkest shade representing the highest number of patented inventions. Two types of labels are on the graph. A dot represents a census metropolitan area (CMA) with less than 10 institutions while a waypoint with a number labeled represents a CMA with at least 10 institutions. The number inside the waypoint represents the number of institutions in that specific CMA.

Number of patented inventions
Province Patented inventions / GDP
Northwest Territories 0
Nunavut 0
Alberta 1.68208E-10
Yukon 0
Saskatchewan 1.48738E-10
Newfoundland and Labrador 1.20333E-10
Ontario 3.71479E-10
British Columbia 3.63912E-10
Manitoba 5.50297E-11
Quebec 2.03698E-10
New Brunswick 1.62311E-10
Nova Scotia 1.35275E-10
Prince Edward Island 1.4298E-10
Geographical clusters of inventive activity by Canadian institutions
Census metropolitan area (CMA) Number of institutions
Calgary 19
Edmonton 4
Guelph 2
Halifax 5
Hamilton 4
Kingston 3
Kitchener-Cambridge-Waterloo 20
Lethbridge 1
London 2
Moncton 1
Montréal 42
Oshawa 2
Ottawa-Gatineau 30
Québec 4
Regina 1
Saskatoon 4
Sherbrooke' 1
St. Catharines-Niagara 1
St. John's 1
Toronto 78
Trois-Rivières 2
Vancouver 30
Victoria 4
Winnipeg 3

There are six institution clusters, each comprised of ten or more institutions, that emerge as key areas leading innovation in the Canadian AI sector. Patenting from these clusters accounts for 83% of AI patented inventions nationwide. Similar to the geographic map in the Canadian Researchers section, the provinces are shaded in orange and normalized by their GDP. Most of the institutions that have inventions patented in AI are located in Ontario, which holds three main CMA clusters: Toronto, Ottawa–Gatineau, and Kitchener–Cambridge–Waterloo. The largest is the CMA of Toronto, which boasts 78 patenting AI institutions. Other CMAs with large clusters include Montréal (42 institutions), Vancouver (30 institutions) and Calgary (19 institutions).

Figure 29 takes a closer look at the relative specializations for each of the CMAs by AI Field. It is interesting to note that most of the prominent CMAs are specialized in a different AI. The Toronto CMA specializes in Networks, and Life and Medical Sciences, whereas the Kitchener–Cambridge–Waterloo CMA specializes in Computing in Government, and Personal Computers and PC Applications. Toronto has become an integral part of the national ecosystem, partly owing to the Vector Institute, a "not-for-profit corporation" that works with start-ups, the marketplace, incubators and accelerators to help drive AI research.Footnote 34 Toronto has attracted a lot of attention from foreign investors because of its reputation as Canada’s financial capital.Footnote 35 The city has also teamed up with Waterloo to gain the title of Silicon Valley of the North.Footnote 36

Figure 29: Relative specialization of census metropolitan areas by AI Fields

Description of figure 29

Figure 29 shows a horizontal bar chart illustrating the revealed specialization index for specified AI Fields for Canadian CMA clusters. Each AI Field has 6 bars representing each of the specified CMA clusters.

Telecommunications AI field
CMA RSI
Montreal 0.4575
Kitchener-Cambridge-Waterloo 0.1139
Ottawa-Gatineau -0.0305
Toronto -0.1806
Calgary -0.3035
Vancouver -0.3848
Physical sciences and engineering AI field
CMA RSI
Calgary 0.6229
Vancouver 0.1583
Montreal -0.0165
Toronto -0.1998
Ottawa-Gatineau -0.6056
Kitchener-Cambridge-Waterloo -0.7325
Personal computers and PC applications AI field
CMA RSI
Kitchener-Cambridge-Waterloo 0.2006
Toronto 0.1829
Montreal 0.0215
Ottawa-Gatineau -0.1314
Calgary -0.1546
Vancouver -0.3360
Networks AI field
CMA RSI
Toronto 0.2703
Ottawa-Gatineau 0.0065
Kitchener-Cambridge-Waterloo -0.0089
Vancouver -0.1400
Montreal -0.1804
Calgary -0.5607
Life and medical sciences AI field
CMA RSI
Toronto 0.1752
Montreal 0.0434
Vancouver 0.0151
Ottawa-Gatineau -0.0982
Calgary -0.2684
Kitchener-Cambridge-Waterloo -0.8570
Computing in government AI field
CMA RSI
Kitchener-Cambridge-Waterloo 0.3425
Ottawa-Gatineau 0.1479
Calgary 0.0853
Montreal 0.0790
Vancouver -0.0961
Toronto -0.8374

Montréal was identified as a cluster specializing in Telecommunications. Relying on Mila, Montréal has a research focus on a wide range of topics, including deep learning, recurrent neural networks and generative models.Footnote 37 In addition to Mila, Montréal has another institution called IVADO, which focuses more on industrial research.Footnote 38 This combination has made them a target for international investors, leading to a wide range of deals and generous research funding. Last but not the least, Calgary was found to be specialized in Physical Sciences and Engineering.

Profile of Canadian institutions patenting in the field of artificial intelligence

While analysis of the patent data itself already yields substantial information on innovation in the area of AI by Canadian institutions and researchers, a more complete picture of the environment can be obtained when this data is linked to other data sources. With the support of Global Affairs Canada and CIPO, Statistics Canada linked the patented inventions for Canadian institutions from the EPO-PATSTAT database to data on the characteristics of institutions held at Statistics Canada. This linked data source sheds lights on the industry, size, and ownership characteristics of Canadian institutions patenting in AI.

Industry

A large percentage of patented inventions in the field of AI by Canadian institutions come from the Professional, Scientific, and Technical Services industry. Over the entire 2001 to 2016 period, 41% of AI patented inventions came from this industry. This is far higher than the 23% accounted for by the next largest source, the Manufacturing industry. When patenting across all fields is considered, the situation is reversed. The Manufacturing industry accounts for 47% of the patented inventions in all fields, while Professional, Scientific, and Technical Services industry accounts for 17%.

Figure 30: Breakdown of patent activity by industry sector across all sectors (left) and in AI (right)

Description of figure 30

Figure 30 shows two pie charts. The one on the left shows the breakdown of total patented inventions in Canada by industry sector. The pie chart on the right shows the same information for patented inventions in AI.

Figure 30: Breakdown of patent activity by industry sector across all sectors (left) and in AI (right)
NAICS All sectors AI
Manufacturing 38,373 81
Professional, Scientific and Technical Services 13,853 146
Others 30,012 129

Institution size

Innovation occurs in institutions of all sizes. While innovation is often associated with younger, smaller institutions, larger institutions have more specialized resources and finances to sustain development efforts. The data on patent activity in AI is broadly consistent with this. Although 59% of the patented inventions in AI were made by institutions with less than 100 employees over the 2001 to 2016 period, the share of AI patented inventions accounted for by large institutions with 500 or more employees was also substantial at 27%. Compared with patenting in all fields, where 44% of patented inventions come from small institutions and 33% come from large institutions, patenting in AI is more concentrated in smaller institutions.

Figure 31: Breakdown of patent activity by institution size across all sectors (left) and in AI (right)

Description of figure 31

Figure 30 shows two pie charts. The one on the left shows the breakdown of total patented inventions in Canada by institution size. The pie chart on the right shows the same information for patented inventions in AI.

Figure 31: Breakdown of patent activity by institution size across all sectors (left) and in AI (right)
Size All sectors AI
Large (500 or more employees) 27,079 98
Medium (100 to 499 employees) 18,671 49
Small (99 or fewer employees) 36,488 209

Ownership

The share of patented inventions in AI by Canadian institutions that are foreign controlled is lower than for overall patented inventions. Over the 2009 to 2016 period, 10% of the patented inventions in AI were made by foreign-controlled Canadian institutions. This is compared to 15% for patented inventions overall.

Figure 32: Breakdown of patent activity by ownership across all sectors (left) and in AI (right)

Description of figure 32

Figure 32 shows two pie charts. The one on the left shows the breakdown of total patented inventions in Canada by foreign ownership. The pie chart on the right shows the same information for patented inventions in AI.

Figure 32: Breakdown of patent activity by ownership across all sectors (left) and in AI (right)
Foreign Ownership All sectors AI
Domestic 34,027 207
Foreign 5,965 23

AI In health

Canada is hailed as one of the leaders in healthcare AI. One initiative that is being heavily leveraged by Canada is the Equitable AI Initiative. This initiative focuses on using AI to provide opportunities in the area of public health not only to analyze complex data but also to design and provide impactful solutions that draw on a wider range of insights. This is done through the initiative’s goals to fund, train and promote knowledge transfer across platforms.Footnote 39 At the University of Alberta, scientists are researching and creating prototypes of a bionic arm that can learn and anticipate the movement of its wearer to allow for better and smoother use of the prosthetic.Footnote 40 At Humber River Hospital, affectionately dubbed the first all-digital hospital, resides a command centre that tracks the flow of patients from their intake to their discharge, analyzes the data and reports back where there are slowdowns in the process.Footnote 41 This allows those working in the hospital to know where and why delays are caused and allows them to fix the problem before it arises. Humber River has taken it a step further and moved from electronic health records to a fully automated hospital.Footnote 42 From robots sorting medicine to machines delivering blood samples from patients to the lab, Humber River has made use of AI in almost all areas of the hospital.Footnote 43

Opportunities and challenges for integrating AI at intellectual property offices

AI and IP Policy

The world of IP is changing at an incredible pace. This change is driven by a number of factors, including rising value of intangibles, emergence of new countries as IP powerhouses, convergence of science and technology, increasing volumes of IP filing, increasing complexity of applications, interdisciplinary nature of innovation, and changing nature of work as a result of the integration of new technology.

CIPO is committed to ensuring that our IP system supports transformative technologies, such as AI. As we consider the implications of AI for the Canadian IP regime, Canada is adopting a whole-of-government approach to ensure that the Canadian IP system is well equipped to support the emergence of transformative technologies. Although we are still in the early stages of considering the implications of AI for the Canadian IP regime, CIPO is working with leading thinkers, policy makers, academics, practitioners and international partners to analyze the implications of AI for IP policy and law. In particular, we are engaged in discussion on:

  • the policy questions being raised for IP, creation and innovation, such as authorship and inventorship;
  • the copyright considerations in using copyright protected work to train AI algorithms and data; and
  • the best practices to address the implementation of AI in a manner compatible with core administrative law principles, such as transparency, accountability, legality, and procedural fairness.

AI operations

Transformative technologies have the potential to benefit the administration of the IP system. CIPO is looking to use AI technologies to gain efficiencies both in the delivery of timely and quality IP rights and the provision of a modern service experience. Toward this end, CIPO is launching and managing multiple AI projects.

Date modified: