Page 44 - CCCA Magazine Winter 2017

{ leGal iNNovatioN }

ArtiFiCiAl intelligenCe:

spotting tomorroW’s trends todAY

By Carole Piovesan and Kirsten Thompson

rtifcial intelligence (AI) is likely one spam legislation (CASL) and the Personal indemnity requirements to protect parties
of the most powerful transformative Information Protection and Electronic Docu- from AI-related harm?
Atechnologies of the modern era. AI is ments Act (PIPEDA), as well as international Competition: It is conceivable that AI sys-
an umbrella term for a computer system’s data protection legislation. Counsel needs to tems could cause or materially contribute
ability to perform tasks normally associ- think practically about the source and man- anti-competitive behaviour. For example,
ated with human intelligence. Examples ner in which data is collected; the purpose if an AI is granted sales/pricing decisions,
range from assisted AI such as GPS and and intended use of the data; whether con- it is possible that one system could collude
virtual assistants, to chatbots capable of sent for the data is required or has been prop- with another AI, exposing the companies
felding customer service questions, to au- erly obtained; and the manner in which data to potential price fxing liability. The Com-
3
tonomous systems such as self-driving cars is stored and accessed, and by whom. petition Act sets out criminal and civil sanc-
and advanced robotics. Intellectual Property: While algorithms tions against persons who are involved in
AI is expected to contribute $15.7 tril- per se are not patentable, computer pro- agreements with competitors that fx prices,
1
lion to the global economy by 2030. Un- gram products or methods that implement
derstanding AI and identifying opportuni- a tangible solution, or produce a discern- restrict supply or allocate customers or mar-
kets; or who are involved in bid-rigging, de-
ties for machine learning is becoming an able effect or change, generally are pat-
important business skill. This is true for entable. Under what circumstances can ceptive telemarketing, or wilful or reckless
2
4
lawyers as well—the ability to support op- and should an AI system, or the methods misleading advertising offences; or who en-
portunities and identify legal risks associ- created or used by the system, be patent- gage in abuse of a dominant market position.
ated with AI will soon be a big value-add ed? What other IP-related issues arise as next steps
for in-house counsel. a result of the creation or deployment of In-house counsel should be ready to con-
sophisticated AI systems? Can AI-created
legal risk management for Ai systems innovations—a song, for instance—be sider, advise on and respond to business
An AI-specifc risk management approach protected as original IP? requests to create or deploy AI systems of
is particularly important as companies are increasing complexity and ability. Devel-
starting to incorporate more complex and Employment & Labour: Greater AI integra- oping an AI-specifc risk management ap-
autonomous systems. Begin by consider- tion has deep implications from an employ- proach requires an understanding of the
ing the nature, purpose and control of the ment law perspective. For instance, the use of technology and its intended use, as well as
system: AI systems in recruitment is gaining in popu- a general mapping of the various potential
■ Is the system best characterized as a larity. At the same time, caution is required to legal issues it could cause and the protec-
ensure systems are used in a fair, transparent tions needed to avoid those harms. For
product (i.e., software) or a service?
What is its purpose? manner to avoid unintended bias and possi- more information, see McCarthy Tétrault’s
■ Is it an instrument, an agent or an au- ble human rights issues. Separately, the likely white paper, “From Chatbots to Self-Driv-
impact of extensive automation may require ing Cars: The Legal Risks of Adopting Arti-
tonomous being? What are the legal 5
implications of the different character- an entirely distinct HR strategy around mass fcial Intelligence in Your Business.” ❚
retraining of staff or their replacement. Early
izations?
■ Who retains control over the system? consideration of how automation may affect Carole Piovesan is a litigator in McCarthy Tétrault’s
HR in a business is critical.
The developer or the entity deploying Toronto offce who regularly consults across various
the system? Contract: Counsel should refect upon the sectors on the evolution and integration of AI and
need for specifc contractual provisions in its legal implications.
Then consider the possible implications of the use, purchase and licensing of AI tech-
AI systems in various contexts: Kirsten Thompson leads McCarthy Tétrault’s
nology. Are there express or implied war- national Cybersecurity, Privacy and Data Manage-
Privacy: AI systems require vast amounts of ranties that can preclude or limit recovery ment group, providing privacy, data security and
data to be successful. Access to data is sub- from liability for a breach caused by an AI information governance advice to clients in a wide
ject to legal restrictions under Canada’s anti- system? Are there unique insurance and variety of industries.

1 PricewaterhouseCoopers (2017), “sizing the prize”: https://www.pwc.com/gx/en/issues/analytics/assets/pwc-ai-analysis-sizing-the-prize-report.pdf
2 hume, Kathryn (20 oct 2017), “how to spot a Machine Learning opportunity, even if You aren’t a data scientist,” harvard Business review: https://hbr.org/2017/10/how-to-spot-a-machine-
learning-opportunity-even-if-you-arent-a-data-scientist
3 McCarthy Tétrault (2017), “doing Business in Canada”: http://www.mccarthy.ca/news_release_detail.aspx?id=7353
4 McCarthy Tétrault (2017), “doing Business in Canada”: http://www.mccarthy.ca/news_release_detail.aspx?id=7353
5 McCarthy Tétrault (sep 2017), “from Chatbots to self-driving Cars”: https://www.mccarthy.ca/pubs/from_Chatbots_to_self_driving_Cars_White_Paper_sePT2017.pdf

44 CCCa MaGazine | WinTer 2017 hiVer