Mr. Speaker, I am pleased to speak to this bill after my colleague from Rivière‑des‑Mille‑Îles, whom I would like to congratulate. I am also pleased to be following my colleague from Trois‑Rivières, an ethics expert who enlightened us on the potential impact of this bill and the dangers involved.
Unfortunately, very few people are interested in this type of bill, and yet, in the digital age, we cannot afford not to regulate the use of personal information. We cannot deny the fact that the digital shift has exploded in Quebec and elsewhere over the last decade, and it has greatly changed our lifestyles.
It is impressive to see which path companies have chosen during the pandemic, and I think it is a timely discussion to have today. However, I would like to draw attention to the new part of the bill that deals with artificial intelligence. I think it deserves serious consideration.
Part 3 of the bill raises many questions, and opinions from experts in the field of artificial intelligence are mixed. The use of artificial intelligence is a rapidly growing field that risks expanding beyond our control and jurisdiction if we do not begin to regulate the practice and define certain concepts.
Recent developments in AI in general and deep learning in particular have led to the creation of autonomous intelligent agents, which are essentially robots capable of deciding what to do without third-party intervention. These agents' autonomy raises new questions about civil liability, so we have to think about criminal provisions that would apply if someone were put in a dangerous situation, for example.
How should we approach this, and what legal status are we granting them? What legislative framework is the best fit for these autonomous agents?
At this point, we think some important definitions are missing. The law clerks who are examining the bill's provisions from a legal standpoint told us that again today. What is a high-risk intelligence system? What is a high-impact system?
The algorithms produced in applications that use artificial intelligence enable artificial beings to create goods or services or to generate predictions or results. If we compare them to human beings and use the existing framework, how will we interpret the notions of independence and unpredictability attributable to these artificial beings? The experts will help us understand all that.
Quite a few goods already exist that have a layer of artificial intelligence built into them, and 90% of those goods should not pose a problem. Experts at Meta have even said that this technology has reached its limits, because the data to train an algorithm is insufficient in quantity and lacks depth.
Let us get back to the main problem we have with Bill C‑27. Until the department clarifies its thinking on what constitutes a high-impact system, it will be difficult to assess the scope of part 3. Let us assume that everything can be considered high risk. This would mean that many companies would be accountable. If we had greater accountability, the Googles of this world might be the only ones that could risk using artificial intelligence.
The bill does not need to cover everything a machine can do for us or everything software can do once it is developed and generates predictions and results like a calculator.
If we compare it to the European legislation, we note that the latter is currently targeting employment discrimination systems, systems that would determine whether or not a permit to study there can be granted. That is essentially the limit of what the machine can do in our place.
Although the law in this document concerning artificial intelligence is far from being exhaustive, I believe it is important that we start somewhere. By starting here, with a framework, we can lay the groundwork for a more comprehensive law.
My speech this evening will help my colleagues better understand what needs to be clarified as soon as possible so we can have an important discussion about how to regulate the applications that use artificial intelligence and how to process these systems' data.
First, we will have to implement regulations for international and interprovincial exchanges for artificial intelligence systems by establishing Canada-wide requirements for the design, development and use of AI systems. Next, we must prohibit certain uses of AI that may adversely affect individuals.
The legislation is very clear on many other aspects, including on the fact that there would be a requirement to name a person responsible for artificial intelligence within organizations that use this technology. The responsibilities are fairly extensive.
In addition to the artificial intelligence and data act, which is in part 3, Bill C‑27 also includes, in part 1, the consumer privacy protection act, as well as the amendments to the former legislation. Part 2 of the bill enacts the personal information and data protection tribunal act, while part 4 includes the coming into force provisions of the bill.
As my colleagues explained, the other sections of the bill contain a lot of useful elements, such as the creation of a tribunal and penalties. One of the acts enacted by Bill C‑27 establishes a tribunal to process complaints under litigation when it comes to the use of private data. In case of non-compliance, the legislation provides for heavy penalties of up to 3% of a multinational's gross global revenue. There are provisions that are more in favour of citizens when a company misuses digital data.
Yes, this bill does have its weaknesses. I believe those weaknesses can be addressed in committee, but they may require the introduction of new legislative measures. Public services, however, are not covered by this bill. Data in the public sector requires a greater degree of protection; this bill covers only the private sector. Take, for example, CERB fraud and the CRA. In 2020, hackers fraudulently claimed $2,000 monthly payments and altered the direct deposit information for nearly 13,000 accounts.
The government can do more to tackle fraud. Unfortunately, this bill offers no relief or recourse to those whose information has already been compromised. There are digital records of nearly every important detail about our lives—financial, medical and education information, for example—all of which are easy targets for those who want to take advantage. It has been this way for a while, and it is only going to get worse when quantum computers arrive in the very near future.
This means that we must find and develop better means of online identity verification. We must have more rigorous methods, whether we are changing our requirements for passwords, for biometrics or for voice recognition.
Recently, at the sectoral committee, we heard about how easy it is for fraudsters to call telecommunication centres and pass themselves off as someone else to access their information. We must improve identity verification methods, and we must find a way to help those who are already victims of fraud. We must do so by amending Bill C-27 or introducing an additional legislative measure.
Since this is a fairly complex bill, it will be referred to the Standing Committee on Industry and Technology, where we will have the opportunity to hear from experts in the field. At this step, I would like to recognize the leadership of the Minister of Innovation, Science and Industry and his team. We have been reassured by the answers we have received.
Since Quebec already has data protection legislation—Bill 64, which became law 25—we want to understand when the federal act will apply and whether the changes we requested to Bill C-11, introduced in the previous Parliament, were incorporated into this bill. I want to say that we are satisfied with the answers we have received so far.
We will do our due diligence because this bill includes a number of amendments. Obviously, the devil is in the details. During the technical briefings held by the department since Bill C-27 was published, we asked how much time businesses would have to adjust their ways of doing things and comply with the legislation.
We expect that there will be a significant transition period between the time when Bill C-27 is passed and when it comes into force. Since the bill provides for a lot more penalties, the government will likely hold consultations and hearings to get input from stakeholders.
In closing, I would like to say that I have just come back from Tokyo, where I accompanied the Minister of Innovation, Science and Industry to the Global Partnership on Artificial Intelligence Summit, where Quebec and France took the lead. The first summit was held in 2020. I would like to list some important values that were mentioned at this summit that deserve consideration and action: responsible development, ethics, the fight against misinformation and propaganda, trust, education, control, consent, transparency, portability, interoperability, strict enforcement and accountability. These are all values that must accompany open data and ecosystems.