Human Resources Committee on Nov. 22nd, 2023

[Yana Lukasheh]

Absolutely. The Disability Inclusion Business Council that was recently formed takes a bit of that perspective as to how we ensure that a lot of what the business community across the board is using—the design of their offices all the way to the IT and the software that they leverage—is accessible. That component of the study could partly be taken with that in mind.

From a regulatory perspective, for example, the federal government in Canada has an accessibility act, which a lot of the providers have to abide by. That is one way we adhere to it.

Regarding how regulations are evolving, I think that's where we can take the conversation. Having an advisory board look at the different and evolving ways that AI technology can play in that role, I think is a very valid conversation and one that we should probably be taking a deeper look at.

[Bonita Zarrillo, NDP, Port Moody—Coquitlam, BC]

Thank you so much.

[The Chair, Liberal, Bobby Morrissey]

Thank you, Ms. Zarrillo.

Ms. Gray, you have five minutes.

[Tracy Gray, Conservative, Kelowna—Lake Country, BC]

Thank you, Mr. Chair, and thank you to all of our witnesses for being here.

My first question is for Mr. Kiron.

You stated in an article you co-wrote that “These analytic systems, which we call smart KPIs, can learn, and learn to self-improve, with and without human intervention.”

Do you believe that due to this, AI would be able to collect private data? If so, are there gaps in privacy legislation that you would recommend the government amend or implement?

[Dr. David Kiron]

That's a fascinating question.

On the whole issue of KPIs and acquiring private data to improve KPIs and help them learn, to the extent that businesses use private customer data and that's part of their datasets, there's definitely regulation that constrains how businesses can use that personal data outside of the organization.

Within the organization, there's obviously.... You can't see social security numbers outside of HR. So with the fact that KPI data is being used to help train new KPIs or better KPIs, and the KPIs themselves can learn from this data, it could be limited to whatever is appropriate within the enterprise's uses of the data, if that makes sense.

[Tracy Gray, Conservative, Kelowna—Lake Country, BC]

Great. Thank you.

I have just a quick question. Do you think the development of AI will pose risks to someone's privacy and intellectual property?

[Dr. David Kiron]

Oh, yes, and it already has.

The large language models, for example, have been trained on datasets that include published works by writers around the world. I think there's a class action suit going on with writers like Stephen King saying, “Look, your tool that you're making billions of dollars from—you have like a $90-billion capital valuation—is piggybacking on my work and it's completely uncompensated.”

There's that kind of rip-off of intellectual property—absolutely.

Then, in terms of privacy, there are so many different ways that AI is going to interfere with people's privacy. If you just take generative AI, we've talked about ChatGPT. There's Claude, and Bard from Google. There are all of these LLMs that are out there.

These companies are trying to stay ahead of the issue by putting in guardrails that are ethical and all that, but what we haven't talked about is that there is going to be a grey market for large language models that are free of these constraints that governments and companies in the public eye are focused on. What do you do with that?

Ms. Chabot, it's very hard to do a moratorium on that kind of thing.

[Tracy Gray, Conservative, Kelowna—Lake Country, BC]

Thank you very much. Thank you for that explanation. It was very helpful.

Mr. Chair, I would like to move in a different direction here for a moment and pause.

I would like to move a motion. This has been circulated to the committee. I'll just read the motion here:

Given that,

the Auditor General of Canada recently issued a scathing report on the Liberal Government’s Benefits Delivery Modernization programme, identifying delays, cost overruns and concerns on the viability of increasingly outdated technology;

this project was budgeted for $1.75 billion when launched in 2017 but has nearly doubled in cost, to $3.4 billion;

new reports from ESDC projects a revised cost estimate of $8 billion marking a 357% increase from the original price tag;

the completion date for the project has been pushed to 2034;

That the committee undertake a study of no less than four (4) meetings to review the government’s Benefits Delivery Modernization program and the Auditor General of Canada's report on this matter and that, the Auditor General of Canada, the Minister of Citizen’s Services, the Minister of Employment, Workforce Development and Official Languages, the President of the Treasury Board, and all relevant officials from these departments be invited to appear before the committee on this matter for two hours each; and that the committee report its findings and recommendations to the House.

Mr. Chair, just to put this into perspective, the benefits delivery modernization programme is the largest IT project ever taken on by the Canadian government. It was projected, as I said, to cost $1.75 billion. According to reports, it's now projected to cost an estimated $8 billion.

Costs have gone up. Expensive consultants have been hired. Timelines are extended. Liberal ministers need to answer questions to be held accountable for this chronic pattern of lack of oversight and mismanagement with yet another IT project. As an example, the ArriveCAN app project didn't work. It cost taxpayers $54 million and is now under criminal investigation. The Liberals recently paid over $600,000 to consultants to advise on how to reduce spending on consultants.

The government does not deserve the benefit of the doubt here. This is a massive spending project of taxpayer dollars. This human resources committee needs to scrutinize this.

I hope to have support of all members of this committee.

Thank you, Mr. Chair.

[The Chair, Liberal, Bobby Morrissey]

Thank you.

For the benefit of witnesses appearing, this is a normal process. A member can use their time to introduce a motion. We suspend the interaction with witnesses while we're doing this. We have now have the floor.

The clerk has advised me that the motion is in order and to be moved today. It's now open for discussion.

I have Mr. Fragiskatos, on the motion of Ms. Gray.

[Peter Fragiskatos, Liberal, London North Centre, ON]

Thank you very much, Mr. Chair.

I'm not so sure that we can count on the veracity of those numbers. I'm not saying I'm not willing to delve into these issues further, but I think we will have an opportunity to do so—not today, not in coming meetings, but when we have the supplementary estimates. I think that offers us a chance to continue this committee's focus on agenda items that we have already agreed to. I think we should resume the meeting at the earliest opportunity.

I'm happy to move to a vote.

[The Chair, Liberal, Bobby Morrissey]

Seeing no further discussion, I'll call for a recorded vote on the motion of Ms. Gray.

(Motion negatived: nays 7; yeas 4)

The motion is defeated. We'll resume to the matters before the committee.

Ms. Gray, you do have about 50 seconds.

[Tracy Gray, Conservative, Kelowna—Lake Country, BC]

Thank you, Mr. Chair.

It's really unfortunate that the members opposite have not supported this motion, considering that the Auditor General has written this committee saying that she is willing to come before the committee to discuss the very damning report that they put together—so that's unfortunate.

[The Chair, Liberal, Bobby Morrissey]

Ms. Gray—

[Tracy Gray, Conservative, Kelowna—Lake Country, BC]

Yes, Mr. Chair.

[The Chair, Liberal, Bobby Morrissey]

—the business before the committee is the witnesses appearing. The motion you moved was voted on and defeated. I would ask you to bring your comments to the agenda item currently before the committee, which is the questioning of witnesses on the AI study.

[Tracy Gray, Conservative, Kelowna—Lake Country, BC]

Mr. Chair, I think I have a few moments here, so—

[Michael Coteau, Liberal, Don Valley East, ON]

On a point of order, I think the 50 seconds is done. This is my point. The clock shouldn't stop because the member is speaking about.... The time has been exhausted, so I think we need to move on to the next speaker.

[The Chair, Liberal, Bobby Morrissey]

Thank you, Mr. Coteau.

[Tracy Gray, Conservative, Kelowna—Lake Country, BC]

Burned up my time....

[The Chair, Liberal, Bobby Morrissey]

There's no direction. When a member moves a motion in their time, we do suspend, and I allow the time left, and that is the procedure of committees, but Ms. Gray used up the rest of her 50 seconds with the discussion.

Now we will move to Mr. Kusmierczyk for five minutes.

[Irek Kusmierczyk, Liberal, Windsor—Tecumseh, ON]

Thank you, Mr. Chair.

I thank the witnesses for an excellent conversation this afternoon.

I have a question for Professor Kiron. Work rarely happens in isolation. Workers rarely work in isolation. They work on teams. I want to ask you whether you've considered how AI may impact teamwork or could impact collaboration in various settings, whether it's an office setting, a warehouse or a factory.

I'm curious if you've given teamwork and the impact of AI on teamwork some thought.

[Dr. David Kiron]

We've looked at this. We did a study with the Boston Consulting Group and a professor from Boston College, Sam Ransbotham, on this very topic.

The ways that machines and humans interact fall into different categories. I'll try to keep this as concise as possible, but you can have the machine doing.... Take decision-making. The machine makes the decision all by itself, and it's an automated thing. Take fraud detection. These AI technologies are sifting through so many parameters that no human could do it, possibly. It's making decisions about what constitutes fraud.

There are other kinds of things where the AI would contribute to a decision, but the human would have final decision-making authority over it. Similarly, the human could contribute to the AI making a final decision. Take fraud. There's another fraud instance, but it reaches a level where it's not really clear whether or not it's fraud, so the human might play a role in that kind of decision.

There's a whole spectrum, and what we found is that AI at a very high level, when humans are working with AI, emboldens and strengthens teamwork on the part of humans. Humans are more satisfied working with AI than teams not working with AI. It increases collaboration.

[Irek Kusmierczyk, Liberal, Windsor—Tecumseh, ON]

That's interesting. The reason I ask is that I'm reading a book right now by Dr. Brian Goldman on teamwork. He looks at the operating room, a complex environment where you have many surgeons, doctors and nurses operating together, and mistakes sometimes happen. Suboptimal decisions are made. I'm wondering how AI might be utilized to prevent some of those mistakes and help optimize decisions in a complex dynamic setting. I very much appreciate what you brought to the table there with your insights.

Ms. Lukasheh, I believe that today you were moderating a panel, if I'm not mistaken, with the Canadian Chamber of Commerce on AI, and you had some really interesting guests on your panel. There were folks from Microsoft and others. Were there any interesting insights? Did anything surprise you from those discussions, anything you'd like to share with us that is pertinent to our conversation?

[Yana Lukasheh]

Indeed, as co-chair of the Canadian Chamber of Commerce's Future of AI Council, we did have our first executive summit today, and it was a successful one.

We had members from all sizes of companies and from all different industries come together. We discussed AI technology as an emerging new technology, where it's going and where it's headed. We all came to a consensus that it is fast-paced. It is consistently evolving, and it is going to continue evolving in all our different sectors.

Currently, there is legislation before Parliament that looks at how to regulate AI. The conversation around whether Canada is going in the right direction, around legislating and regulating AI, is a mixed bag in terms of the sentiment around the current legislation. Overall, we can all agree on the fact that we do need some level of principles and regulations in this space.

We appreciate that different companies than are currently leveraging this AI technology are unlocking value and benefits from it. They're seeing realized and happen fairly quickly.

Looking at the productivity of AI and looking at how we in Canada can create an ecosystem that is both domestically and globally competitive was also an interesting conversation that we broached,in terms of how AI can play a factor into that.

I'll stop there, but I'm happy to speak more about it.

[The Chair, Liberal, Bobby Morrissey]

Thank you, Mr. Kusmierczyk.

Go ahead, Ms. Chabot. You have two and a half minutes.