Page 19 - CCFA Journal - Third Issue

CCFF2020/加拿大中国金融论坛 2020
主持人：谢谢！Touyz 博士，对此您的观点是什么呢？
Touyz 博士：这个问题很有意思，在金融科技或者任何一
个存在数据的地方，隐私总是一个让人担心的问题。我觉
得现在的领域尤其是网络生成，你可以设置一个系统去模
拟你可能会有的数据类型，这些数据有相同的类型的统计
特征，在此基础上去构建模型，而不用去担心潜在的隐私
问题。我认为这是一个理想状态，也是一种防范此类攻击
的措施，同时也确保了我们的客户隐私。
主持人：好的，谢谢你。卓女士呢？您怎么看在聊天机器
人的隐私问题？
卓女士：有两种不同的情况。当客户打开网站时就会有一
个问题，他们可能不会提供个人信息，一个普遍性的问题
是怎么定义公共信息。除此之外，一旦客户登录，就等同
于同意与我们共享他们的数据，这样我们就能更好的满足
客户需求。从聊天机器人的角度来看，他们会征询客户的
同意去完成一些特定交易。客人是从机器人而不是人工获
得帮助。我们不给机器人起真人的名字而称之为虚拟聊天
的目的就是为了让客户知道，这个交易的过程中我们是无
法接触到任何数据的，除非涉及欺诈相关的特殊调查。这
种情况下，需要得到客户的特别许可，同时我们遵循非常
严格的程序去获取数据。
主持人：好的，谢谢你的回答。由于时间有限，我想问卓
女士最后一个关于加拿大未来的数据科学人才的问题。
卓女士：就目前的研究领域来说，加拿大在这方面处于领
先地位，我们在蒙特利尔和阿尔伯塔都有研究机构，我们
也在向美国或其他国家输出人才，包括很多领先科技公司
的首席数据科学家比如杰夫•辛顿博士的博士学生。但在
接收方面，我们是落后者，这也许是因为文化原因。不过
我说的金融业，而不谈论其他行业。
主持人：感谢所有今天参加活动的人！
CCFF2020
CCFA JOURNAL OF FINANCE
加中金融
Moderator: Thank you. Dr. Touyz, what’s your view?
Dr. Touyz: I think it's a fascinating question I think within
FinTech or any place where you have data, privacy is always
going to be a concern. I think that what's really exciting within
the current field is that especially with generative networks, you
can now set up systems where you can emulate the type of data
that you get so that it still has the same type of statistical
properties on which you can build models without worrying
about the underlying privacy. So I can see that being as a
destination state, being a safeguard against these types of
attacks in the future and also ensuring that there really is a
lockdown in terms of privacy to our customers.
Moderator: Right! Thanks for that. Ms. Zhuo? Especially on
the chat bot.
Ms. Zhuo: I guess there are two different scenarios here. There's
a general question when customer comes onto the site. They
may not give us personal information. It is just a general
question which is considered public information. So beyond
that, once the customer signs in, they essentially give the
consent to share their data with us so that we can better serve
their needs. From the chat bot perspective, they actually ask you
to consent on the certain transactions and get the assistance
from the bot instead of real agents doing it for you. So
customers need to consent for these transactions. We purposely
do not name the bot like a real person's name, and we call it the
virtual chat. So that means that we want to let the customer
know that it is all about helping them complete the transaction.
None of my team members have access to any of the data unless
it's a special case of fraud related investigation that requires our
attention. In this case there would be special permission
required and it follows very tight procedures to gain access to
the data.
Moderator Okay, thank you for the answers. Because of
time constraints, let me just ask the last question to Ms.
Zhuo with regards to the future data science talent in
Canada.
Ms. Zhuo: So, if you look at the research field, Canada is a
leader, with institutions in Montreal and in Alberta. There are
leading the pack, but Canadians are exporting the talents to
other countries like the U.S. or other countries, such as the many
chief data scientists in many of the of the leading technology
companies who were Dr. Geoff Hinton's PhD students. We are
the laggards in terms of adoption. Perhaps this is because our
culture, but I can only speak about financial industry and cannot
speak about other industries.
Moderator: Thank you to all participants today!
MARCH 2021 Page 19