Spread over two days, Big Data TO and AI TO hosted thousands of attendees and nearly one hundred talks covering all possible elements of big data and artificial intelligence. A common vein throughout the talks, whether focused on the business side, infrastructure side, or user side of these technologies, was the importance of collaboration and building trust with stakeholders.
Highlighting a few of the talks from the event, speakers brought their insights from a variety of experiences, education, and research to discuss how these topics permeate our lives today – and how they will further impact human life in the future.
Privacy and user security: the onus should be on the business
“Many people don’t read the terms and conditions of websites; companies should take on the responsibility of disclosing what you’re using that data for,” said Charlene Cieslik, Chief Anti Money Laundering Officer and Chief Privacy Officer at local cryptocurrency startup Coinsquare.
Continuing her commentary, Cieslik said she worries marketers want every possible piece of data, which opens up companies to data breaches. Instead, she recommends partnering with compliance professionals as business partners, working with them to identify the easiest – and safest – ways of identifying what data teams need to do their jobs well and provide high quality service.
Agreeing with her sentiment, Dean Dolan, the general counsel and Chief Privacy Officer for Staples, added that customers should be given communications preferences and disclosures about what their data is used for. He shared an example of an angry customer who, he said, gave consent to be emailed and have his data used on email but didn’t consent to any other communication. His anger stemmed from the fact that the organization sent him physical mail, which to him was a violation of his privacy and consent.
Dolan noted that these examples are not uncommon, which is the problem – if companies are going to collect data for a stated purpose, they have to respect and honour that purpose.
Artificial intelligence should be proven trustworthy
“Explanations are for understanding, not for [building] trust,” said Peter Hall, senior director for data science at H2O.ai, stating that AI should be proven to be trustworthy, not just explained for what it does.
Getting to a trustworthy state for AI means identifying and rectifying any bias in the technology. As Hall said to the audience, there’s more to trusting AI than simply knowing that it can analyze someone’s performance – you also need to know that its analysis mechanisms are not biased against certain people, for example.
Andrew Graham, co-founder of FinTech firm Borrowell, spoke on Day 2 of the event and came with a similar refrain. Instead of talking about AI in the abstract, though, he discussed how to leverage AI and big data for personalized financial advice.
Borrowell’s goal is to democratize financial advice and financial services, and one way the company hopes to do that is with AI. By building an AI-powered financial coach, for example, the company is able to provide the kind of personalized financial advice that only wealthy people typically have access to, said Graham.
Financial advice is a unique industry right now, as Graham noted that unemployment is at historic lows and yet financial stress is the leading cause of stress in both Americans and Canadians. The problems causing this, he said, are relatively low income growth, increasing pressures to consume more, and an explosion in high-interest consumer credit.
For someone to get out of a debt spiral or make the right financial decisions for them, though, requires a lot of data. Because of technology’s ability to scale advice out easily, trusting the technology is paramount. At Borrowell, this is top of mind and the team uses data analysis to ensure they are not only finding good recommendations, but showing them in the most impactful order.
“When you have a lot of something, how you order that something matters a lot,” Graham said, explaining how Borrowell analysis shows that many users only look at the first page of recommendations, so the company has a huge responsibility to put the best possible recommendations on the first page.
Getting to trusted data technologies
Aaron Swanson, VP of Talend Cloud, offered some reprieve to the calls of building trust: he shared a three step framework.
First, he said, you have to discover and clean your data. This will help take data from disparate systems and bring it into a common system of record, reducing duplicate work and improving trustworthiness of the data because it’s the single source. After that, systems need to organize and empower their users to keep data up to date and usable, reducing analytical errors that so often cause distrust from end-users. Finally, if the system automates and enables the user, they will be in more control of their data and how its used.
Swanson spoke in the context of enterprise technologies, but the principles apply to end-user focused data technologies as well. Getting to trust is not about having efficient data, but about offering users control and empowering them. Do that and you do data right.