As data becomes more pervasive and more accessible, cybersecurity and privacy concerns continue to be on the rise as well. Can AI can continue to advance without risking highly personal and valuable information? Dan Baird of Wrench.AI shares his views.

The management of digital and online data continues to be on the rise as emerging technologies, like AI and Analytics, makes it more convenient for businesses to gather vital details and insights on their customers. In fact, by 2020, the world will have accumulated 44 zettabytes of information, according to market research firm International Data Corp.
The solution, according to Dan Baird, Founder & CEO of Wrench.ai, is to use “private” machine learning approach that allows the technology to only pull insights from data sets, while not having to access peoples' personally identifiable information. He explains how in an interview with Digital Journal.
Digital Journal: How important is digital transformation for business?
Dan Baird: It’s critical because there’s no slow down on the increasingly fast-paced evolution of digital and AI impacts to products and services, across a variety of industries.
It’s not that folks are stuck in analogue mode and need to make the switch to digital, it’s that most businesses made it a long time ago. The challenge now is keeping up with new solutions that enable businesses to maintain their competitive edge. Here’s why: according to a recent report from Forrester, AI-driven companies will take $1.2 trillion from competitors by 2020. And there’s this: By 2021, Gartner is predicting that AI technology will generate an estimated $2.9 trillion in new business value, as businesses look to use AI-powered tools to generate insights, create efficiencies, and personalize customers’ user experience. These are strong incentives to get on the AI train—but to do it strategically and deliberately.
DJ: Within this, how important are data analytics?
Baird: First, data is ubiquitous because in the broadest sense it’s information, and information is everywhere. It’s literally as important as air. You could put the phenomena of Big Data on par with the paradigm shifts that resulted from the discovery of electricity and the Industrial Revolution.
Businesses typically traffic in an overwhelming amount of data. In fact, the sheer volume of it makes it challenging to know how to use it, track it, and establish the right metrics from it. That’s why knowing what information (images, text, audio, video, websites, apps, etc.) prospective or current clients are consuming, creating, and considering can be extremely valuable.
The most astute businesses will be discerning about what data they need, and what they don’t. They’ll be able to find the data point that shows what drives customers and what alienates them. They’ll understand how to strengthen operational weaknesses to get better. However, most businesses don’t fall into this category. It’s a lot easier said than done.
DJ: To what extent does this use of data present cybersecurity concerns?
Baird:To a great extent; you need to assume your data will be compromised at some point—not if, but when.
In general, organizations have a ton of information on customers and prospects, some of which they willingly give (like email addresses and phone numbers). Companies combine that information with publicly available data to build rich profiles of the contacts in their CRMs.
None of that data should ever be made public unless customers allow it. Sure, a nefarious party can attempt to extract that private data—that’s why companies need to use good data practices to keep it secure—but business leaders have to exercise highly ethical data governance at all times.
DJ: Where are these risks coming from?
We are absolutely surrounded by data; it’s a part of every aspect of our lives. That’s why data is so valuable. Think about your day, from start to finish. Unless you’re intentionally living off the grid, it’s rare that you can’t track data activity, measure, or pay for it in some digital fashion. There are already more data collection devices on the planet than people. Yeah, we know, read that sentence again. Gartner estimated that in 2017 we’d have 8.4 connected devices, and that by 2020 that figure will hit more than 20 billion!
Each one of us is a constant data producer. With all the humans on the planet, and their devices, just think about how that adds up—not just day to day, but over time.
Consider also that those devices have unique security vulnerabilities, and that’s why they’ll always be targeted in order to get at the valuable data they contain. Even if the data was not hacked or stolen, it can still be easily lost or mishandled.
DJ: What measures can businesses take to protect their data?
Baird:They need to understand their data flows really well, and they need guidance from security experts to assess vulnerabilities, fix them, and constantly monitor for breaches. If there’s no crying in baseball, there’s definitely no room for complacency when it comes to securing data.
We’d also recommend that companies do all they can do to prevent taking on unnecessary data. Purge what’s dated, or no longer useful, and focus on the information you really care about.
DJ: How central is artificial intelligence to this?
Baird:Artificial Intelligence should be able to provide an advantage, but because no solution is perfect, organizations need to understand the risks it can pose. Believe it or not, AI can often behave counterintuitively. Behavioral economics, for instance, demonstrates several areas where human bias can cause us to choose to take or avoid risks. Having more data and an ability to review probable outcomes could result in a different decision.
Machine Learning, Deep Learning and AI provide an opportunity to logically consider a lot of data—through models—and extract useful outputs (a robot baking bread, types of marketing campaigns that increase conversion rates, etc.), but those outputs are only as good as the data on which those models are fed and trained.
That’s why it often takes a lot of data and training (this means human intervention) to even come close to making anything like a prediction. Consider that our intuition is powerful, but like AI, it is limited by the amount of information we are able to consume.
DJ: Does this wider use of data also create privacy concerns?
Baird:Yes and no. Thanks to GDPR, global companies have much more stringent regulations on how they can collect and use peoples’ data within the European Union, but similar regulations will hit the US; it’s just a matter of time. California just passed one of the toughest data privacy laws in the country. We predict other states will follow suit, and eventually we’ll see policies enacted at the federal level.
But those policies won’t necessarily guard against bad actors making attempts to get to peoples’ personal data and exploiting it, or companies here and there falling down in their responsibility to keep the data in their possession safe. Companies need to be constantly vigilant and proactive in their approach.

DJ: What measures can be taken to protect privacy?
Baird:We do what a lot of other companies do: we “hash, encode, and encrypt” personally identifiable information (“PII”), which means we don’t access contacts’ information in a client’s data set. Our models don’t use PII as a source of data on which to build predictions.
But they can still “read” the information and use it, in combination with other hashed data, to create outputs like lead scores, segmentation clusters, campaign recommendations, etc. We could get into a lot greater technical detail, but suffice to say that there are established best practices that can make storing and securing personal data relatively safe. But it’s also important to be clear that nothing is perfect.
DJ: What services does Wrech.AI offer?
Baird:We’ve built a proprietary Platform-as-a-Service (PaaS) business model and Software-as-a-Service (SaaS) products that make sense of large data sets (including video, images, and audio) to uncover data and establish key performance indicators that can be measured, tracked, and improved upon over time, using deep learning neural networks.
We have a plug-n-play product, Serendipity, for CRMs that predicts the ideal buyer’s experience, applies lead scores, recommends a messaging strategy, and personalizes and tracks marketing campaigns. We have another version that’s geared at financial service providers, VCs, and investment advisor professionals, predicting investor status and probable investor life stage and product recommendations.
Currently in beta mode, our product Parkour can automate a product launch. It incorporates a neural network that tracks marketing campaign results to get faster and more accurate over time.
We’ve also developed Executive Decision-Making-as-a-Service. Decision-makers at all levels can use AI to make faster, more informed decisions and those in a C-suite role are no exception. For enterprises that need a more custom solution, we work closely with top decision-makers to build a tool they feel comfortable using and will get more precise over time.

DJ: Which types of business do you work with?
Baird:We work with companies that need to get more out of their data, like a CRM, or need a solution to extract insights from formats like video, images, and audio. We collaborate on establishing or revising key performance indicators, and figure out what their sales funnel looks like, from start to finish so they know what works and what needs to be fixed. The goal is to automate what can be automated, optimize where possible, and increase a client’s ROI, and more than that, allow for more time where the human touch is most effective.

This article originally appeared in The Digital Journal