Interset and Responsible AI - Part 2: Solidarity, Democratic Participation, and Equity

by in Security

Read the previous chapter in this series: Interset & Responsible AI – Part 1: Well-being, Autonomy, and Privacy.

Interset and Responsible AI - Part 2.pngIn my previous blogI looked at the Montréal Declaration for Responsible AI’s first three principleswell-being, autonomy, and privacyas we explore how Interset, both our technology and company as a whole, intentionally align with the values of the Declaration.

Today, I’d like to take a look at the next three principles of solidarity, democratic participation, and equity. Let’s jump right in.

Principle 4: Solidarity

“The development of AIS must be compatible with maintaining the bonds of solidarity among people and generations.”

Human relationships are at the heart of the fourth principle, which seeks to preserve and foster relationships and reduce isolation. Of particular interest to us is the second sub-principle, which states:

“AIS must be developed with the goal of collaborating with humans on complex tasks and should foster collaborative work between humans.”

A common concern around AI is the potential negative impact it could have on the workforce. What if automated technologies replace human workers? At Interset, we view AI as a partner to humans, not a replacement. Cybersecurity is different in some ways to other industries. There is a human talent shortage in our space; there are not enough people with the right experience to fill many security operations center (SOC) gaps. This is where automated technologies can help significantly. Interset’s security analytics enables the analysis of billions of data points—a task that would take even a fully staffed SOC far too much time. In this way, we foster a human-machine teaming whereby our technology does the “boring” work of analyzing data and humans are free to do the interesting, complex and important work of investigating and responding to the surfaced threat leads.  

At the G7 Conference on AI in December, Geoffrey Hinton, the ‘Godfather of Deep Learning,’ drew a helpful parallel. When ATMs were first introduced, many people feared they would replace human tellers. Of course, that never happened. In fact, we have more bank tellers now than ever before. ATMs have automated “boring” transactions like deposits and withdrawals, leaving the human tellers to interact with customers on more complex transactions and provide meaningful engagement and relationship building.

Principle 5: Democratic Participation

“AIS must meet intelligibility, justifiability, and accessibility criteria, and must be subjected to democratic scrutiny, debate, and control.”

The second sub-principle states:

“The decisions made by AIS affecting a person’s life, quality of life, or reputation should always be justifiable in a language that is understood by the people who use them or who are subjected to the consequences of their use.”

This principle explains why at Interset we selected the machine learning algorithms we did, and why we present our results the way we do. As I mentioned in my previous blog, our security analytics outputs probability-weighted risk assessments that are accompanied by natural language explanations of why we believe a behavior is unusual and should be investigated (i.e. “It was unusual for Joshua Newman to access servers 6 times in an hour; Joshua Newman typically logs in 2 times per hour”). Some machine learning techniques are relatively straightforward to generate explanations from (e.g. tree methods), while other models are much more difficult to explain (e.g. deep neural networks). In fact, there is an important area in AI research called “AI explainability,” which is actively trying to further this area. At Interset, we choose certain algorithms instead of others to ensure that our AI decisions are always explainable and transparent: for us, there has to be a balance struck between effectiveness and explainability.

Further to explainability is the third sub-principle:

“The code for algorithms, whether public or private, must always be accessible to the relevant public authorities and stakeholders for verification and control purposes.”

Openness and transparency are critical to ethical AI, and it has always been critical to Interset. We operate in a highly competitive space that does not foster a sense of collaboration for fear of losing competitive advantage. Many vendors are hesitant to speak openly about their mathematics, which they describe as their “secret sauce.” Interset is unique in that we have always been happy to describe our algorithms and math.

Principle 6: Equity

“The development and use of AIS must contribute to the creation of a just and equitable society.”

In the same vein as the fifth principle, we believe equity rests on openness and transparency. The math we are leveraging itself is not altogether new, but the way we that we have learned to effectively apply it for cyber threat detection is new. Openness fosters this type of innovation, which is why we agree with the seventh sub-principle of equity, which states:

We should support the development of commons algorithms — and of open data needed to train them — and expand their use, as a socially equitable objective.”

We are continuing to develop our product to further enable this type of transparency, and you can expect to see important, new advancements from us this year to foster an open, collaborative environment where security companies can share and exchange analytical models.

Stay tuned for my next entry in this series, where I will look at the next three principles of the Declaration: diversity, prudence, and responsibility.

Read the next chapter: Interset & Responsible AI – Part 3: Diversity Inclusion, Prudence, and Responsibility.