MiQ Unlocked: Artificial and Human Intelligence - by Tess Harkin

An important part of digital advocacy is learning how to build and target the audiences you want to receive your message. In recent years, access to available personal data that could be used in crafting strategies has grown exponentially. However, with this growth in readily available data comes an added burden of best using this novel information in order to derive useful insights and create meaningful action. Artificial intelligence can play a role in the digestion of data in order to provide substantial observations into how people interact with digital media, and the influence of advocacy and awareness based campaigns. However, human intelligence is still a crucial part of driving impactful online efforts.

On Tuesday, I attended an event called “MiQ Unlocked: Artificial and Human Intelligence.” The event was hosted by MediaIQ, a global analytics technology company specializing in programmatic digital advertising efforts. I’ve worked with MediaIQ in my job as a media planner to successfully implement digital media campaigns. The event featured two speakers, with lengthy backgrounds in media analytics.

The speakers presented the idea of thinking about the role that machine learning plays in evaluating the success of campaigns as similar to the concept of linear regression, meaning that it is a predictive analysis that evaluates relationships between variables. MediaIQ can use their data to build models and predict future outcomes based on past behaviors. The important thing to know is which variables might affect conversion rates in online campaigns. As an example, if you were running an ad trying to convince people to purchase a car, things like gender most likely would not affect conversion rates. However, whether or not someone how children, suburban vs. urban living, and income would. Once you can determine what your variables are, you’re able to adjust your equation or algorithms to reflect goals from past findings. Machine learning can use this information to help design audiences and other strategies that will influence campaigns by closing the gap between large amounts of data, and how much of it can actually be analyzed.

However, machine learning doesn’t fundamentally understand data. It cannot be proactive, as it is always pulling from and looking at past data. Additionally, there is no way to transfer learning ability and reuse models, meaning that new models will constantly have to be developed and updated. Above all else, machine learning cannot tell a story.

This is where human intelligence can play a role. Humans have the ability to apply logic and reason to data. Whereas artificial intelligence is assumptive, human intelligence is informed, predictive and outcome based. The panelists argued that there are three key tiers when working with data: infrastructure, analysis, and reporting. In order to achieve success at each level, human and artificial intelligence must work hand in hand.

The type of research MediaIQ is doing relates to what we’ve spoken about in class with regards to the role that technology can play in digital campaigns. We’ve discussed tools to define and catalog populations, vessels for transmitting information, and tools for citizen interaction and engagement, as well as innovative ways to use old technology. Artificial intelligence can play a role in each utilization of technology for election campaigns mentioned above. When it comes to defining and cataloging populations, artificial intelligence can run analyses to determine which people are best to reach out. Using voter file data, coupled with either IP/cookie pixels and targeting, machine learning can determine look-alike audiences to compliment first-party targeting efforts. This means that establishing which people are most likely to vote/volunteer/donate can be simplified. Additionally, using patterns established by artificial intelligence analysis can help to determine what the best method of communicating information is by answering the question, how do specific groups of people interact with different mediums?

Additionally, machine learning presents other potential when it comes to evaluating success of online advocacy campaigns. In the reading, “Assessing Success in Internet Campaigning,” by Yana Breindl, she presents three standards for evaluating a campaign’s accomplishments: awareness, credibility, and changes. Machine intelligence, as discussed in the panel, would be most applicable to assessing patterns in cultivating awareness. Using regressions established by prior research as to best ways to cultivate metrics to determine awareness.

Discussing the role that artificial intelligence and machine learning play in manipulating data is especially crucial now, as we begin to come to terms with the sheer magnitude of personal data readily distributed to digital marketers. Primarily, how vulnerable is artificial intelligence in the management of these mass amounts of personal data? As previously mentioned, artificial intelligence is unable to be proactive, thus exposing itself to weaknesses for intentional manipulation. Additionally, if artificial intelligence continues to evolve in ways that can create even more specific and powerful targeting, where does the line get drawn for the scope of personal inventory?

Artificial intelligence has been established as powerful tool, and certainly the direction that data analyses are heading in. However, this tool which presents gains in one aspect of digital advocacy poses several threats and issues to still be addressed in order to assure the validity of machine learning moving forward.