IBM just made it easier for Cambridge Analytica v2
Taraaz wrote recently about IBM’s development of a personality insights tool that allows someone to create personality profiles using machine learning and online activity obtained from messages, social network, emails, blog articles etc. By doing so, they have made it easier for others to apply the same techniques Cambridge Analytica reportedly used to help their clients persuade people to take certain actions or hold certain opinions in their lives — such as who will they vote for come Brexit or U.S. presidential elections.
This is because the ad targeting techniques within Facebook and other social networks have not changed since the Cambridge Analytica events came to light. Anyone can still target populations of people with ads, observe engagement rates and sentiment, and adjust their message until they get the desired the result.
What was reportedly unique about Cambridge Analytica was that they obtained large amounts of personality information of US citizens enabling them to create more specific ad creative and aided their ability to refine this ad creative over time: The net result being a much more targeted persuasion campaign. All that separated Cambridge Analytica from others in their field was this personality dataset, and the financing made available by their clients. Anyone who could gain access to similar datasets, and with the money to run social ads could theoretically repeat what Cambridge Analytica did.
Which brings me back to the IBM personality tool. Theoretically, someone could use this today across public datasets to create similar personality datasets. Some obvious public datasets include Twitter, Medium, and some aspects of your LinkedIn activity. Less obvious datasets could be obtained by more motivated individuals through other techniques of social engineering, perhaps befriending people in Facebook in order to gain access to their personal information, before scraping it and using it in their profile tool.
Personality profiling on an individual basis can be a very helpful diagnostic for this individual or a team they work with. However, making this technology available at scale provides the tools to profile populations with great fidelity. This is fantastic news for digital marketers, as it provides the tools to be much more efficient in targetting customer segments, and persuading them to engage with their brand or message. But this same approach can be used by sinister actors to do just what Cambridge Analytica infamously did, offering a rich dataset of individual personalities enabling effective tools of persuasion at scale, influencing specific political views at the individual level and at the aggregate shifting the views of populations, and potentially altering the outcome of elections.
This article isn’t meant as a criticism of IBM: Once technology is available to us, containment is near impossible. If it weren’t IBM releasing this, someone would have. However as the power of these tools grows, we need to balance our “fast thinking” culture of “move fast and break things” by developing a “slow thinking” culture that helps us anticipate how these technologies may be abused and attempting to build in appropriate counter-measures from the beginning. We need a cultural shift in technology teams akin to the agile movement, recruiting the creative energy of entire technology teams and organisations to incorporate “slow thinking” into product design if we are to minimise the unintended consequences of technology, and create space for ethical considerations in technology policy and design.
