Meet Polly
The Future of Information
By Lisa Sadach
What do you think about zebras? Chances are, for most people, the honest answer is probably “not much.” However, when asked this question in a formal survey or by a market researcher, nearly all participants will dig deep to form an opinion on the spot. Because of this, traditional market research is not adept at measuring natural, authentic engagement and opinions. In short, it cannot answer the question:what would you be talking about if a survey wasn’t asking you and guiding your thoughts?
Where traditional market research methods come up lacking in the pursuit of authentic engagement, Polly shines. Polly, AI created by Ottawa based Advanced Symbolics Inc. (ASI) is being used to gather market research that reflects genuine consumer engagement and authentic consumer attitudes.
Erin Kelly, president and CEO of ASI, says Polly started with the search for perfect, unbiased information gleaned from natural human behavior. “With the advent of social media, this was the first time this became possible because it created a kind of living laboratory that allowed you to observe people without asking them questions and without disturbing the environment in any way,and while still protecting privacy. We could get a good idea of what the population really wants.”
Having an accurate sampling of a population would allow companies, governments, communities, and other interested parties to make better decisions for everyone concerned because the information would be reflective of the true opinions of the majority of that population.
Polly goes beyond market research for products and services and can answer questions about a population’s feelings towards policies, mental health, and even provide customized recommendations about what solutions are likely to work best for a given demographic based on the trials and errors of similar participants.
How Does Polly Work?
Polly is, above all else, an observant student. Her empathetic and sensitive nature allows her to take in information and learn from it and the population she is tasked with studying. This ability to learn allows her to create a representative sample of the population – one that includes enough people of a certain age, ethnicity and economic background to give an accurate view of what the attitudes are of the total population.
As Kelly explains, “two things have to happen when Polly is tasked with a problem: she must find an accurate sample of the population, and she must also be educated on whatever the topic is that we want to explore. She does that by reading, just like humans do. We give her newspaper articles and she might read Wikipedia. We have her read the articles and then we ask her, ‘what did you learn from this article?’ This ensures she has learned enough about all the key players involved in the issue.”
Polly can work in a multitude of ways, but Kelly gives a simple example of measuring engagement with a bus ad for a university. With Polly’s observation capabilities she is able to accurately find the social media posts of people who would have ridden the bus and are now talking about the ad. Though there is a chance that a passenger will post about the university because a friend or another source told them the information in addition to seeing the bus ad, this still tells marketers that this is a touch point for the message.
And Polly doesn’t stop there. Her ability to learn from the information she pulls in allows her to answer increasingly complex questions including predicting how a population will react based on previous knowledge.
The human-like way she empathizes makes her a good learner, but also a bit “quirky.” She can misinterpret opinions, sympathies, and other signals much like a person can. This is why questions can only be input by the ASI team who are familiar with the way she works.
Kelly emphasizes that Polly is commercially ready, but needs to be trained in order to understand and solve the client’s problems. Her abilities will increase the more she learns, further opening up possibilities for research and understanding various populations.
Secure Science is Good Science
With Polly’s almost omniscient quality, it’s reasonable to question if the technology is recording personal information or invading privacy. But, according to Kelly, “microtargeting” is just plain “bad science” and Polly would have no use for info on any one individual.
“It’s important to know that everything is population level because having one person say something doesn’t tell us anything.When we see 100 people suddenly have more discussion about a topic, then we know that there has been a significant impression within the population … it’s not just an outlier. If you’re microtargeting, focusing on a single post or impression, you’re going to be wrong, you’re going to have a huge margin of error.It’s really not significant unless we see a number of people doing something.”
Kelly stresses too that microtargeting is unethical and that ASI makes protecting personally identifiable information a top priority. Doing so is in line with the company’s overall goals and philosophies as a Canadian company structured around democratizing access to quality, unbiased information.
More than Marketing
Because of the intuitive nature of Polly, her applications go far beyond measuring consumers’ opinions on goods and services. Her ability to read and learn people’s honest likes, dislikes, and mood changes is setting Polly up to be an important part of the future of how we will manage and prevent mental health concerns.
“She understands people often better than they understand themselves,” Kelly says.“We do a lot of work in mental health because she’s very, very good at detecting changes in personality—worrisome personality quirks. In some cases, she can detect mental illness in people better than they can detect it in themselves.”
This is not about “microtargeting” or monitoring any one single individual or “outliers.” As with everything Polly does, the process is anonymized and is intended to help a larger population. For example, Polly would be able to see the fact that a thousand more people than usual are expressing grief in a certain area. This large spike could indicate a variety of tragedies or disasters like a flood, a factory closing, or a school shooting that is affecting a large percentage of the population. Polly would be able to see this sooner and alert healthcare professionals to have additional counselors on the ground to provide support.
The Future is Alien
Polly, who originally got her name from the first study she did regarding opinions on the ”poli- tics” of Brexit, has used her abilities to become a team member to her ASI coworkers. “In our company, AI doesn’t stand for artificial intelligence, it’s alien intelligence. For us, she has intelligence that is different from our intelligence, but it is not artificial,” says Kelly, who notes that Polly, like all other AI in her experience, is here to help people be better, more productive and healthier.
“In keeping with our origins as a Canadian company, we see Polly as an information peacekeeper. On our team, we often see studies are used to prop up agendas…So, information becomes a bit of a weapon. We want to demilitarize the information world.”
As part of this effort, the company is developing a public interface called Ask Polly that is set to be available in 2020. It will allow people to fact check broad and subjective statements. For example, if a politician states they are seen as the greatest in their field, Ask Polly can answer if the public truly shares that sentiment.
“Ask Polly is an unbiased observer. There is no agenda there. We want people to be able to use Polly as a source that they trust, and a source that they can always go to for an unbiased observation of people. At the end of the day, that’s what we’re trying to do – make accurate information available to everyone. That’s the goal of the company: to provide perfect information that allows people to make the right decisions for the majority of the population…because they can trust the information they’re getting.”
Audio Version
Female Full Audio Magazine
Female Voice Audio