what if there's a way to prevent Canadians especially young people from taking their own lives by tracking their posts on social media about 4000 Canadians and their lives by suicide every year the Canadian Mental Health Association says it is the second most common cause of death for young people in this country men in their 40s and 50s in Canada have the highest suicide rate and that brings us to the what if part a company that uses artificial intelligence is being hired by the federal government and the hope is what that AI finds can anticipate and
even prevent suicides in Canada Brett Ruskin is in Halifax with the story this morning Brett we know this is a difficult time of year it certainly can be for people who struggle with mental health problems but we have been watching spikes of suicides in Canadian communities no matter the season so tell us what this company is going to do to try to reduce those numbers and good morning Heather well this company is called advanced symbolic and what it's going to do it's going to use artificial intelligence and social media to try to predict and prevent
suicides so what they do they're going to be watching four previous instances in spikes in suicide to try to predict those events that if they reoccur so one question is if you post or tweet something that is considered suicide related behavior is there going to be a mental health expert or a government official who is going to be knocking on your door to do a wellness check no that's not what's going to be happening here again this company is looking for spikes in suicide so looking at previous events identifying precursors using their AI and watching
for those red flags in other communities across the country whether it be the spike that we saw here in Cape Breton for example northern communities even University campuses so that is what the government is going to be watching for in a pilot project that launches soon Heather so we're seeing more and more Brett just how AI can be used to sift big data spot things that people may miss human eyes how precise is this company's program at predicting the things that you're talking about the social trends well this company claims to be the only social
the only marketing firm that is correctly predicted the Canadian election that US election and brexit spot-on and so they have made those predictions in the past using this technique of using AI and social media monitoring that was to assess public opinion this is more so to assess personal emotion and again it's being used in a pilot project that'll last about three months until the government decides whether it will continue with this coast-to-coast monitoring of public social media accounts whether in Ottawa this morning Kenton white who's the chief data scientist for advanced symbolic that's the AI
company the federal government has hired mr. white good morning to you and Happy New Year hi good morning happy New Year to you Heather let's begin simply but specifically tell me about the technology you're going to be using and what you're going to be doing with it thanks our key technology is that we're able to create randomized controlled samples using online sources this is equivalent to how we used to would run telephone poles where we can create large numbers of representative populations and see what they're saying and see what they're saying now see what they
were talking about many many years ago force anna de yeah go ahead so when you use that when you look through these social media posts what are you specifically going to be looking for the warning signs are gonna be watching for when it comes to suicide well AI works a bit differently than us telling the AI I telling her what she's looking for what we'd start with our past events past tragic events that have happened and our AI she reads and learns the patterns leading up to those prior events and then applies those to future
events so at this point we really don't know what it is that she's looking for she's going to find and even once she finds them we have to ask her what she found and she has to tell us as much like an interview to find out what she is looking for my goodness AI and even with with a she so there's this entity that's going to be looking through these social media posts why is that more effective than our own human eyes or humanized monitoring these same sorts of posts it's a it's a question of
sky so we have over 150,000 people that Rai she's keeping track of that she's reading and learning from and she's doing this 24 hours a day seven days a week and she's going back four or five years to learn this is a task that if we apply people to it's a little bit beyond our scope so how long before you know what she is picking up well it usually takes us about 90 days for the AI to Train I mean she learns just like a regular person well she's working with clinicians and our staff and
after about 90 days she'll start to learn and be able to pick up on these signals and then from there we can understand what it is that she's learning and what she is seen okay and then what do you hope to do with that information mean what what do you hope how do you think this could be effective in stopping suicide spikes well we don't want to be Minority Report we don't want the government coming and knocking on doors instead what we want to do is be able to bring the resources to the communities that
need them most before they need them so that we can have extra counselors or we can have extra outreach to help the communities that need it before the tragedy strikes and then when you say you don't want government knocking on doors there are going to be anytime you're talking about she and collecting data and looking at social media posts people with privacy concerns personal information concerns how do you answer those yes so we are dealing with a very large anonymous sample we use the privacy techniques at our industry standard the same kinds of techniques that
are used by Statistics Canada and all the other organizations out there that are collecting and dealing with personal information we're not micro targeting we're not reaching out and knowing anything about any one individual but knowing everything we can about a population of over 150,000 Canadians Canton white a pleasure to have you on this morning thank you very much chief data scientist advanced symbolics and more on that story coming up here
