Description: How can the often dreaded commute be transformed into a more efficient, productive experience that simultaneously tackles the climate crisis? The answer may lie with smart commuting technologies enabled by AI and machine learning. This episode, we speak with Corey Tucker, Director of Innovation at RideAmigos, about AI and smart commuting.
Corey explains how AI can be harnessed to track daily habits and provide users with optimized personal commuting strategies, many of which utilize shared transportation to reduce carbon emissions. Deep and Corey also touch on how the COVID-19 pandemic has changed the approach to solving commuting problems and highlighted the need for flexibility of work hours and commuting options. As we learn, improvements in AI-powered smart commuting technology and increased data homogeneity promise radical, positive change in the way that we approach commuting.
Listen on your preferred platform: https://podcast.xyonix.com/1785161/11077335-ai-and-smart-commuting-with-corey-tucker
[Automated Transcript]
Deep: Hi there. I'm Deep Dhillon. Welcome to your AI injection. The podcast where we discuss state of the art techniques and artificial intelligence with a focus on how these capabilities are used to transform organizations, making them more efficient, impactful, and successful.
Welcome back to your AI injection. This week, we'll be discussing the role of AI in smart commuting with Corey Tucker Corey received a BA in civil engineering and public policy from Carnegie Mellon university and an Ms. In civil engineering from MIT. She's now the director of innovation at ride Amigos, an online platform seeking to redesign commuting systems to be more sustainable and improve the commuting experience.
All right. So Corey, start us off by telling us a little bit about yourself and how you got attracted to this idea of the commuting experience and trying to improve it.
Corey: Sure. Yeah, absolutely. Um, it was actually sort of accidental. I'm a big biker and I raced bikes when I was in grad school. I was doing completely different work.
Deep: Like what kind of bike? Sorry, I'm a bike nerd. So we're gonna have to still interrupt you right there.
Corey: I started racing road bikes. Okay. I was a grad student at MIT and for those who don't know, and this will surprise everyone. Uh, MIT has an excellent cycling team.
Deep: That I didn't know. I lived in Cambridge for a number of years, but I did not.
Corey: I know it's mind blowing, like national champions out the Wazo, um, like very, very talented.
Okay. And, uh, I started road biking. I did a little bit of cycle cross racing. Nowadays I live in Oregon, so I do a lot more mountain biking. It's just bikes, bikes are a thing for me. But while I was at grad school, I was doing actually environmental, um, work on mercury poisoning. So that was what I was studying as a grad student.
Mm-hmm um, I got, I got into bikes and then I get hit by car several times. Ooh. Um, and that's not just like a bad thing. It's just like a. Shit, we should do something about this, right? Like this is a problem. Um, let me see.
Deep: Anything serious? Were these like?
Corey: No, I got, I got doored a couple times and I got, uh, turned into not, no, no like high speed, you know, no high speed incidents or anything, but, um, I was sort of, not particularly that interested.
I wasn't having a great time. Um, in grad school, I wasn't enjoying what I was researching and my background was in civil engineering. So I'd done some transportation work in the past. So I went, I just basically walked into the civil engineering department one day and was like, Hey. Are you guys doing anything related to bikes related to transportation?
Could I come over here and could I basically change the direction on my grad, my grad school? Yeah. And they just so happened to have, uh, a project that was funded by the federal highway administration at the time. Uh, that was, um, looking at, you know, can we influence basically people's behavior around how they commute by doing a couple different things.
There was a parking component to it. There was a, an incentive component to it. Um, and a technology C. and I sort of jumped out the opportunity and, uh, and headed over there and that's, so I sort of stumbled into this particular space. Mm-hmm um, and, and as much as I care about bike safety and other things, this feels like a good, a good place to be in it's peripheral or it's tangential, I guess.
To that work. And certainly the less cars we have driving, the less incidents potentially we have with bikes . So it, it felt like a kind of an easy transition to make. Um, and that sort of started it. And, uh, I worked with my company that I work for now as a grad student. Okay. Um, and so they sort of pluck me out of grad school when I was done, kind of said, come, you can choose your own path.
Um, come, come work with us and you have expertise that we need. And that's how I got to. Awesome.
Deep: And so, and tell us a little bit about, like, what does your company do? Like what does it have to do with smart commuting and what does that mean?
Corey: Yeah, absolutely. Um, so my company, uh, makes software mm-hmm , um, and the software we make two.
Pieces of software now, um, one is, uh, a pretty robust, kind of like large scale ride matching commute management service or platform. And that one is, was driven by city and state government. So for the audience like city and state governments often have a mandate to provide. Certain services to their citizens.
Yeah. And, um, one of those services often is ride matching. So the ability to find a carpool match, um, sometimes those services include trying to get less people driving, so to reduce congestion, um, that can include, you know, other kind of. More niche services, like guaranteed ride home, where basically they say, you know, if you were willing to carpool to work, we'll make sure that you can get home.
And so that was sort of our driving factor. Yeah.
Deep: And like here in Seattle, we have all kinds of stuff there. Like the city will actually give you a van. Yeah. Uh, that you can, you can, if you can formulate a carpool, you actually get one. I think there's like a million things, but yeah. Most cities have something going on.
Corey: Yeah. And like a carpool of. Four plus you get a, you can get access to a van and, and there's tax subsidies and all kinds of implications to that. So that was sort of where we started. Um, what I've been working on for the past couple of years is, is recognizing that while that exists and that type of product can often serve really large organizations.
Cause large organizations often operate like little cities. I mean, like talking about like big companies in the bay that have campuses, you know, they, they operate a little differently. is that, you know, smaller companies also have commuters. And, um, and so we now have also have an app that's more of a turnkey turnkey solution.
So it's both to support commuters. One is sort of run by an entity, right? The government or the, or the company runs it. They use it at like kind of like Salesforce or other large scale software. They like have to implement things and U and utilize it to get benefits out of it. And the other is more of a turnkey app that we are, uh, kind of leveraging into smaller.
Business spaces, because at the end of the day, a 20 person company still has commuters. They still pay for parking. They still do all that kind of stuff. Um, so they don't, they need not be left out in the cold.
Deep: And is the, is the core of the app, find people to ride to commute with
Corey: the core of the app actually is, uh, is sort of, yes, I guess the core of the app is that, but it's also a, the origination of the app was around this idea of can we personalize someone's commute.
So, you know, every. Apps, um, other services like Netflix, Amazon tell you what to do on a day to day basis. They tell you, you know, what to watch, what to buy. Can we actually suggest to you how you should commute and then help you by finding people for you to commute with. And we do that. We've actually adopted that in a Mo diagnostic kind of approach.
So it's not just carpooling. We find people for you to bike with if you're close enough to bike or walk with, or take transit with. Um, and then. offer kind of incentives and gameification on top of that to keep the, the, the app sticky and make people engage with it a little bit more.
Deep: So where's the AI piece? Is it in the recommendation engines to try to identify like who you're compatible with? Let's just start there or are there other places as well, you know, beyond the recommendation engine?
Corey: Sure. So, um, there's a couple places where, where we leverage machine learning primarily, but, um, The recommendation engine is certainly an area, right?
So it's not just to sort of help determine who you're compatible with, but you know, the day of the fixed schedule is pretty much out the window. So it's also to sort of try to gain insights about how every, how people actually commute on a day to day and week to week basis. So if you have patterns like gym, gym, attendance on Tuesdays and Thursdays or other things, like we wanna be able to pick up on that and we.
I guess this gets to the other kind of area where, um, where we use AI is, um, we actually use passive tracking. So we'll use your, the sensors on your phone. Of course, assuming you've given us permission to do that. We use the sensors on your phone and other things to like, determine how you commuted and that's how we, we reward folks that.
Obviously, you know, has to leverage some amount of machine learning to determine even what mode you you've taken and all of that kind of stuff. But beyond that, we're trying to recognize patterns in your commute. So we're trying to recognize, do you take your kids to school on Mondays, Wednesdays, and Fridays and go to the gym Tuesdays and Thursdays.
And does that change the recommendation because. In the PA in the past or in the more rudimentary form of this, right. All we know is that you leave your home and you need to get to work and you need to do it every single day or, you know, three days a week by a certain time. And if we don't have any information about.
How you do that or how sensitive you are to weather, you know, weather changes or other things, our recommendations and our matches start to become pretty useless because like your circumstances change on a regular basis. Um, and so we're trying, or just traffic or something like Google now, right.
Deep: Is like, Hey, you know, there's a traffic jam on your commute home. That kind of thing. I imagine you probably have to deal with.
Corey: Yeah. Traffic transit delay. I mean, like anything, right. I used to live in DC. People, I hate to say this people would jump in front of the trains a lot, or, you know, like who stuff happened like on the, on the Metro and the Metro would be shut down and I'd be standing at the train station waiting.
So all kinds of all kinds of stuff.
Deep: So I imagine in addition to your own sort of direct data, you have like with respect to, you know, who's matched with whom and what that individuals are doing. It sounds to me like you have to also triangulate some external data sources. So, you know, like, uh, whatever is telling you that somebody jumped in front of a train, um, you know, like the traffic sources.
So how do you. Source that data. And how do you think about it in the context of your company?
Corey: Yeah, I mean, I think to be perfectly honest with you, this is an area that is, is actually probably one of the more difficult ones because it's data that we in some ways sort of can create for ourselves, but we do, we do access other data sources.
So traffic. Is sort of, I don't wanna say it's an easy one, but like it's sort of a straightforward one, right? Google offers that like a lot of entities out there that do, uh, mapping or routing offer real time traffic stuff and that, and that's at least a place to start. Um, you know, you have an organization like ways that histor has historically done this idea of crowdsourcing information about what's going on in the roads.
Um, that's something we obviously could also implement. So like there's, there's ways to do that. Where holes exist more often than not is in the transit space. Um, transit agencies don't have a centralized repository for data, nor do they have like a kind of like a fixed way, uh, to, to represent their data.
Deep: And so, you know, when you're like, yeah, I, I know this really well. I used to be the CTO at Sora and most of the data that you're getting there is all piped through our systems. Yeah. And we had a massive heterogeneity problem. Like the data. Super. It's all, it's all over the place. And they're, you know, like depending on the kinds of cuz like municipalities generate their own data and due to the layering in our, in our Federalist government structure, you know, you've got federal government, which whose data sets are, you know, typically quite normalized and geofenced, if you will, based on census codes or zip codes, et cetera.
But. Municipal data. I mean, some of this stuff goes back to the pioneer era in terms of like how they classify traffic crimes or, you know, and it, and like mapping it to something sensible and common, uh, is something that sort of sounds straightforward and easy, but ends up being really tedious and, and challenging.
Corey: Yeah. It's, uh, it's really difficult and all the data sources are separate, right? So like anytime we go, you know, Into a new space or a new city. It's hard sometimes to track that down. Um, especially if we start serving rural or suburban areas, like that's, you know, those might be a couple bus lines in the middle.
Yeah. Right. It's not, um, it's not sophisticated. , there's certainly not a centralized data store. And actually remarkably in some, we work in Europe quite a bit and uh, some European countries are great and some European countries are really stingy about their, like, they just don't wanna. Their data's even harder to access and it's not as well done.
And it blows my mind cause our data in the us for transit is so bad that like, it can be worse.
Deep: Yeah. I mean, I know there's a lot of companies doing things, um, you know, like selling services to cities, you know, with sensors on the buses, for example, so that the buses can be tracked and they can, you know, and, and we can sort of have like a full, real time map, but any company that sells that.
To any, anything to a city is not typically guaranteed to sell it to all the cities. And so, you know, you wind up with at your best case scenario two or three vendors, but oftentimes it can be split and fractured even further. Some of the more, some of the more endowed cities will have their own technology groups in house that will go off and try to solve these problems and then they'll wind up getting somewhere and, uh, and then you wind up with a pretty fractured landscape.
Corey: Yeah. It's um, it's interesting actually, similar. Similar problems too, with like parking vendors. Right. We work with parking vendors because like, we wanna know how occupied the parking spaces are. Especially if we're working with a company that like has a parking, you know, a parking entity, maybe they own their own parking lots.
There's so many parking vendors. So like it's not, and there's no, you know, specific way. It just becomes kind of a, kind of a mess at all times. But yeah, it's interesting. I guess that you say like, Sometimes they decide to take it in house. And I feel like oftentimes that doesn't work out that well either it just becomes messier every time.
Deep: Uh, I'm almost wondering, like, is anyone trying to force some centralization here because everyone kind of benefits, uh, from some consolidation and centralization of the data, but at the same time, everyone's trying to maintain their individual advantages from their data. So like, you know, if I think about Seattle, like we've.
Every fruit flavored bicycle, color of bike share that you could imagine has been here. It's gotten so bad that by the time you fill it out and get your app going and you write a few, then they're gone and there's a new one. And there's like, you know, we've been through probably eight or nine, then there's the scooters and the bikes, and then the moped looking things.
And, and then there's the car shares that are there one day and gone the next, I, I kind of feel like the cities need to force some sort of like, if you wanna run. Share app in our city, you've gotta plug into this common infrastructure and it's gotta adhere to these schema and, you know, and, and this sort of latency budget, et cetera.
And you can keep a little bit to be private, but mostly we just want it all like, yeah,
Corey: yeah, no, I have you seen any
Deep: kind of movement like that or, or something actually happening?
Corey: It's actually funny that you say that, cause I've said I've made this comment many times it's um, and I think you recognize one of the issues quite well, which is that.
There is a power dynamic associated, particularly when you start talking about these third party mobility providers, like, you know, bird, when you're talking about bikes, obviously like bird line, whatever Lyft, yeah. Scooters the same, but like, even with like Lyfts and Ubers and where like they don't care.
Right. They have enough money. They have enough leverage that they don't really, they don't really want to play ball in that, in that regard. And I do think that cities. At the end of the day, have the only, basically one of the only levers, which is to say like, you cannot operate in our city, if you don't.
Yeah.
Deep: I mean, like in Seattle, they're very hard about it. Like everyone has to renew once a year. Yeah. And so cuz that keeps the, the lever in the hands of the city, because there's a, there's just like an evolution of new problems. Like, I mean it used to be the bikes. Didn't even have they had no batteries or anything.
And then they were just like sprawled out all over the place. Yes. Deemed relatively useless in a hilly city like Seattle. So people started throwing 'em off the bridge or putting them like hanging 'em upside down some weird crap with those making sculptures out of 'em. Then they started putting batteries on 'em and then people started grabbing the batteries and doing stuff with the, uh, or just knocking 'em down.
Then there were scooters and then there's, you know, there's just always something going on. So they realized that they can't force compliance of their regulations unless they can hang something. Big and heavy over these vendors. And yeah, it just feels to me like there's a federal and even international role to be played here.
That there's an ecosystem of data that we really need centralized because at the end of the day, society is not trying to solve the green bike problem. Society's trying to figure out the, ultimately it's the carbon reduction problem, uh, in large part something bigger, you know? Yes. I'm curious. If you see any progress
Corey: there , there are groups, um, certainly that are doing work like this.
Um, there's a, there's a group that we have done some work with. Um, that's kind of a, a global sustainability group. It's the called the world business council for sustainable development. But at the end of the day, like these are just working groups that are making suggestions, right? Yeah. And, um, and unfortunately, I think that, I mean, like, I don't wanna comment on the status of our lovely federal government and how much of a crap shoot it is.
But like, you know, it's, it's low priority for them. So like the they're they're doing things like leveraging, you know, funds through, uh, the D O T like the federal D O T or the federal highway administration to do research projects, but they kind of don't go anywhere. Right. They go to. We're not gonna do anything with this.
So how can you guys do something with this? And so like anyone that gets granted, those, those funds, um, you know, has to say like, well, this is how I'm gonna commercialize. This is how I'm going to make this sustainable, you know, going forward into the, into the long run and the federal government will fund it for a few years and then kind of.
Kind of walk away,
Deep: Need help with computer vision, natural language processing, automated content creation, conversational understanding time series, forecasting, customer behavior analytics? Reach out to us at xyonix.com. That's X, Y O com. Maybe we can help.
Corey: I think like there isn't, uh, a clear like, understanding that this is can have an impact. I think like, yeah.
Deep: Perhaps, I mean, climate is, you know, a bigger, yeah. I mean, you mentioned like the D O T and I think that's like a good analogy, cuz the like in the, in the crime world, the FBI. Uh, has a lot of levers and to the extent that we have normalized crime data, it's put out and sort of forced by the FBI, actually, this, you can, you can generalize this to any, any data that goes down to municipal county.
It's, uh, different layers of government state levels is that there exists a federal agency that had some kind of leverage and resources to force some nationalization like the national education, uh, administration puts. Probably the best national data sets around school, school systems. Yeah. Um, like, you know, every school's in there, high school, middle elementary, uh, and then they're all like geo fenced within various census boundaries.
I'm thinking like that this is really pretty much the DOT's, uh, arena. And the question is, can they incentivize all these various layers of government? To like force something like
Corey: this to happen. I mean, there, I guess like what I'm curious about and I'd be interested, like your thoughts on this, especially like, as you're referencing the FBI or like, um, the education, you know, is like, the FBI makes sense to me, right?
Because if they get centralized data, like they can do something, they potentially get to do something with it. The D O T is unlikely to leverage this data themselves. So they'd be doing, they'd be making basically state cities, municipalities, whatever. Do some, and I guess private, you know, micro mobility providers, et cetera, do something that doesn't necessarily serve the D O T like they don't they.
Deep: Well, I would push back on that, like, cuz if you think, if you think about like the NEA, you know, with respect to like education as an example yeah.
They have to report up. Yeah. So every administration has an education policy and they have to, and they have metrics that they get attract bureau of economic analysis, same thing like their, their basic first and foremost tasked with. Measuring the state of the nation for a thing, education for the, uh, you know, for the economy, et cetera.
And then that data is typically the most respected data in the country for the thing. And then at a minimum, you know, administrations. Have to talk about how they're performing relative, you know, it's cuz that's where those numbers that wind up in the news that we talk about come from and you know, research is benchmarked against these numbers.
So I would argue that D O T like some, like if the Biden administration said, look D O T you have to tell us, like, if they painted a specific vision here saying we want all transportation, whether it's public private or sort of public private partnerships, we need all that data. Reflected. And then they enumerated what it was.
And this is probably the hard part. And they said, we withhold traffic by, you know, like, uh, um, you know, traffic related funds. If you don't do these things or have a plan to get to them, there is precedence for it. Like, you know, it can happen. It's just, it requires political will, which you know exactly.
Corey: Yeah. I guess like, that's, that's sort of like, what I'm wondering is like, um, 100% they could do that. I'm not disagree. I don't disagree with you at all. I'm thinking like,
Deep: You know, where does the will come from? Or so, yeah,
Corey: like, like the public, I, I guess debatably, right? Yeah. Hears about the state of education in the United States.
Like we have kids, we want our kids to be educated, et cetera. Um, other than, and like this it's comical, cuz it's sort of, I guess, wraps back around to like whether or not we can have an influence on computer. Other than like my drive into the office every single day. If you, if I read like the traffic in LA is terrible and these are the metrics, or like, this is, you know, the data about micro mobility.
Like I care cause I'm in the industry, but like as a normal human, I'm just not sure that we care. Um, if we can't get ourselves to care about like, Unless we frame it in sustainability terms. And even then I'm skeptical because we don't seem to care that much about, you know, pollution from other sources or landfills or, you know, any number of well it's, it is not easy, right?
Deep: Cause on the left and right. You have very different. Sort of belief systems around what matters to them, but I, and I think you'd have to somehow speak to them. Um, you know, you'd have to like speak in a way that made sense, like on the left, I think, you know, creating an ecosystem, um, where, you know, carbon reduction is like a core principle probably will work, um, or, you know, stands a good chance.
That's good. Yes. On the right. You have to find something that the right cares about. So like commute times might be one, you know? Yeah. That is the struggle. I mean, you know, like I remember. In a lot of the cities I was dealing with in Europe had very different orders of magnitudes of budget. Like I remember working closely with like Bristol and England and I mean, they had a couple of orders of magnitude, more money to spend on problems.
And so they, those, some of those places almost went into a different. Problem space where they got, they had so much money to spend that they didn't focus on just what government should do. They started falling into like what we would think of as private companies filling the gaps and like, and they would start to imagine like excessively fantastical scenarios in a way it happens here too, honestly.
yeah. Maybe cuz it is more fun to think about that stuff yeah. Than it is to talk about like the mundane, you know?
Corey: Yeah. And there is more money I think there. Um, especially after, I don't wanna say after COVID cuz like every time anyone says that it makes me wanna like yell it's not exactly, we're not, there's no after, but like after the past couple of years where people are now, I guess sort of going back to work, um, the, and, and there was a pretty big outlay to the department of transportation.
Once, like, there has been a bigger outlet to D O T um, in the Budha judge Biden, you know, period, mm-hmm, there's money. Right. And, and we talk a lot to government ABC. Oh, there's big bucks. Like there that are rolling in money and they, and all of a sudden they have no idea what to do. Right. So like, there is sort of now this, this, this disconnect where like, they.
Deep: Have more money than they know what to do with, and, and they have this kind of like naive way of spending it. I mean, I feel like it's very 1972. It's like, oh, we're gonna have shovel ready projects. And we're just gonna start building roads and bridges and, you know, fixing roads and bridges. And I'm not saying that those are bad things to do.
We may need to fix roads and bridges, you know, but it just feels very. Like not reflective of the reality of all of this energy in the private sector. That's trying to address the last mile commute problems. That's trying to do it in, in ways that, you know, that shake up the usual political dynamic. But I feel like the political conversation doesn't really keep up with the reality of who's actually trying to solve these problems.
And then there's just gaps because the private sector just doesn't wanna deal with the non-profit. Parts of the space.
Corey: Yeah, absolutely. I mean, I think I will say like we do have crumbling infrastructure, so like there is something to be said for like, yeah, we have to fix some of it. But you know, one of the things that I think about a lot is the industry that, that I operate in is this industry that is generally called TDM it's transportation, demand management, mm-hmm
Um, but a lot of people in my space and outside of it have started to really just sort. Contextualize that as meaning, like, we want less cars, right. We want less people driving. And like, of course we want less cars, but we do have a certain supply of infrastructure. Right. And so like for me, when I think about actually what out, where the outlets of carbon come from and other things and sustainability as a whole, like, One of my goals also is just like, let's not build more infrastructure if we can avoid it.
Right. Maybe we build more, more transit or other things, but like, I don't wanna build more roads just to like, address that problem of demand. Right. So it's not just for me, like, can we or widening them.
Deep: And we're kind of in Seattle here, we're doing the opposite. Like we're narrowing roads. Yeah. Don't cases.
Corey: So if we could use the roads more optimally, we'd be a little bit better off. Right. So. We have a certain amount of supply of infrastructure. Um, if we could optimize our use of it, I think that's a good step that doesn't necessarily make it feel like we're just sort of shoving this idea of don't drive down everyone's throats.
Um, it could be that we spread it out a little bit and we're not as hard on the roads and we don't have as much congestion. There's so many externalities associated with how people get around. Um, and there's a lot of externalities with the work to do to widen roads or build new roads or, you know, build new bridges, um, that, you know, if we could avoid that as a first step, it's kind of like, what I think about a lot is like, how can we spread people out around, like over the modes that they could be using, not just be like, don't use a car at all costs, because that really is, is a turn off to a lot of people.
Deep: Yeah. And it's also a bit of a red herring. Like not all cars are created equal. I mean, do I really care if a car is full. And electric and getting power from a green source. Like, I don't know that I care at that point. I mean, there's just like multiple factors. There's like the, the environmental carbon footprint issue.
There's the social equity issues of just having mobility and having access to it. So. Changing the topic back to AI a little bit. Oh yeah. Sorry. so, so let's go, like, thinking about AI and like the lens that we've kind of like covered in the transportation arena. Like if you had to name a handful of use cases, if you will, where some machine learning can like really make a difference, like, what are those arenas look like from your vantage?
Corey: Yeah. I mean, I think the sort of lowest hanging fruit from, from my perspective is this idea of. Can we pretty accurately determine when people are using certain modes using certain like routes and obviously to a certain extent that, I mean, the route piece doesn't necessarily involve AI, but the mode certainly does.
If we wanna be sure that they did it and not have them tell us. Uh, a certain thing, right? So it's quite hard walking, biking, um, things that your, you know, your phone picks up on pretty easily are pretty easy. Um, the phone phone sensors are pretty accurate at picking up on that. The patterns are pretty, pretty common.
The way that you like spin the way that you walk, et cetera, where we really struggle is, and I'll get into why this is important for us in a minute, but where we really struggle is like understanding. Once you're in a motorized mode. Are you in a car by yourself? Are you in a carpool? Are you in a van pool?
Are you in a, like a shuttle bus or a train? Um, those things are actually quite hard to determine, and that's where, you know, really robust machine learning and, and large data sets are going to become important or are already important for us. And the reason that that from my vantage point is, is important is because, um, you know, In our space, a lot of people offer incentives.
They offer tax subsidies, they offer, um, the ability to use lanes or go through carpool, you know, or go through like different toll lanes and all kinds of stuff that has an impact on how people perceive their commute. Right. Or share rides or do other things.
Deep: Um, how are you doing that right now? I mean, are you just tapping into the mic and listening for conversations and like doing some speaker diarization or something to figure out.
Corey: We, we actually work with a third party that does pretty robust, kind of has a pretty robust mode detection. They generally are able to differentiate between like a car versus a shuttle or like something that has a moderately fixed route. Now.
Deep: Yeah, cuz you're gonna see the same route and you're, you're gonna see the same.
Corey: You're gonna go to match the route. So you might be able to get operation out of the stop, all that kind of stuff.
Deep: Yeah. Plus they move slower. Like a bus moves way slower than on car typically. Yeah.
Corey: Interestingly, we, we still run into issues with this in Europe because they have these like long distance, high speed bus routes that look much more like a car.
Deep: So it is not, yeah, they always a perfect solution and they also just think it through a little bit more like their buses can like move and yeah. You know, bus rabbit, cetera.
Corey: Um, but, but where we really struggle right now is, is carpooling. So to your point, and, and I think, you know, that's an interesting point that you bring up about, um, Mike, Mike, some other things.
We do try to limit how many sensors we access on people's devices. Because at any time you have battery well battery, but anytime you have to ask for more permissions, people are inherently skeptical. Permissions are something that we struggle with on a day to day basis. We had a meeting about it today.
People yelled at each other. It's fine. Um, but like, you know, accessing sensors is difficult. What we do now is mostly sort of trying to match roots much like a transit route buzz. But like, if we. We try to like pick up on who you're carpooling with most frequently or who you're, you know, van pulling with most frequently.
Can we actually match the route that you took with those folks? So, um, you know, oh, I see. That's less, less AI machine learning than, than, you know, we would pick up on this is motorized, right? Like there's kind of levels to this. Like we do leverage the machine learning into you're in a car and then we'd say like, okay, these are the people that you most often carpool.
We'd we'd look at like arrival times departure times, and then like, you kind of narrow down until you get, like, we think these two people probably did.
Deep: And then you look at routes and other things, and then you probably also think these two probably should cause leaving like five minutes apart and are in a couple of blocks of each other.
Corey: Yeah. I mean, one of the pieces of the, the app itself is that it matches people. Right? So like part of it. We say like we, you and I, we told the app, told you and I to carpool together and we're chatting with each other already. Like, we know that you we're chatting. We're gonna, we're gonna suspect basically if we did that, that, that you and I are carpool together.
And then we would have a reason to check our routes against each other. Um, so we are sort of, it's kind of an, you know, it's like a whittling process. It's not the entire breadth of the number of people. Like I'm not gonna drive to the other side of LA to pick. Someone it's unlikely unless they're no,
Deep: no, but yeah, like you're looking for, you know, some like radius around two people's starting point end point and time of, uh, departure and return. Yes.
Corey: And like, yeah. And also like. Our ability because we have chat built in the ability for us to monitor the chat too. Right. So like, we can actually, we can pick up on, you know, did you guys exchange chats this morning, right before the commute or maybe last night? Um, was there activity in, in your chat feed, all kinds and you little pieces of information and you can probably.
Deep: It may be not across all cases, but you can judiciously like get some explicit feedback where they just tell you yeah, that's a fellow carpool. Um, so that you can train some models off of, off of that. Do you do that sort of thing too?
Corey: Yeah, we, we're trying to sort of phase that out a little bit. In our early versions of the app, we had people sort of self-report so that we had a data set.
We, we, so that like we could confirm, then we basically had a test and train data set and we could use it to. Um, assess whether or not we were doing a good job, but, uh, we still do that periodically, but we've sort of tried to phase that out a little bit. So it is a more, we're kind of launching in a broader way.
So we don't wanna have people having to feel like they're reporting all the time.
Deep: Have data, have a hypothesis on some high value insights that if extracted automatically could transform your business? Not sure how to proceed? Bounce your ideas off one of our data scientists with a free consult! Reach out at Xyonix.com and you'll talk to an expert, not a salesperson.
Deep: Yeah. I imagine with an app like this, how people engage and what they do is probably a constant challenge.
Corey: Um, it's really hard. It's like edge cases that you could possibly never possibly expect, you know, like that's I think, and that that's the challenge, right? The challenge is. People are different people's schedules change on a day to day basis, but they're often patterns, right?
So like, it, it is not really common for, I mean, my mother maybe is a great exception. She'll love me when she listens to this later, but, you know, get up, go to work at the same place every day at the exact same time, leave work at the same time every day. And oh, now post pen pretty much go home, but that's like, it's not a thing anymore.
I mean, like people go to happy hour people. There's so many things, but. There are patterns, right? And so if we can get close enough, so what happened?
Deep: So, so I always love talking to folks about what happened to their data pre and post pandemic. So your data must have gone a little whackadoodle if you were mostly catering to, you know, Urban core office workers who are like, you know, living in dense areas.
And then all of a sudden nobody's even going to the office. And all of a sudden you must have, you know, March 20, 20 senior data just go off a cliff or something. Yeah. And so how so, so like what happened then? And like, what's happen. What's like reconnecting now and what's, what's different and new.
Corey: Yeah. I mean, March, 2020 is like the day that we were like, our job is done. We've done.
Deep: We've done our job and no, yeah. Cause everyone's staying home. So like yeah. Carbon not being admitted. Everyone's sitting at home yeah, just a beautiful,
Corey: beautiful. No, it was a terrible moment, but, um, you know, I mean, I think even to the thing that we're discussing, right? Like I, I don't work in an office now and I've been a remote worker for a while.
You know, I, when I graduated from college, I lived in DC and I went into the office every day. I took transit. Um, and even then, like, my schedule wasn't necessarily fixed, but it was way more fixed than people are experiencing now. And I think there's a couple things that we are noticing, and these are more industry trends that, that are reflected in the data.
But. Is that like, while people had flexibility in how they got to work and what they did afterwards and before bef like before the pandemic, most people had pretty fixed schedules. Like most people still, there was like this expectation that you're in the office around nine, or, you know, whatever, depending on where you worked.
And there was this expectation that you worked until. Five or six or seven, and then you went home. Right. And, and it was, and it was every pretty much every day. Yeah. Now like back to work policies are just like, they're across the, they're crazy across the board, but more often than not what they are is like, most companies are like, please come in a couple days a week.
Right? Yeah. And so what we're seeing is. And this is something that has not a lot to do with data, but sort of a little bit to do with transportation is like something that we used to really advocate for when we talked to companies is like, you need to introduce flexibility in how your people like a little bit around time, but also in like don't charge them for an annual parking permit.
Don't lock them into things you need to introduce flexibility so that people can make good decisions on a day to day basis. Everyone's reality is different on a day to day basis. You know, people have kids, people go to the gym, people do all these things. If you lock them into a certain way of getting to work, you're gonna have your parking.
Lot's gonna be full every single day, because basically you've made someone pay for a parking permit for a year and that's just bad policy. We're coming out of that where employers are sort of being forced into that, right. They're being forced into this. Like my employees are only coming in two days a week.
They hate it. We have to make it as flexible as possible. They can come in, you know, whatever time they want and go home at whatever time they want. And in a way it's sort of helped us because we designed some of our products to really try to feed into that flexibility. So. um, we are seeing people go in all at all kinds of times.
We're seeing people, you know, go in usually on a set schedule, but like, definitely like not every day, maybe Tuesday, Wednesday, Thursday, or what have you, um, more flexibility around that. I will say though, that we're, and this is, this is a common, this is everyone knows this transit ridership and other things like that have definitely dropped off, um, because people are nervous.
Um, it's obviously it's ticked back up, but it hasn't reached free pandemic levels by any stretch of the imagination. So. In the data. What we're seeing is like people driving more, but also people biking, more people walking more. Cause they got used to doing that kind of stuff during the, during the pandemic time, you know, a lot of people got outside and oh yeah.
Deep: My, my mountain bike trails are like packed now. Crazy. I've had the funniest conversations with people. Well, I used to just sit in coffee shops all the time and now I go riding. But anyway, I, so yeah, so we have this phenomenon and I have a few friends that like. You know, businesses with outdoor apparel and it's just like, that whole world has just grown.
Yeah. Terrific. To get people outside, um, in a level that they haven't before, but yeah, very. Confusing from somebody sitting around looking at data and trying to figure out what's going on.
Corey: Well, yeah, I mean, it's been, yeah, it's funny. It's like hard for, uh, those of us who are hobbyists to be like, oh, my trails are all full, but like at the same time, if I can convince the companies that we work with to leverage that into a back to work strategy, that's like, We're going to raffle off 10 by 10, really nice bikes to get you guys, you know, to, you know, whatever.
Deep: I feel like. I mean, like, I don't know, this is a bit of a side note. It's not like quite on the AI bucket, but the model to getting people into the office is just really not that complicated. Like every graduate school in the world has a simple model. There's a place you come when you want and you don't when you don't.
And as long as, um, you make it interesting and fun, and they're a reason to come, which is namely, like other people are there and wanna be there and it's quiet and it's a good place to work. Then. No problem. People show up going back to the. Machine learning thing though. You, you mentioned this idea of being able, there's like certain problems in right.
Amigos that you think are kind of challenging from machine learning, vantage, uh, probably due to like your kind of availability of the types of data you've got, but. If we step out of that particular context and think about it more from like a city's vantage, like, what do you think are the big machine learning problems there that like really matter, like what, what is a great vision that we could have and work towards?
That's sort of digestible by, you know, somebody who has, who, who doesn't do any of this stuff at a party on a Friday night.
Corey: Yeah. And I don't think it's, I don't think it's everyone or no one ever, you know, with a vehicle. I think, you know, you made a good point before, which is. Some of these entities, like Lyft, an Uber potentially like use electric cars or use other, you know, less, uh, emitting vehicles.
And I think in, in some ways that's, that's really good, but I, I do think, and I like to, I wanna just like circle back on this idea, like, um, and it's, we have the same problem. I think that cities, the city, the cities have this problem at a, at a, almost like a higher, a higher level, which is that like, They can't solve all the problems, right?
Like you and you, we talked about this earlier, like, and they don't even often solve, you know, the transit problems, but they fund a lot of these things. But at the end of the day, you know, you have your first mile, you have your middle, you have your last mile, you were getting around to a lot of different places.
And I think this concept around, like, can we, can we do a little bit more to sort of guide. People and how like, and how they make their decisions and then use all of the data. If we could, if we could aggregate all of this data from all these different entities, from the micro mobility providers, from transit, from vehicles, we could do a better job of optimizing people's experience.
So like what I'm thinking at the end of like the, the like dream, right? Yeah. It's like, You know, it could, I'm not saying it's our app, but like some, some like your watch or whatever mm-hmm is like, I know your calendar, right. I know that you deep are like gonna get up at this time, whatever. Um, you're going to go to the gym.
You're gonna go to work. You're gonna go to happy hour. You're gonna go home and like your device or your phone or whatever. Can basically guide you through your day in the most optimal way possible. So if the optimal is for you to walk to the gym, because it's down the street and then take this bus from the gym to your, your place of work.
And in some cases that is going to be, you were going to drive your car. In some cases, that's going to be, you were going to drive your car and pick someone else up. But at the end of the day, we can't do that. If we don't aggregate all of these data sources and then like the, the, so I think the, the place that cities and municipal.
Have to play in. This is one trying to enforce that kind of data, collaboration, that collaboration across all of these entities.
Deep: Yeah. And I, I would also, I mean, I would argue like if like one of the things I kind of can't stand about the bus system is like, nobody can tell me when the stupid bus is gonna get there.
And this is like an literally trivially solvable problem. Right. Duct tape a $10 Android device to the top of the bus. I mean, and problem solve. Like it's
Corey: hard problem solve on the bus regardless.
Deep: yeah. Like it's just, I mean, I've literally thought about just duct taping, like, uh, you know, an Android device on the inside of a bumper of a bus.
So I can know when the hell it's gonna
Corey: get there every day. Yeah. The more people have bad experiences using transit or using other things, the less likely they're going to do it. Right. Totally. Yeah. And so like, All of these things, add up to being a bad, like being bad experiences. And so, and they're not in Europe
Deep: or even like when I lived in Boston, this wasn't a problem because like
Corey: heavy rail,
Deep: easier, heavy rail was a problem.
But like the tea was, uh, was great. Like you go down plus your minus is five minutes, but like, you know, like taking the local train in Oslo it's plus or. A minute. Yeah. I mean, if that, I mean, it's there when it says it's gonna be there, but not in America generally outside of maybe New York, Boston, a few cities,
Corey: it's like, yeah.
Maybe, but even then it's just like, actually it's probably not. It's just that they're running on such short headways. That it's fine. Right. So like you're going at rush hour. They're running on. Three to five minute headways. And of course, like, you feel like that's fine, cuz you're not, you're not counting on a time.
You're counting on the fact that you can walk into the station and there will be a train that comes in the next few. Yes.
Deep: Yeah. Which is what I wanna do, cuz I mean, I'll be blunt. I don't wanna reach in my pocket and pull my stupid phone out to like figure out how to get somewhere. And I'm like, I wanna be an autopilot.
I just wanna go one. This has been an awesome conversation. Thank you so much for coming on. I'm gonna ask you. One final question, paint me a picture 10 years into the future, where everything goes perfectly and AI systems are able to be able to route writers to ride, like to, to ride shares. Um, there's machine learning, able to maybe figure out who can go, where, when, like what's the perfect scenario.
So for, for me and as an individual, um, or for somebody El or you as an individual where you can like wake up in the morning, you, I don't know, what do you do? You check your phone, you walk outside, you get on a thing, you get to your work, like, how is your life better? And like, what does that life look like?
Corey: I. Yeah. Um, I'm a little bit, I'm just gonna caveat this by saying that I'm a little bit of a, like a behind the times human. Like I still read books and don't have a Kindle, et cetera. So like, I don't wanna believe that we'll have like high tech glasses that you wear that tell you where to go all the time.
But, um, but I, you know what I think that looks like is you get up in the morning, you know, we have become a, you know, zero car, maybe one car per family country. You, you know, your watch or your, your smartphone tells you when to leave your ha like what time to leave your house to get to the bus, stop in time for the bus.
That will be there on time. For example, mm-hmm um, that takes you to work. Um, there's wifi on the bus. So you can get 30 minutes of work done on the bus. So maybe you get to leave 30 minutes earlier, or maybe even an hour earlier, if you're gonna do 30 minutes of like we've created a situation where your commute time is productive time.
Um, when you go to leave work, you know, your device, or otherwise also tells you to take a different bus to pick up your kid at school, um, you meet your kid and you walk, you walk the rest of the way home. So it's, you know, I don't wanna say, like, it sounds like I want people's lives to be dictated. What I want is them not to have to one.
Drive their car every day. But in order to not do that, I don't want them to have to think about all the other options and, and kind of dive through them. I think one of the places that I see issues is that people drive their cars every day, because it's the easiest thing. They don't have to think about it.
They can just drive and park. Right. Um, and the activation energy of considering other things is. And so what I would like to do, you know, what I would like to see is that we can sort of remove that activation energy that people generally have good experiences doing. The, the thing that, you know, whatever it is, your device, your, your watch, whatever tells you to do so you, you take transit it's on time.
Um, now of course, there's so many things that go into this, but like you are enjoying your experience. You're not stuck in traffic. You're having a more productive time during your commute. Your employer takes that into account. And so like you don't have to stay at work an extra hour, um, and you get to have more, you know, quality time at home and a better work life balance.
And you don't lose time during that, that period.
Deep: So I was kind of thinking you might do something crazy and say, oh, I don't really think we're gonna have any buses because there's gonna be self-driving cars.
Corey: I think there's gonna be electric buses, but I don't think, and, and maybe there will be self-driving cars, but I don't believe that that solves our problem right.
Deep: The cars. But does it like, like. Like I, if the, let's say two things happen, a cars are self-driving, um, you know, within reason, like, you know, maybe not in a hurricane or something, but they're, self-driving in, in, in like reasonably predictable scenarios. And, uh, and let's say that two they're totally electric.
And let's say, um, in that world, Your ride share algorithm is like figuring out how to like route smaller pods of vehicle. Like you don't even necessarily need to own, at least you can still meet your, maybe you just have your own car to get out of the city or something. But like you, you know, you, you press a button on your app.
Uber shows up. Um, you know, or Lyft, but it's electric and there's no driver and there's already two or three people in there because it's already figured out who's gonna drop off whom, where, and the cost drop. And, uh, you walk out and now I don't have to figure out how to get on my bus and I don't have to figure out any of that other stuff. I just have to walk on my door, get in and go.
Corey: Yeah. I mean, I think, I think that's possible. I do think it's probably more of a micro mobility solution than a self-driving. Like, so, and, and maybe. You know, there's no reason to say that a bus couldn't be self-driving or a van couldn't be self-driving. Right.
So like, yeah, I guess I don't see right now, what I see is that we're moving in that direction for smaller vehicles. I don't see that solving that many of the problems, because I think some of this. Some of the issue with carpooling will transfer over to that, which is that people don't like this shared, like sharing there's this cultural norm around sharing a ride when you're on a bus that doesn't exist around sharing a ride when you're in a car and.
Um, while a self-driving car,
Deep: You're talking about the, kind of the trust issue, because you're in a car, it's a smaller space. You have
Corey: to, there's a smaller space. You have to interact. There's, there's a feeling that you have to interact with people where you don't have to interact with people on transit, I think is sort of one of the driving things, right?
If you and I were gonna carpool. And I got in the passenger seat or wait even better. You picked me up and I got in the back seat and just sat there the whole time. That would be a little awkward, whereas like, there's this expectation when you get on a bus or transit that you don't have to interact with any of the people, you can have your headphones in and it's all good.
And, and while I see. You know, I hear what you're saying and I fully see that like a self-driving car could sit firmly in the middle of those two things. There ha there's a little bit of a culture shift that would have to occur that doesn't that's not just technological.
Deep: Well, thanks a ton. I feel like we had, um, we covered a lot of good terrain.
This has been, this has been a lot of fun. That is all for this episode of your AI injection as always. Thank you so much for tuning in. If you enjoyed this episode on AI and smart commuting, please feel free to tell your friends about us. Give us a review. And check out our past episodes podcast.xyonix.com.
That's podcast-dot-Xyonix-dot-com. That's all for this episode, I'm Deep Dhillon, your host saying check back soon for your next AI injection.
In the meantime, if you need help injecting AI into your business, reach out to us at xyonix.com. That's X, Y O N I X.com. Whether it's text, audio, video, or other business. We help all kinds of organizations like yours automatically find and operationalize transformative insights.