Latest / Paths of Curiosity in an AI World / Q3 - AI in Data, Behavior, and Human Systems with Alex Rosenblat, BA, MA

Transcript

Q3 - AI in Data, Behavior, and Human Systems with Alex Rosenblat, BA, MA · Paths of Curiosity in an AI World

• 0:44
  SPEAKER_00: Alex Resin Black is the director of sociotechnical research at the Markup and Cal Matters, where she investigated how AI and dark patterns affect patient privacy and healthcare.
• 0:56
  SPEAKER_00: She is the author of Uberland, how algorithms are rewriting the rules at work based on more than 5,000 miles of rides of kid workers across 25 cities.
• 1:09
  SPEAKER_00: She later spent three years inside Uber as head of marketplace policy awareness and research and has since returned to independent investigation.
• 1:20
  SPEAKER_00: She holds a BA in history from McGill University and an MA in sociology from Queen's University and a PhD in sociology from the University of Portugal.
• 1:32
  SPEAKER_00: Dr.
• 1:33
  SPEAKER_00: Rosenblatt, thank you so much for joining us today.
• 1:36
  SPEAKER_01: Thank you so much.
• 1:37
  SPEAKER_01: I'm delighted to be here.
• 1:39
  SPEAKER_00: You started as a sociologist, not an engineer.
• 1:43
  SPEAKER_00: Walk us through how you ended up as one of the most recognized researchers on how tech systems affect real people's lives.
• 1:52
  SPEAKER_01: Well, I joined the Data and Society Research Institute back when it was only a couple of employees.
• 2:00
  SPEAKER_01: And the institute was focused on the social impact of emerging technologies.
• 2:06
  SPEAKER_01: So I found myself embedded in an institution where public scholarship from people who had one leg in academia and one leg in the public sphere was designed to flourish.
• 2:20
  SPEAKER_01: The goal was let's try to anchor narratives about what tech can do or be in the experiences of people who are directly affected by them.
• 2:34
  SPEAKER_01: As an ethnographer, which is someone who really immerses within different areas to sort of experience it for themselves and have touch points with people who are experien who are experiencing similar things, I did it by riding around in a lot of cars with men and women across the United States to understand how new sharing economy technologies were affecting people who were finding new opportunities to work within that environment.
• 3:03
  SPEAKER_01: So the time at which I started my process, like what kind of work am I going to do, coincide with an emerging technology movement that was really centered around jobs and how technology can change them.
• 3:17
  SPEAKER_00: In our last quarter, I talked with Professor DuBall about what Uber is doing to its workers through the legal frame.
• 3:24
  SPEAKER_00: Your focus, however, is on the human behavior and everyday systems.
• 3:29
  SPEAKER_00: What did going inside Uber Tech teach you about human reality that we could never have learned from the outside?
• 3:37
  SPEAKER_01: Well, I joined Uber a couple of years after I'd finished seven years of field work and had published a book about the experiences of drivers who were being managed through algorithms, which really meant that instead of having direct control over how people do their jobs, you have a system of algorithmic management that becomes an intermediary for how to control the labor process people subscribe to without necessarily telling them exactly or precisely what they have to do to achieve their performance metrics.
• 4:11
  SPEAKER_01: That has real implications for how workers are classified.
• 4:15
  SPEAKER_01: If you can achieve control without that direct relationship, employers can elide some of the rules designed to protect workers.
• 4:25
  SPEAKER_01: So I was really just trying to understand against these claims that drivers are entrepreneurs and independent contractors who can become that way just by downloading an app, what is it like for them?
• 4:36
  SPEAKER_01: And my experience of that was that a lot of drivers did actually appreciate having more flexibility about when and where they could work.
• 4:47
  SPEAKER_01: And they liked and almost preferred having a computer sort of tell them what to do rather than having a human supervisor tell them.
• 4:55
  SPEAKER_01: But where the rubber really hit the road centered on accountability.
• 4:59
  SPEAKER_01: If you're a driver who's taken out a lease or you've invested significant capital or time in learning these platforms and you're suddenly deactivated one day, what's your recourse?
• 5:09
  SPEAKER_01: What do you do when things go wrong?
• 5:11
  SPEAKER_01: And that's really where this sort of technical governance system runs up against the realities of accountability.
• 5:18
  SPEAKER_01: So having experienced all of that and having written broadly about what it was like for drivers to have an algorithmic boss, a couple of years later, I was looking at what else could I know or learn or do to try to improve the outcomes or experiences for drivers.
• 5:36
  SPEAKER_01: And I had an opportunity to join Uber, where I tried to work really in evaluation and mitigation.
• 5:44
  SPEAKER_01: So I stood up ethnographic programs to help employees of the company understand and empathize with what the data-day experiences and challenges of drivers were like.
• 5:55
  SPEAKER_01: I tried to organize feedback that we could channel internally for internal advocacy about policies or processes.
• 6:03
  SPEAKER_01: And that taught me a lot about what goes into setting certain policies or moving new products and how to gain traction within a company to find outcomes that align with both the drivers and the company itself.
• 6:18
  SPEAKER_01: It's just a different avenue, I would say.
• 6:21
  SPEAKER_01: From the outside, what you can do is uh advocacy.
• 6:25
  SPEAKER_01: You can write publicly, you can document, you can engage with different stakeholders, and from an inside perspective, you have more opportunities to directly interact with people who are working directly on the products that people down the line are affected by.
• 6:42
  SPEAKER_00: You rode over 5,000 miles with gig workers across 25 cities to write your book, Uberland.
• 6:49
  SPEAKER_00: What did you learn from the drivers about their personal experience?
• 6:54
  SPEAKER_01: Well, I learned a lot, and it wasn't one homogenous experience.
• 6:58
  SPEAKER_01: A lot of people do gig work in particular to help them manage different facets of their lives.
• 7:04
  SPEAKER_01: So the context in which they're coming from really matters for how they how they do this job, how they approach it, and ultimately how they rationalize the same policies across different domains.
• 7:16
  SPEAKER_01: For example, one person I interviewed in New Orleans had a son who suffered from bipolar disorder, which meant that she sometimes had to drop a shift or quickly be responsive to an unexpected medical event.
• 7:32
  SPEAKER_01: And having flexibility about when to log in or log out at work was really meaningful to help manage the many different facets of her life.
• 7:43
  SPEAKER_01: For other drivers, you know, if you're a driver who started working for a Rydehale company in New York, you had to go through a lot of licensing and you had to pay a lot of fees to be able to operate in that city.
• 7:56
  SPEAKER_01: So a typical New York driver is not going to be part-time.
• 7:59
  SPEAKER_01: They're not balancing some driving work with some work at a Starbucks or work somewhere else or work at a company.
• 8:05
  SPEAKER_01: They're generally full-time workers, which means that they take their everyday choices about their livelihood seriously in a different way.
• 8:15
  SPEAKER_01: For example, if a driver is being sent to the sent on a trip destination that isn't going to result in a lot of money, or is just going to leave them in a place where it's difficult for them to pick up a return trip and they have to do what's called deadhead miles on the way back, they might grumble about that because that's they're absorbing the cost of providing customer satisfaction to a rider without maybe getting the economic benefit that they would want out of that trip.
• 8:40
  SPEAKER_01: And so their concerns might be really different.
• 8:43
  SPEAKER_01: Like everyone cares about their livelihood, but the proportion to which it matters in how you do this work means that drivers had really different investments in what they wanted this work to be.
• 8:51
  SPEAKER_01: For some people, it was a full-time job, some of them came out of the transportational industry already.
• 8:56
  SPEAKER_01: For some, this was just like a new kind of uh work they could easily do, but they'd already been in the service industry for a while, and this was just another version of that.
• 9:05
  SPEAKER_01: Um, for some people, they had really different attitudes about how they'd you know rationalize work that not wasn't always profitable.
• 9:11
  SPEAKER_01: They might say, Well, I'm gonna take this particular ride, like I'm gonna take every ride.
• 9:15
  SPEAKER_01: I'm just gonna let luck balance it out.
• 9:17
  SPEAKER_01: And I I generally believe in that.
• 9:20
  SPEAKER_01: And so they might get a trip that didn't seem very lucrative, but they would take it and sort of think, well, maybe the next one will be really lucrative.
• 9:27
  SPEAKER_01: And for other drivers, they'd focus on those really specific trips.
• 9:30
  SPEAKER_01: What is my actual take rate on this trip?
• 9:31
  SPEAKER_01: And they'd be really offended if the take rate that the company was keeping from the overall amount that the passenger was paying was really high, they'd be offended at that instance.
• 9:41
  SPEAKER_01: So people just really have different beliefs and cultural attitudes and really different costs, actually, and investments in this work that provides a real range of perspective on how they'd be responsive to any specific policy or feature or product.
• 9:56
  SPEAKER_00: You documented how Uber's apps were designed to make drivers feel like they were making free choices when the system was actually steering every decision.
• 10:06
  SPEAKER_00: How does that kind of invisible design actually work?
• 10:11
  SPEAKER_01: Well, drivers never received, and it's not just Uber, drivers from Lyft, from others don't receive an employee handbook saying these are what the rules are and this is what your expectation is.
• 10:22
  SPEAKER_01: They generally have some community guidelines, and they have instead performance metrics that are very changeable.
• 10:31
  SPEAKER_01: So they might receive notice that their rating has fallen too low and they're at risk of deactivation, which was a sort of prominent fear that drivers held about whether or not they could lose access to the platforms that they started to rely on for work.
• 10:48
  SPEAKER_01: And they'd have to sort of guess at what the right behavior was to please the algorithm.
• 10:53
  SPEAKER_01: And that behavior could change.
• 10:55
  SPEAKER_01: Those expectations could change and they're regionally different.
• 10:58
  SPEAKER_01: So a policy that was in force in one city might simultaneously not be enforced in another city at the same time.
• 11:04
  SPEAKER_01: So, for example, I remember a discussion drivers were having, I think, in a city in the Northeast, and they wanted to unsubscribe, basically, or avoid being co-opted into doing carpool rides because they didn't pay enough to justify the passenger dissatisfaction that could also result in lower ratings from people who were stuck now on a nonlinear route to their destination.
• 11:28
  SPEAKER_01: And I remember watching in forums where drivers, I think in Seattle were saying, Oh, we were actually able to opt out of this here, and drivers in another city, you know, in the Northeast could not.
• 11:36
  SPEAKER_01: And so you'd sort of see this emerging discourse about their about the rules as they were figuring them out, which goes to show you that there aren't clear rules.
• 11:46
  SPEAKER_01: There's these limited inf there's limited information that the drivers get.
• 11:50
  SPEAKER_01: They crowdsource a lot of it amongst themselves, and they have the power to make limited choices, which can contribute to a feeling that they're making independent choices because they are.
• 12:01
  SPEAKER_01: They're choosing whether or not to take a ride, they're choosing whether or not to participate in a promotion.
• 12:06
  SPEAKER_01: But on the big picture things, like negotiating the rate at which they are paid, they have no power.
• 12:12
  SPEAKER_01: And also they face the potential disciplinary action of not meeting certain performance benchmarks.
• 12:19
  SPEAKER_01: And so they can decline the ride, but they might worry that affects their drive or their ride cancellation rate.
• 12:24
  SPEAKER_01: And if the ride cancellation rate falls too low, maybe they lose access to the platform, maybe they're at risk of deactivation.
• 12:30
  SPEAKER_01: And while they might receive warnings about this, and drivers themselves might spread either rumors or information or experiences about that happening, they never had clear insight into, for example, how often a deactivation would occur.
• 12:44
  SPEAKER_01: That's information that only the platform would hold.
• 12:46
  SPEAKER_01: And so they're operating in this in-between zone where they have some information, some limited power, but no real power over the big things, and some information, but not all the information, they would need to make informed economic choices.
• 12:59
  SPEAKER_01: And I would say that this has evolved over the years.
• 13:01
  SPEAKER_01: Like when I was starting my work, it could be something as basic as a blind passenger acceptance policy, which meant that a driver didn't know the destination of the person they were going to pick up before they arrived to pick up that passenger and started the ride.
• 13:16
  SPEAKER_01: What that means is they don't know what that trip is worth.
• 13:18
  SPEAKER_01: There's not an amount on it.
• 13:20
  SPEAKER_01: And so if they decline it, they risk having their ride acceptance rate go down or the ride cancellation rate go up, which can have disciplinary effects.
• 13:29
  SPEAKER_01: And it's not that it's a terrible thing for an employer to tell a worker what to do and this is where you have to go, but the difference is these drivers were being valorized through corporate rhetoric about how they were entrepreneurs.
• 13:41
  SPEAKER_01: And an entrepreneur, or even just an independent contractor, is supposed to be able to make informed choices about their decisions.
• 13:49
  SPEAKER_01: And they really couldn't make those decisions in their own best interest under those limited and constrained circumstances.
• 13:56
  SPEAKER_01: So the sum up of it is, I would say, after all of these years, is that they have a lot of freedom to make really constrained choices.
• 14:04
  SPEAKER_01: And they operate in that space, I think, to this day.
• 14:10
  SPEAKER_00: That is so crazy.
• 14:12
  SPEAKER_01: What I've been doing in my current role as the director of sociotechnical research at CalMatters and the Markup is exploring manipulative design that guides people to make choices they might not otherwise make about, for example, how their data is used.
• 14:28
  SPEAKER_01: Where I've been exploring this is at the registration desk of physicians or nurses or other provider practices in clinics or in hospitals.
• 14:38
  SPEAKER_01: Typically, when a patient registers for their appointment, they either receive a link to their mobile phone or they are handed a tablet and someone says, please fill this out and let us know when you're done, or hand it back when you're done.
• 14:50
  SPEAKER_01: Those tablets or those form flows contain a lot of paperwork.
• 14:55
  SPEAKER_01: Some of them are notices of privacy practices, some of them are consent to treatment forms, some of them are consent to certain financial policies that a clinic might have.
• 15:06
  SPEAKER_01: But I've been reading all of those really carefully for a number of years, and now as really a method to audit where you can make choices when the forms say you have a choice.
• 15:18
  SPEAKER_01: So some of these forms will say, we'd like to use your data.
• 15:22
  SPEAKER_01: And some of those forms could say, and this is one that I encountered, but here within the bundle of consent forms that are required as a form that is legally optional.
• 15:31
  SPEAKER_01: But the form that I saw, which was offered by Frisia, a patient registration platform, said this is required.
• 15:41
  SPEAKER_01: Like in red letters, it just said required at the top left corner, which might render the impression to a patient that this is in fact required.
• 15:47
  SPEAKER_01: Is a legally optional form by which they request to use the data you have entered through your health registration, which is meant to go to your doctor, and use your other records to target you with advertising.
• 16:00
  SPEAKER_01: And I started to say, what happens when you try to say no?
• 16:06
  SPEAKER_01: You're often given the option, say I accept or I decline.
• 16:10
  SPEAKER_01: What happens if you say no?
• 16:12
  SPEAKER_01: In that particular case, which is from 2023, I wrote to the vendor, their privacy email, and said, Hey, do you have any records of consent on file for me?
• 16:23
  SPEAKER_01: Just on a hunch.
• 16:24
  SPEAKER_01: And they wrote back and said, Yes, we do.
• 16:27
  SPEAKER_01: And I said, Show me.
• 16:28
  SPEAKER_01: And they said, No, you have to get this form directly from your doctor's offices.
• 16:33
  SPEAKER_01: And they sent me on a hunt for a phantom form across all the doctor's offices I'd been to that year, which were a lot just because I was pregnant and you do a lot of routine doctor's visits when you're pregnant.
• 16:45
  SPEAKER_01: And I found that they had never had a consent on file for me.
• 16:50
  SPEAKER_01: And they finally admitted it after a three-month hunt.
• 16:55
  SPEAKER_01: That gave me a lot of pause.
• 16:57
  SPEAKER_01: And I was like, well, even when the form says no, can you say no?
• 17:01
  SPEAKER_01: No, how what how do you keep your records of that as well?
• 17:05
  SPEAKER_01: A couple of years later, I launched a broader project, which is what I'm doing now, basically auditing these intake systems and saying, okay, well, when it says you can restrict how your data is going to be used or shared, you know, how do you do that?
• 17:17
  SPEAKER_01: And what I've come across is really basic design infrastructure that manipulates your choices.
• 17:23
  SPEAKER_01: Those are often called dark patterns.
• 17:25
  SPEAKER_01: And that could be as simple as the form will say, you have three choices about how your data can be used.
• 17:33
  SPEAKER_01: And when you scroll to the bottom of that form so that you can make your choice, there's one choice and it's a button and it says I accept.
• 17:41
  SPEAKER_01: And to do anything other than accepting, you'd have to send a bunch of emails back and forth with the vendor.
• 17:47
  SPEAKER_01: You'd have to verify it later.
• 17:49
  SPEAKER_01: And so I've just been looking at a really basic thing.
• 17:52
  SPEAKER_01: You know, if you have the ability or the right to say no, can you?
• 17:56
  SPEAKER_01: And the answer is kind of shocking because often it's basically impossible.
• 18:00
  SPEAKER_00: Your investigation found you couldn't get a copy of your own child's hospital consent forms after surgery.
• 18:07
  SPEAKER_00: What happened?
• 18:09
  SPEAKER_01: In that particular instance, I, as is my habit, try to keep records of all the health forms that I'm signing so that I can have a record of whether I said yes or said no.
• 18:19
  SPEAKER_01: And I was asked to sign a number of forms when I came to the hospital with my son for a minor surgery he was having.
• 18:27
  SPEAKER_01: And I asked if I could just take a photo of the consent form they showed me to verify my signature before they wheeled him away.
• 18:35
  SPEAKER_01: And they said no.
• 18:36
  SPEAKER_01: They said this violates federal privacy law, which it absolutely does not.
• 18:40
  SPEAKER_01: But you're not gonna fight with the person who's about to do surgery on your son.
• 18:44
  SPEAKER_01: I saved it for later.
• 18:45
  SPEAKER_01: And I said, just tell me how to get a copy of this form.
• 18:47
  SPEAKER_01: And they sent me to another hospital and said, you can get a copy at medical records.
• 18:52
  SPEAKER_01: And I walked over to the other hospital through a wind tunnel in New York and asked the information desk where this office was, and they said, Oh, it doesn't exist.
• 19:02
  SPEAKER_01: Unless you want x-rays.
• 19:04
  SPEAKER_01: So in the end, it took me a pretty dedicated amount of time to merely get a copy of this form that said, I consent.
• 19:14
  SPEAKER_01: And I did it, of course, I did consent, but I like to keep records just in case.
• 19:18
  SPEAKER_01: And just getting your own copy to verify what you said turned out to be like a crazy rabbit hole that I wasn't quite expecting.
• 19:26
  SPEAKER_01: And so uh I I did ultimately get the form, but I think a typical patient would not prevail with the number of steps it would take to get to that outcome.
• 19:34
  SPEAKER_01: And it seems like such a basic problem to solve.
• 19:36
  SPEAKER_01: Send me a copy or make it available, or print it out, or anything.
• 19:41
  SPEAKER_01: Like just taking a photo proved to be a barrier.
• 19:44
  SPEAKER_01: And I'm not totally sure why.
• 19:46
  SPEAKER_01: Like you're technically legally entitled to a copy of these things, but you're not entitled to it easily.
• 19:50
  SPEAKER_01: And that's what I've discovered is like this major friction that is running through these different circumstances where you should be able to do A, but it's really difficult to do it.
• 20:01
  SPEAKER_00: So if a student listening right now wanted to find out what data a company or a hospital already has on them, where would they even start?
• 20:13
  SPEAKER_01: They would probably start by asking for a copy of their medical records.
• 20:17
  SPEAKER_01: Sometimes you can get some of those in a portal, but they're not complete, is what I've discovered.
• 20:22
  SPEAKER_01: A lot of the um paperwork you sign can be sort of shifted between some of it's in medical records, but some of it's not.
• 20:28
  SPEAKER_01: It's in like patient administrative services.
• 20:30
  SPEAKER_01: So even knowing the names of those different categories that can exist would be helpful to start.
• 20:35
  SPEAKER_01: It was new to me as a non-medical professional.
• 20:39
  SPEAKER_01: Uh and then you might have to go through different vendors to try to get answers about what you signed or what you didn't.
• 20:47
  SPEAKER_01: And that can be a surprise to people too, and to doctors as well, who will say, I don't really know who's in the examining room with me.
• 20:53
  SPEAKER_01: Like they're aware that all these different vendors are there.
• 20:56
  SPEAKER_01: Um, but it might have been organized by their clinic manager or their hospital system, you know, and they're like the first maybe touch point you would have, the first human in the loop you might ask.
• 21:04
  SPEAKER_01: But don't be surprised if the doctor or the nurse doesn't actually have the answer for you.
• 21:08
  SPEAKER_01: It's unlikely their front desk will have the answer for you either.
• 21:10
  SPEAKER_01: You'll probably have to go to a dedicated privacy or compliance office, which means lawyers.
• 21:16
  SPEAKER_01: That is code for lawyers.
• 21:18
  SPEAKER_01: Um, and you might or you might end up with an office worker who's sympathetic or seems like eager to help, and they might just get you what you need, even if you're being sent you know across the street to a ghost office.
• 21:27
  SPEAKER_01: It might be that the lady at patient registration who seemed helpful before is a good is a good place to start.
• 21:34
  SPEAKER_00: Healthcare systems can make you sign privacy notices even when that's not required under health privacy law.
• 21:41
  SPEAKER_00: How is that legal and what are the implications for patient privacy?
• 21:47
  SPEAKER_01: That one was pretty interesting to me.
• 21:49
  SPEAKER_01: So there are amongst the patients who I interview, I've interviewed dozens of people, both patients, providers, advocates about how they navigate these registration forms.
• 22:00
  SPEAKER_01: And how they think about their own privacy.
• 22:02
  SPEAKER_01: And a common point of resistance among patients and providers who are also patients when it comes to their data is to say, okay, I want to get a copy of the Notice of Privacy Practices, which hospitals and clinics typically ask you to acknowledge you have received.
• 22:20
  SPEAKER_01: And it's hit or miss whether you've actually received it.
• 22:23
  SPEAKER_01: But you can insist and get a copy of it.
• 22:26
  SPEAKER_01: And once you get a copy of it, under HIPAA, providers are required to make a good faith effort to have you sign an acknowledgement that you've received it.
• 22:37
  SPEAKER_01: That's it.
• 22:37
  SPEAKER_01: It's not supposed to be a consent form.
• 22:40
  SPEAKER_01: But what's actually happening, which I've been documenting, is that those forms are being filled with additional terms that elicit your consent, even in a tablet, for example, in a bundled consent.
• 22:53
  SPEAKER_01: So maybe there's forms you have to consent to, and this form you don't technically have to consent to under law, but they're all bundled together and you click like the same thing to say, I have consented to this, and you check the box and you sign it at the end.
• 23:04
  SPEAKER_01: And so they're sort of blurring that category, which seemed a little amiss.
• 23:09
  SPEAKER_01: But when I explored this further with legal experts and with health and human services, it I discovered that although providers are not legally required to have you sign this form, which gives the impression that it is optional, there's also no law in the privacy rule or in uh uh under federal privacy law that prohibits providers from requiring you to sign it, um, which is pretty interesting.
• 23:33
  SPEAKER_01: And when I talk to experts about this, they basically say, well, like they just never envisioned this happening.
• 23:38
  SPEAKER_01: Um there's some movement within HHS to change the requirement that providers try to get you to acknowledge it.
• 23:45
  SPEAKER_01: But I think it's an interesting space because the privacy advocates and patients and providers that I talk to will say, just don't sign it.
• 23:53
  SPEAKER_01: You know, but it it which seems reasonable, but actually a provider could say, well, then you can't seek care here because I'm I'm requiring you to sign this.
• 24:02
  SPEAKER_01: So they have the impression it's optional, and then there's this like stuff happening where it's not really being framed as an acknowledgement.
• 24:08
  SPEAKER_01: It says, I consent, I I agree to all these terms, and moreover, there'll be additional verbiage later in the forums that will say it is your responsibility to opt out of any term you don't like.
• 24:18
  SPEAKER_01: And the terms that you maybe didn't like are buried in that policy, which you have to read really closely.
• 24:25
  SPEAKER_00: You've investigated from the outside, worked inside, and come back out to investigate again.
• 24:31
  SPEAKER_00: What does that full arc teach you that most researchers never get to see?
• 24:37
  SPEAKER_01: I think it's been really interesting to work it from a number of perspectives.
• 24:41
  SPEAKER_01: I find it really valuable to understand how different stakeholders might be thinking about any given issue.
• 24:46
  SPEAKER_01: Uh by uh when I first started working at the Data and Society Research Institute, I was there at a nonprofit full of academics and para-academics who were doing uh public interest work but through their academic publishing and academic research.
• 25:01
  SPEAKER_01: And that was really interesting.
• 25:03
  SPEAKER_01: But I also appreciated working inside a company that I'd studied from the outside for so long because it gave me just a new perspective and appreciation for the complexities of that world.
• 25:13
  SPEAKER_01: You know, how do you affect change that makes sense to different stakeholders is a real question that I hold more so uh because I've I've been able to have like the privilege of accessing that logic.
• 25:25
  SPEAKER_01: You know, how do you think about this?
• 25:26
  SPEAKER_01: How can I then structure an intervention or how could someone else's intervention map onto the different reasoning that different stakeholders have?
• 25:34
  SPEAKER_01: And I don't tend to have a specific picture of like what intervention I would want, but what I really like to do is create a map of all the different reasoning of different stakeholders in the picture, sort of present, you know, here's how a number of interventions could work, here's how it could overlay with the with that cultural logic.
• 25:50
  SPEAKER_01: And it's not really for me to decide at the end, it's you know, it's for a regulator, it's for a judge, it's for whomever, but like here's how it's gonna work out, you know, I think.
• 25:59
  SPEAKER_01: And that's what I find really is is a value that I can bring to any given discussion in these in these areas because I've been able to work, you know, inside, outside.
• 26:10
  SPEAKER_01: And now again, I'm sort of coming full circle to work at uh in a reporting capacity at a journalism institute, which has given me, of course, a new appreciation for all the different ways that journalists establish what is fact.
• 26:21
  SPEAKER_01: You know, all these different areas, how academics establish what a fact is or what an what a good analysis is and how to cite people in it, and what journalists do, they all have their own distinct flavors.
• 26:32
  SPEAKER_01: Um it's like learning new languages.
• 26:34
  SPEAKER_01: You're like, okay, I'm I'm now a peer to this, I understand what they're doing here, I understand what they're doing over there, um, which might not even help you understand how it maps onto a legal argument somewhere, but this kind of exposure can at least bring together that that um that qualitative feeling for what could work out here, what could the consensus be?
• 26:57
  SPEAKER_01: You know, what are the range of possibilities that would be reasonable to all of these different stakeholders in a room who may not know that that they're in the same room together, but who are all just affecting maybe a public discourse, for example, or product change through their own avenues.
• 27:11
  SPEAKER_00: What is one skill that most people would not associate with a tech researcher, but has been highly important to your work?
• 27:20
  SPEAKER_01: Well, I would say that ethnography is itself a little bit on the margins of most research.
• 27:27
  SPEAKER_01: A lot of research, especially within tech, really values and prizes quantitative knowledge.
• 27:32
  SPEAKER_01: Put a metric on this, put a number on this.
• 27:34
  SPEAKER_01: You can do it, it's a little bit easier to make decisions around things that you measure.
• 27:37
  SPEAKER_01: But not everything can be measured in a way that gives you the right proportions for how this might affect sentiment or how that can affect uptake.
• 27:44
  SPEAKER_01: Ways in which people can vote against their own interests, for example, that would benefit from explorations of ethnographic reasoning, of like, okay, well, how are people really thinking about this to know where there's a match?
• 27:55
  SPEAKER_01: And so I'd say that that's probably it.
• 27:57
  SPEAKER_01: It's the it's the fact of doing qualitative research that gives you a grounding for how big promises are actually landing on people affected by them, the distribution of harms amongst people who are affected by those same technologies, as well as the benefits.
• 28:11
  SPEAKER_01: You know, what is that feeling people have about uh going to work or using a particular product?
• 28:17
  SPEAKER_01: And I think that that is really important to consider, you know, like you could talk all day long about how someone could be exploited by a data-centric system, it could be meta, it could be anything, but they also get gain some affective benefits, some joy every time someone clicks a like button on their post, you know, so you have to contend with these different issues.
• 28:37
  SPEAKER_01: It's never as simple as, well, here's how much people are using it.
• 28:40
  SPEAKER_01: It's like, well, what is their feeling about using it that like would anticipate either resistance to or support for a variety of interventions that ultimately have to map on to the communities that they represent?
• 28:52
  SPEAKER_01: Um, I think that's been true more broadly in these debates about AI and how to regulate them.
• 28:58
  SPEAKER_01: It's like, well, what are really the nodes of either resistance or adoption that people are experiencing?
• 29:04
  SPEAKER_01: And so what frameworks can we use for governing these technologies?
• 29:09
  SPEAKER_01: Um, that's like a really ripe opportunity for someone to explore if they're a student now or are anticipating going to university in the future or you know, working uh in these areas.
• 29:20
  SPEAKER_01: It's mapping those different uh perspectives and going a little bit beyond what the numbers can tell you to what people can tell you.
• 29:28
  SPEAKER_00: What do you feel is one of the most underreported ways technology is shaping people's lives right now?
• 29:36
  SPEAKER_01: Underreported is a good question.
• 29:38
  SPEAKER_01: Um I don't know that I have a good answer to your really excellent question.
• 29:43
  SPEAKER_01: I always looked for where are people on the margin starting to make noise?
• 29:49
  SPEAKER_01: You know, like they're seeing something that we're not.
• 29:52
  SPEAKER_01: And so I always think like, where is there a pre-movement uh developing because there's a clash between how a new form of power, whether that's a technology or an economic structure or a legal structure, you know, affects people, and where might they be looking for opportunities for change?
• 30:10
  SPEAKER_01: And it may be that their logics don't cohere to yours or their politics don't cohere with yours, but usually that seed of resistance is an indicator or a red flag that something has gone amiss.
• 30:20
  SPEAKER_01: And you might find that there's uh odd bedfellow alliances amongst people who wouldn't normally be in the same room together, who end up forming a common uh resistance to something, uh for example, like surveillance pricing.
• 30:33
  SPEAKER_01: You know, there's a new sort of politicization of this well-documented uh phenomenon, which is economic and social and technological, where when you use a product or you use a service, you're often being monitored.
• 30:46
  SPEAKER_01: And that data that's monitored or collected about you and your use of it is fed amongst a lot of data brokers that can build profiles on you or inform future choices.
• 30:55
  SPEAKER_01: And I think that while that's been abstract, no one's all that concerned about it.
• 30:59
  SPEAKER_01: But as it becomes more visible, like I'm getting a different price, for example, at the grocery store than my friend is, then you have a real like breaking of that slim layer that protected your visibility from caring about the outcome.
• 31:15
  SPEAKER_01: And as those as those uh start to become visible, people have a break between what they thought mattered and what they didn't think mattered before.
• 31:25
  SPEAKER_01: And that's always like a really ripe opportunity for looking for what's being underreported.
• 31:29
  SPEAKER_01: You know, all of a sudden this thing that's been happening for a long time really matters to people.
• 31:33
  SPEAKER_01: You know, why is that?
• 31:34
  SPEAKER_01: Uh is it because they're suffering from lack of opportunity?
• 31:37
  SPEAKER_01: Is it because it seems or feels unfair?
• 31:39
  SPEAKER_01: You know, what what is changing?
• 31:41
  SPEAKER_01: And that's that's where the under-reporting is.
• 31:44
  SPEAKER_01: It's like, okay, go there, go to that like odd friction where it doesn't really make sense what's happening, but something is happening and we're gonna find out why by talking to people.
• 31:53
  SPEAKER_00: Before we wrap up, are there any books, podcasts, or other media you would recommend for a student who wants to learn more?
• 32:02
  SPEAKER_01: Uh I read all sorts of things, but I think what I would recommend would really depend on the interests of any given reader.
• 32:08
  SPEAKER_01: Uh, I would welcome anyone who has an interest in these areas to send me an email at info at alexrosenblatt.com and I'd be happy to provide them with resources or uh future reading that might help them on their journey.
• 32:23
  SPEAKER_00: Thank you so much for joining us today on paths of curiosity.
• 32:33
  SPEAKER_00: This has been Paz of Curiosity in an AI world.
• 32:37
  SPEAKER_00: Exploring careers in the land of AI.
• 32:40
  SPEAKER_00: Until then, curious.