Introduction - Hodman Murad
🫛 This week’s edition of AI, Software, & Wetware features an audio interview on ethical AI with Hodman Murad, the 🇨🇦 Canada-based founder of Asaura AI. Hodman writes Between Thinking and Doing and The Data Letter on Substack. We discuss:
Why task initiation is a harder problem than motivation, and what a decade of diagnosing broken AI systems taught her about building one she’d actually trust
How she uses a daily stack of AI tools, from Perplexity to NotebookLM to Claude, to run Asaura as a solo founder
Why the 3Cs rule: Consent, Credit, and Compensation should be a baseline for ethical AI development, not an aspiration
What LinkedIn’s opt-in by default change and Spotify Wrapped reveal about the gap between data transparency and genuine user choice
How the AI companies worth trusting are the ones that lead with their limitations, not just their capabilities
and more. Check it out, and let us know what you think!
Audio length: 51:37. Reading time estimate: 40-45 minutes.
This post is part of our AI6P interview series on “AI, Software, and Wetware”. Our guests share their experiences with using AI, and how they feel about AI using their data and works.
This interview is available as an audio recording (embedded here in the post, and later in our AI6P external podcasts). This post includes the full, human-edited transcript. (If it doesn’t fit in your email client, click HERE to read the whole post online.)
Note: In this article series, “AI” means artificial intelligence and spans classical statistical methods, data analytics, machine learning, generative AI, and other non-generative AI. See this Glossary and “AI Fundamentals #01: What is Artificial Intelligence?” for reference.
Interview - Hodman Murad
Karen: I’m delighted to welcome Hodman Murad from Canada as my guest today on “AI, Software, and Wetware”. Hodman, thank you so much for joining me on this interview! Please tell us about yourself, who you are, and what you do.
Hodman: Thank you so much for having me. I really appreciate the opportunity to be on here, and great pronunciation of my name, by the way. I’m Hodman Murad. I am a data scientist turned AI founder. Earlier this year, I founded a productivity company called Asaura AI. I spent the first 11 years of my career working as a data scientist in the consulting world, so I was really lucky enough to work on projects with companies like Google and Sony, TD Bank as well. And then earlier this year, I thought it was finally time to start this AI company that I’ve wanted to for years now.
Asaura AI is a productivity tool, a productivity system built to help people when they can’t get started on something, whether it’s because they have ADHD and they’re dealing with executive dysfunction, or they’re neurodivergent, or it’s 4:00 PM in your office, or you’re trying to study at the library at night and your brain just happens to be tired. So I built this AI to help people complete their tasks a little easier.
Karen: Okay. Can you share a little bit about what the name Asaura means and how you came up with it?
Hodman: Yeah. Asaura is actually a combination of my friend Asia’s name, who went through a similar thing that I did and didn’t really come out the other side of it, and Aura, because she always had a great energy around her, so put together, it’s Asaura.
Karen: Wow, that’s a really nice recognition to your friend.
Hodman: Yeah.
Karen: You mentioned on another podcast, I think, about how you are neurodivergent yourself, and you’ve had some learning disability challenges, and then that you had all these friends who have been also struggling with some of the same things, and that you’ve kind of founded Asaura to be able to help them.
Hodman: Yeah. I was diagnosed with ADHD when I was a teenager, and I was lucky enough to have two type A parents who worked with me to kind of experiment and develop systems for myself so I could be successful in school or just life in general. Then when we started getting into our 30s, more and more of my friends were being diagnosed as ADHD or being on the spectrum of neurodivergence. So I was watching them relearn how to live their lives now that their brains made a lot more sense to them. And I wanted to help them the way that my parents helped me.
So I started reading all of these scientific, academic papers on non-pharmaceutical treatments of ADHD. Reading all the different types of treatments out there really inspired me to build this system right now that helps people break down tasks into smaller micro steps by focusing on the physical entry point of doing something so somebody can actually get started.
Karen: That sounds good. And we met on Substack. Could you say a little bit about what you do there?
Hodman: I have two publications on Substack. One is The Data Letter, which I started when I started writing on Substack last June, because I just had all these notes that I took from the span of my career, on these data projects that I worked on.
And I wanted to start The Data Letter because a lot of data people were getting laid off. And there’s also this whole new generation of aspiring data scientists coming into the field, and they weren’t really getting hired. So I wanted to share my expertise with people and put my work out there for them.
I also write Between Thinking and Doing, which I started this past January, originally to just build in public, build Asaura AI in public. But now that’s done and we’re definitely out of the beta now. I write a lot about ethical AI, the behavioral science behind AI and productivity, and just AI news that interests me.
Karen: So you started in January and you’re already out of beta. We’re recording this in mid-May?
Hodman: Yeah, yeah. It was really hard getting it out that quickly. But I was really happy with the way that people responded to the beta. I ended up having 51 beta testers and they gave me a lot of good feedback. And I ended up putting out the paid version on May 4th. It’s been good so far.
Karen: Yeah. Sounds like an interesting ride and I hope we’ll hear more about that as we go through the interview.
Hodman: Yeah.
Karen: So it sounds like you’ve been using AI professionally. Can you tell us a little bit about your level of experience with AI? How did you get into data science and how did you learn it?
Hodman: So I got into data science accidentally. When I was younger, I wanted to save the world, and I was in school for public policy and economics. But I ended up having to take this data analysis and economics course during my master’s and that’s where I just kind of fell in love with data. So I ended up working as a beginner in data and most of the tools we were mainly using back then was Excel. But working with these larger companies that adopted reinforcement learning systems and AI systems a lot earlier than what most people are picking up now. I spent a lot of time building and evaluating different models, understanding what these companies were actually learning, where they were failing, and whether the outputs from their systems could be trusted.
I’ve worked across industries diagnosing where AI systems went wrong, which gave me a fairly unsentimental view of the technology early on. But now as a founder, I think Asaura has added a product and design dimension to the technical foundation as well.
I don’t know if you wanted to ask any specific questions around that at all.
Karen: No, just in general, I’m always interested in how people learn. Some people learn from taking classes, some people go through tutorials, and just different ways. So I’m always curious about how people learn, what’s effective for them?
Hodman: When you work in consulting, it’s always on the job. When I have new graduates asking me how they should get started in data, I always recommend them getting started with consulting because the projects that you work on are so diverse and you’re completely managing yourself, so you really do become a full stack data person.
I call myself a data scientist because I like the model version of it and the analytics version of it. But as a data person in consulting, you learn to be like a data engineer, you learn to be like a back-end engineer as well. So it was all just like being thrown into it. Even at the beginning of it, I had to relearn calculus in order to understand the vector system that the data science works on as well. Yeah. Mine was definitely on the job and I was lucky enough to do that.
Karen: That sounds great. Well, it sounds like you started well before the generative AI tools really burst onto the scene in late 2022. So do you use AI tools, generative or not, to help you do your work with Asaura? Or do you use them in your personal life?
Hodman: Absolutely. I use them a lot, and I rely on a stack of AI tools every day. So I use Perplexity for market research because it’s really good for giving sourced answers. When you ask Gemini or Claude or even ChatGPT for links, they’ll give you a lot of dead ones, and you have to go back in and ask them. But Perplexity is very good for market research. For the research foundation for the academic and scientific papers that I’m always reading on cognition, executive function, task behavior, I use NotebookLM and Gemini, because it helps me organize these huge papers and connect ideas across these different sources. But it would take me hours to manually do it.
For all of my marketing, my content creation, I use Claude. It has much better writing quality than the other tools. And then, making Asaura, I coded everything because I didn’t want to get sucked into making an AI wrapper.
But now to maintain everything, ship updates, I’m using Codex. I use some Copilot in VS Code, but that didn’t really stick. I don’t like the UI of it. Just like using VS Code.
Karen: You mentioned writing Asaura and not wanting to write a wrapper, so I’m curious, is Asaura built on top of any of the foundation models?
Hodman: No.
Karen: So these are all machine learning models, but not generative models?
Hodman: Yeah. So I use all the skills that I acquired building different reinforcement learning models. You can kind of separate Asaura into three layers.
So the first is a curated set of templates for the most common stuck moments, like washing dishes, writing a cover letter, packing a bag if you’re about to go on a trip. Those override everything because I verified them and their accuracy, and the way that they break down.
The second is a large library of breakdown patterns that I got from open-source how-to data, and they’re indexed by verb plus noun. So “Make a smoothie” would get matched to the smoothie template.
And then the third is the feedback, which is how it learns about you and the user in that one account. When you’re looking at the UI, every step has a too small, too big, still stuck button. So if you tap too big in Asaura, Asaura kind of reframes that step as a 30-second commitment. So the goal of the too big button is just to get somebody moving. If you tap still stuck, you get a physical anchor so that you can work through the steps. Everything always goes back to a physical step just to get somebody moving.
And then there’s also the low energy mode that will start somebody on the easiest step within that sequence that was presented to them, within that micro step sequence. Using all those feedbacks that the user of that account can give, it adapts to them as they go on.
Karen: What would be an example of a physical anchor or a physical step? You’ve mentioned that here and on some other podcasts. I heard you mention the physical aspect, and I think that’s an interesting dimension.
Hodman: Yeah. So let’s say washing dishes will break down to “walk towards the sink”, “grab a plate”, or “grab the sponge”. It’s just things that you can do without having to think about how to start a big task.
For me and my ADHD, let’s say that it’s spring cleaning time. I have to clean out my closet. There’s a lot that’s baked into cleaning your closet, like grabbing garbage bags and cleaning supplies, deciding what to throw out, what to keep. So it would be at the point of “I have to clean my closet” that I would just completely freeze and not be able to break down what needs to get done in order for that closet to get cleaned.
So something that helps people, and that I read in a lot of these studies, is that just getting them to think about just committing to one physical movement is enough to get somebody going. And even if they don’t finish that entire task, at least a bit of a dent was made into it.
Karen: So the physical act would then be, in that example, just going to the kitchen and getting out some trash bags?
Hodman: Yeah.
Karen: Okay.
Hodman: Get trash bags, or grab hangers, or yeah, grab some Clorox wipes, whatever you use.
Karen: So that helps to get them unstuck from the potentially overwhelming decisions about what do I keep, what do I throw away – that sort of thing?
Hodman: Yeah.
Karen: Well, that’s really interesting, so thank you for sharing that.
Hodman: Absolutely.
Karen: Yeah. So you’ve mentioned using a lot of different AI tools. It sounds like a lot of it is for analysis and for synthesis, like analyzing articles. Can you share an example of a time that an AI tool did this well, and an example of a time where it was off the mark? You mentioned dead links, but beyond that?
Hodman: Claude does a really good job, especially with the release of Opus 4.7, of helping me organize a content calendar. I’m not a creative. I am getting to be one. I’m getting better every day. But when I first started with Asaura, and I had a mentor help me with the content creation part, but I just could not for the life of me think of a way to coordinate my Substack articles, my Notes on Substack, my LinkedIn posts, and just how to manage everything and coordinate everything so that I wasn’t creating different content or copy for every little thing that I was doing, and I could just cross-post and manage it.
Claude was super great at putting together a three-month content calendar for me, and now I just follow it. I’m actually starting another three-month stint tomorrow. But yeah, it’s really good for marketing or, at least, somebody who’s a beginner at it.
Karen: Thanks. That’s a good explanation. So when you think about a time when you’ve used it and you got something that was not very useful or that you recognized that, “No, this isn’t right,” I assume that’s happened at some point when you’ve been using these different tools?
Hodman: Gemini is not that great at synthesizing big projects. You have to keep repeating things to it. If it wasn’t for NotebookLM, I don’t think I could have used it for any research papers either. NanoBanana is good for images. But I wouldn’t do big project planning with Gemini.
Karen: When you’ve tried using Gemini for a bigger project, what happened?
Hodman: It’s what didn’t happen, and just constantly trying to prompt it to give you a bigger output.
Even when I’ve tried to write articles with it. I keep trying with different tools and seeing what they’ve gotten better at. This is very recent. When you try to write an article with it, even if you give it the amount of words that you want, you give it the template that you’re looking for, it’ll produce, like, two paragraphs. And it just doesn’t do well with deep thinking or giving you something of substance. At least not what I’m trying to use it for.
Karen: You mentioned NanoBanana and images, so have you been using the image generation tools?
Hodman: Yeah. If you look through Between Thinking and Doing or even my LinkedIn, all of those images that I’m using, all of those illustrations are from NanoBanana. It’s good for that. Another creator, Daria Cupareanu, just did an article on NanoBanana and she showed the realistic images that can be created with NanoBanana and how marketing teams are using it. I haven’t used it for realistic images yet, just little branded illustrations.
Karen: Yeah, Daria’s very good about explaining that, and Pinkie – you know, the AI Meets Girlboss account – she’s very good at talking about how to use it for branding.
Hodman: Yeah.
Karen: I see a lot of people are using it effectively for that, and making a recognizable brand, and that’s something that I’m relatively new to. It’s interesting to see people making headway with that.
Hodman: There’s a lot of really good marketing advice on Substack that’s really helped me a lot.
Karen: Yeah, that’s great. So it sounds like you’re using AI for a lot of different purposes. Are there any things that you would avoid using AI tools for, or that you do avoid using it for?
Hodman: I don’t use it for anything financial, just because I am acutely aware of how much data these tech companies are collecting from you. I don’t use it for budgeting, any of my personal financial planning. I just keep that in spreadsheets.
This isn’t any of the LLMs, but there was this deep fake video that was going around and scamming people in Canada last year. It was a deep fake of Mark Carney approving of this new investment fund. And whoever the scammers were sent it out, and a lot of people ended up investing in this fake fund because of this. That rubbed me the wrong way. So yeah, I don’t use AI for anything financial.
Karen: So it sounds like it’s more a concern about confidentiality than about the fact that it may get the numbers wrong. Some of the earlier versions were very bad about numbers, although I understand they’ve improved.
Hodman: Yeah, it’s definitely about privacy.
Karen: Okay. Do you have any concerns about using the tools, for instance, with images, or with writing, about either something leaking out or concerns about confidentiality of what you’ve put in there?
Hodman: No, because I don’t use it for anything that needs to be private. I don’t share passwords or anything with it. I don’t really want to talk about this too much, but I don’t like the way that Grok isn’t really managed well. And the explicit images that are made using Grok and how they don’t put any sort of barriers on it. I think it’s just gross. I would say, mainly, that’s my other concern, and that’s mainly a Grok thing.
Karen: Yeah. One of the concerns that comes up is, where did these tools get the data that they’re using? Where do they get the pictures that they use to generate these deepfakes and pictures or videos and such? In a lot of cases, they’re using data that people put online even decades ago, not thinking that it could ever possibly be exploited for some of these purposes. And a lot of companies are not very transparent about what they’re going to do with our data when we sign up, or they change the terms afterwards. We’ve seen a lot of that lately.
Hodman: Yeah.
Karen: There’s a concept called the 3Cs which came from the Cultural Intellectual Property Rights Initiative, CIPRI. They advocate that creative people, or people who create things in general, should be entitled to the opportunity to give consent if their work is used, or their data, and to be credited, and to be compensated if their works are used for training AI tools or such.
Hodman: Yeah.
Karen: I’m wondering what your thoughts are about that. Do you feel like it’s appropriate for them to have access to all the data they can scrape from the internet, or do you feel like the 3Cs ought to be applied?
Hodman: I think the 3Cs should be a baseline. I am not concerned with it for myself. All of these tech companies that you’re using, they’re all scraping data from you and using it to better whatever their technology is. But when it comes to musicians or writers, even developers, medical patients, they all contributed to the foundation that these models are built on, often without any awareness that their work or their health data was being used that way.
Consent should be explicit and informed. Nobody reads the terms and conditions. And I think, especially for musicians and artists whose heart goes into everything that they create, the credit matters because attribution is how we preserve the integrity of knowledge and creative work. Again, compensation, because the output of somebody’s labor is generating commercial value. They deserve to have a share of it. So it’s complicated. But it’s not complicated, really, and I feel like the companies that build this in from the start are going to be the ones that people actually trust in five years. If it evolves that way, if these companies start actually compensating artists.
Karen: Yeah. Some people say “It’s simple, but it’s not easy” is one way of putting it. The concepts of the 3Cs are very simple and really straightforward.
Hodman: Yeah ...
Karen: And there’s so much bias built into these AI systems, which as a data scientist I’m sure you’re very aware. But one way of offsetting the bias is potentially to allow them to have access to data from the groups that are currently underrepresented. And I’ve talked with some people that say, “You know what? I’m a Latina in data. I don’t see my voice, my perspectives, my experiences represented. I want them represented. So yeah, I want them to take all my data. Anything I post online is theirs and I want them to use it.” And other people say, “You know what? My experience has value, and I want that value to be compensated.” I hear both perspectives.
Hodman: Yeah. I wonder what are the laws for images and stock photos? I don’t know anything about stock photography, but I wonder if anything had to happen to, let’s say, Shutterstock having to compensate artists that are on their website, and photographers that have images on their website. I wonder if anything had to happen in terms of getting artists protected that way, that we could apply to how AI is using all of this information and art and music out there to train, to sell their products.
Karen: Yeah, art is an interesting area. On the one hand I’ve heard people talk about how Adobe has been making these efforts with their Firefly product. And they claim that they’ve licensed what they’ve used in their AI tools, or that they’ve compensated the people who have been using Adobe tools, Photoshop and such, for decades. But I’ve also heard, from some artists who use those tools, that they haven’t been compensated at all for the use. Or if they did it was pennies, for professional work that was maybe commissioned, that they did for an employer, and by rights that work belongs to them and the employer, and it’s basically just been stolen.
Hodman: Interesting.
Karen: There’s a lot of controversy about that. And different companies buy up sources of images. Obviously if something’s true public domain, that’s a different story. But when these people started using Photoshop online 20 years ago, they never dreamed that their work could be stolen and used basically to compete against them. And that’s where a lot of the lawsuits are coming from. Some of it is on this, the theft and violation of copyright, but some of it is the unfair competition as well.
Hodman: Yeah.
Karen: Most of the lawsuits that we see, there’s over 100 right now in the US and worldwide on copyright infringement relating, a lot of them are based on music. And the music industry actually has a better precedent for protecting the rights of musicians and performers and people who write music and produce it and perform it. So we may see some precedents come from there.
Hodman: Yeah. There was this singer musician on Spotify that had 1.4 million monthly listeners, and she just doesn’t exist. Like, the image is AI, the music has been made with AI, and everyone is just grooving to her music. There’s so much of it right now, and I don’t think they’re very forthcoming with how much music they’re using to train their AI tools or sound tools. It’s going to be really interesting over the next few years when we start seeing a lot more lawsuits out there.
Karen: Yeah, when I first started writing on Substack a little over two years ago, AI and music was one of the areas that I focused on, looking at what they’re doing and who’s doing what, and what ethical tools are out there, and there weren’t very many at the time. But there were a lot of stories about different companies that are basically stealing musicians’ work.
Hodman: Yeah.
Karen: Stealing copyrighted works from online sites, and using it. There are studies looking at the impact on musicians’ livelihoods as a result worldwide, and it’s pretty devastating for them. And Spotify has done some things. I know a lot of people – again, they’ve been using it for years, and they love it – but at the same time, some of what they do, the way that they treat human artists is just not right. Actually, this interview won’t go out on Spotify because I cut them off. Not that that’s going to hurt them, but I just feel better about not using them. But there’s a lot of movement in a lot of different areas of creation. And, you know, regulations always lag.
Hodman: Yeah.
Karen: Just the nature of it.
Hodman: So, yeah. I think it’s been really cool watching how streaming has evolved. But I didn’t really understand what it was doing to all of these creatives until the strikes in Hollywood. Was it SAG?
Karen: SAG-AFTRA, I think.
Hodman: Yeah, when the Screen Actors Guild went on strike a few years ago, or a couple years ago, I didn’t realize just how much they’ve dwindled the compensation that all these creatives get that are working on these shows and these movies. Yeah.
Karen: Yeah, there was a lot of activity there. I interviewed a retired actress earlier this year, and she was talking about how they basically wanted to cut out extras. You know, you just record some people, and then you use them forever in generated videos and movies, and they never have to pay for extras again. And just even using the voice of a well-known actor or actress. They felt strongly, they saw this early, that they needed to really stand up for it.
Hodman: Yeah ...
Karen: There are people like Joseph Gordon-Levitt who are on Substack and write about this and about the need to protect the rights of creative people.
Hodman: Yeah. Oh, God, was it a documentary or a movie where they used a deceased artist or a famous actor? It was Anthony Bourdain. But the family was watching this documentary on Netflix about their family member, and they were just shocked to hear this AI-generated voice.
Karen: Oh. Oh, man.
Hodman: Yeah. So the Roadrunner documentary, which was really good. It came out in, like, 2020 or 2021. Everything during COVID just bled together. But yeah, it was an AI-generated voice of his.
Karen: Yeah, there’s the whole right to control our digital legacy after we’re gone, who’s allowed to use our voice, our data, our everything. We could spend a whole episode on that, I think. Maybe we’ll do that sometime.
But the AI-generated movies and videos, there’s a – they call it an “AI actress” by the name of Tilly Norwood. And now the company that created, let’s say, it, is getting such a good reaction that they are trying to create, like, 30 more AI performers. And that’s just not a direction I want to see, you know? I think the human aspects of that, it’s why we make art, and it’s why we enjoy art. And I just … I can’t see endorsing that.
Hodman: No.
Karen: You talked earlier about consent or informed consent, and people not reading terms and conditions. I mean, there are studies that show that something like 80, 90, maybe even as high as 95% of people don’t read it, but usually they’re legalese. People can’t understand them anyway, so, can’t really fault people for not reading them. But how do you navigate that in Asaura, as far as giving people enough information that they know what you’re doing, but not so much that it’s overwhelming?
Hodman: I can’t see most of what people are putting in. Asaura recognizes verbs and nouns in order to help somebody get to that physical step, but I can’t see the details of what they’re working on. And that’s a decision that I made for myself. It does make it more difficult for me to use people’s data to generate more revenue from the tool, but giving people, my users, their privacy when it came to that, was very important for me.
Karen: So when you, when people sign up for Asaura, you have some sort of disclosure that says, “Here’s what I do and don’t do with your data”?
Hodman: Yeah.
Karen: Yeah, that’s good. So you’re being transparent with Asaura and setting an example there. But a lot of the companies are not following that. If you think about the AI tools that you use, do you feel like any of them have been transparent about sharing where they got the data that they use for their models and whether the original creators consented to its use?
Hodman: Yes, but only because they have to. Only because it hit the news, and there were journalists that did the work of actually researching this and seeing that OpenAI didn’t really ask for permission. OpenAI got the brunt of it at the beginning just because they’re the most well-known LLM. But now they’re publishing detailed accounts of the internet data and the academic literature their models are trained on, but only because they have to now. I’m glad that they have to now. That kind of disclosure matters, and it’s worth acknowledging, so at least they’re finally doing it. I think where the industry as a whole still falls short is on the specifics. And we were just talking about a bit of proprietary or user-generated data, what was collected when, and whether the people who created it have a meaningful say. We just talked about that.
On the Asaura side, the data informing how the product works comes from research literature that I read during the validation phase. The nouns and the verbs, the action inputs that people are put in. I also use a feedback software called Canny, where my users can anonymously rate their feedback to me, and that’s their deliberate choice.
Karen: Yeah, when you think about where OpenAI and some of these other tools got their data, and Anthropic had the big lawsuit settlement and such, we know that they’ve gotten their data without consent and credit and compensation.
Hodman: Yeah.
Karen: Yeah. We both feel, I think, that people whose works get used in training an AI tool have the right to the 3Cs. To the right to be able to consent to whether their work gets used, to be credited, to be compensated. Most people do feel that creators are entitled to that. At the same time, there are these tools out there and we all feel pressure to use them so that we don’t fall behind. We’re using tools that were not built with that. By using a tool like ChatGPT or Claude or Perplexity, we’re sort of rewarding the companies who didn’t act ethically in that regard. By using their tools, and in some cases by giving them money, but in other cases just by giving them our interaction data, which they then use.
So it’s a bit of a conflict, reconciling what we think and what we do, and it’s a difficult one because there’s so much pressure to use these tools and to not get behind. How do you feel about that, or how do you navigate that?
Hodman: Yeah ... you can’t get away from it. And the use of AI isn’t limited to using these LLMs. I know that everything I do with my phone, there’s some form of AI there. My email. I use so many different forms of technology that I just can’t get away from rewarding these giant tech companies. And for me, just because I worked in data for so long, I’m so aware of everything that’s being collected from me just, like, existing in this world.
I don’t know what the solution is with these large LLMs, because it feels like they’re just getting bigger and bigger. And it’s not going to be me that does it, but I really do hope that they are held accountable for the original creations that they’ve used and at least have to compensate artists and musicians and writers somehow. I really hope we see those cases come up over the next few years. I’m sure there’s somebody working on it.
Karen: Yeah, there are lots of lawsuits. There’s a writer, Edward Lee, he’s actually on Substack as well (ChatGPTIsEatingtheWorld). He tracks all the lawsuits that are active in the US and worldwide. And he’s got a map that shows where they are and lists the cases and keeps up with when they move in different directions.
Hodman: Yeah, I’d love to see more countries and states hold them accountable as well. Like, France is not happy with Elon right now because of those images that I mentioned that are being made on Grok. It’s not going to be really good for him. And I hope it does something to lower the value of the SpaceX IPO. I hope that more governments take the lead on this and stand up for their people.
Karen: Yeah, I think it’s very tempting a lot of times for people to say, “Oh, they’re just so big, there’s nothing we can do.” But we do have some influence and we do have some power. We can influence our legislators, we can act individually and collectively. And I think that we shouldn’t give up our agency on that so easily.
Hodman: Yeah. There is a case with OpenAI in Canada right now, because unfortunately their lack of oversight led to a mass shooting in British Columbia. That doesn’t happen in Canada. We don’t have mass shootings here.
Karen: Yeah, not like the US.
Hodman: Yeah. And I really hope that they’re held accountable. I know that our government is not very happy with them right now. So we need to keep putting pressure on them in order for them to actually act and follow through with what they’re saying.
Karen: As consumers and members of the public, even when we’re not actively choosing to use an AI tool, we are nonetheless still subject to having them used on us or on our data. There are a lot of cases: taking tests online for certification exams or going through TSA when we’re traveling and photo screening, social media, websites that ask for data. So there are a lot of ways that our data gets captured, gets used. Do you know of any examples where your information has been used?
Hodman: Mine specifically, no, but I think an example for me comes from working with publicly available data sets in my data science career. A number of widely used training data sets like Common Crawl, certain image repositories, scraped code from GitHub, like public repositories, were assembled and distributed without any sort of meaningful engagement with the people whose work was in them. And that happens a lot when it comes to data science and how corporations use data to train their own models, their own internal models.
So I used some of those data sets myself early in my work. But again, for me, like everything that I do, like I’m just aware of all of the data that’s being scraped for me at all times just because of the work that I did during my career. So it doesn’t really surprise me when I learn of a new way that somebody’s collected data on me.
Karen: Yeah. Our phones are one thing, and so many people have these doorbell cameras, and if you walk past their house down the street, you’re on their camera and you don’t even know it. And the smart glasses that some people were hiding the LEDs.
Hodman: That’s weird. Apple just got a patent passed. I don’t know what the stage is called, but they’re testing AirPods right now that have cameras in them.
Karen: I just saw that, yeah, a few days ago. I saw that announcement. I think the patent has been granted.
Hodman: Okay.
Karen: Now the question is: what will they do with it?
Hodman: Yeah ...
Karen: Companies patent a lot of things that they don’t ever act upon, or they decide by the time the patent comes through they’ve gone in a different direction. So we never know. But it’s certainly indicative that they’ve put research effort into that.
Hodman: Yeah.
Karen: So yeah, we’ll see.
Hodman: We’ll see.
Karen: Do you know if anything that you’ve written has been scraped and used in an AI tool?
Hodman: I’m sure, because I’m getting inbound subscribers from ChatGPT.
Karen: Oh, okay.
Hodman: So my stuff is out there. My stuff is out there, and yeah.
Karen: Substack gives us a setting that lets us control whether or not they tell third parties they are or aren’t allowed to train on our newsletters. Do you have that setting turned on or off? Do you know?
Hodman: No, I didn’t even know that existed!
Karen: Yeah.
Hodman: I just thought it was all under the same TOS.
Karen: Yeah, no, there’s a setting. And it’s not that Substack is building models with our data. As far as we know, they’re not. But this is basically, it’s like the robots.txt, telling the AI scrapers, the crawlers, whether or not they’re allowed to look at this newsletter or that newsletter.
And some people have turned it off for reasons, and other people have it turned on for reasons. One of the questions that always comes up is discoverability. Do you want what you write about to be discovered when someone does a search? A lot of people use LLMs like search engines nowadays.
And so if you want to be discovered there, as opposed to in Google Search, then it supposedly hurts discoverability to not allow that there. And there’s a bunch of stories around that too, but we won’t go there now.
Hodman: Yeah. I remember my friend was telling me that she applied for a job at Accenture, and when you’re putting your resume in, when you’re sending it in, they have a box that you tick. It asks you, “Would it be okay if this was read by AI?” And you can press yes or no. She pressed no, and it automatically rejected her application, which kind of sucked. There isn’t really a way of getting around it right now. But, yeah, I’m going to go check out that Substack feature.
Karen: Yeah. It’s a setting that’s per newsletter. There’s no control over Notes. Notes are public. But yeah, with newsletters we do have that option at least.
Hodman: Yeah.
Karen: Yeah. So do you know of any companies that you gave your data or your content to that made you aware they might use your information for training AI and machine learning?
Hodman: Yeah. Grammarly was actually the first tool that I used that was explicit about it. Although, when you read their disclosure, it’s varied over time, but it’s never really specific of how they’re using it. But now, they also have AI features on their tool.
I think It was, like, a couple years ago when LinkedIn just automatically opted in for you. And LinkedIn does that a lot. They’ll automatically opt you into something about your data, and then a power user on there puts a post that goes viral.
Those are the two that I can think of. One was honest and the other wasn’t.
Karen: Yeah. A lot of companies have these clauses that they can use your data for product improvement, and that’s just wide open. You have no idea what that is, and I guess it covers them legally.
But LinkedIn has been odd. On the one hand, yes, they added this opt-out option. But they opted all of us in by default for everything that we’d put in for the previous 20 years, and there was no way to undo that. And they also didn’t actively tell us. Like you said, a power user or an influencer had to say, “Did y’all realize this is what they just did?”
Hodman: Yeah, exactly.
Karen: And that was not cool. It’s not a good look for them when talking about professional work and behavior and a lot of our personal data. At one point there was even talk that they wanted to use our DMs, the texts of our DMs, and what about that leaking out?
Hodman: X does – so Twitter does that. Yeah.
Karen: I left. I mean, I kept the account there just so no one else took over the name, but I left several years ago from that. I couldn’t take it anymore.
Hodman: I don’t use it anymore, but I didn’t realize. So I still have it to communicate with brands whose products that I use. A lot of brands use it. It’s much quicker than calling in customer service. So I sent a screenshot to this one brand that I use, and they were like, “Don’t do that. Please send all ...” They had to explain what was going on to me, and I had no idea, and I couldn’t delete it. They were like, “Please send all photos or evidence to this email.” And that was very recent, so I didn’t know that X was, ... of course he is, though. It’s Elon.
Karen: Yeah. Yeah. Has it ever created any problems for you, with any of your personal information being shared or exposed against your knowledge or against your wishes? Or phishing attacks, anything like that?
Hodman: Yeah, actually. I guess it’s common. I love that you just asked about that. It was a few months ago where I got a phishing attempt of, like, somebody saying, “We have explicit photos from you that we took off your computer, and this is how much ...” They’re trying to get money off of me, but I never take photos like that ever. And it made me really sad that there’s people out there that are doing these things.
Karen: Yeah. There’s phishing attacks. I don’t get too many of those. I do get a lot of data breach notices, though. Companies that I thought could be trusted, a medical provider or other companies that were given my data. At one point I was on COBRA insurance, the insurance that you have temporarily after leaving a job. And my COBRA insurance provider just informed me recently that they had a data breach. I’m like, “Come on.” There’s really no way to prevent that. There’s nothing I could have done.
Hodman: Yeah, there’s nothing you can do. Yeah. Do you guys have Chapters in the States?
Karen: Not sure what you mean by Chapters?
Hodman: It’s a bookstore.
Karen: Okay.
Hodman: Oh ... so it’s just Canada. Okay.
Karen: Yeah.
Hodman: So it’s a big bookstore chain here. And in 2024, maybe 2023, all of the data that they had on their employees, like banking information, like their social insurance number, were hacked. So their system was hacked, their internal system was hacked, and all of this information of their employees was stolen.
They demanded a ransom, and they wouldn’t do anything with the employee data. And Chapters just decided, “No, we’re not paying for this.”
My friend works there, and she sent our group chat the email that she got from head office saying, “This is what happened. We’re not going to pay for this data, but we wanted to let you know.“ So yeah, that happens a lot. I don’t think I’ve gotten any ... Oh, wait, no, there was the Substack leak. Was it last year?
Karen: I think it was just earlier this year, because I remember when I changed – I use a unique email address for Substack that I don’t use anywhere else, so just, “Yeah, okay, okay. Kill that one.” I just create a new alias, and I use that. That was just a few months ago that we changed it over. But yeah. Yeah, that was one for sure.
Hodman: Yeah. I totally forgot about that. I wonder if there’s any other ones that I forgot about. I have a lot of friends in South America, and I know there was a big, like, Santander, if that’s how to pronounce the bank’s name, leak that happened, I think, maybe 2024. And so they all had to change their passwords and stuff. But yeah – it’s everywhere.
Karen: Yeah. And I think one thing that I think is worth pointing out is that this is not new with AI or even with generative AI. People have been scraping and stealing data and abusing it for decades. So it’s not new. What’s different, I think, is that if the scraped or stolen data gets trained into a machine learning model, then even if you get it corrected, there’s no way to get it out of the models. It’s gone.
Hodman: Because there’s more being created with that, and it’s multiplying. Yeah.
Karen: Right. There’s no way to really stop it from there. So that sort of makes it worse. But it’s actually not a new problem, it’s just amplified.
Hodman: Yeah. Yeah. I think the targeted content that I get is – I don’t know, it kind of creeps me out that I can say something about needing a new watch and get an ad served to me. So I use, like, three different ad blocks, so I never see ads. Even the Spotify Wrapped makes me think. It always rubs me the wrong way every year. It’s fun seeing how many minutes I listen to Beyoncé. But it’s a reminder of how much of my data is constantly being collected.
Karen: Yeah. That’s true. Well, last question. So we see public distrust of the AI and tech companies has been growing. A lot of studies show that here in North America and worldwide. Do you think that it’s possible to trust one of these companies? And if so, what do you think is the most important thing that they would need to do to earn and keep your trust?
Hodman: Maybe just more honesty and transparency around what they’re doing. Or at least own what they’re doing. And I think, just being honest around that. And I don’t know how much of that honesty we’ll get. Or at least them owning the things that they do.
I was at Anthropic’s developer day last week and several of the presenters were using the exact phrasing that people mock LLM generated copy for. Like the head of product for Claude Code was saying, “That’s a shift and it’s not just this, it’s that.” And there’s something honest about that. So a company whose presenters have kind of absorbed the patterns of the model they’re building, whether it’s self-aware or not. I’d rather have a company that owns its quirks and the bad things than one that performs a polish it doesn’t have. And in the end, it’s the consumer’s choice to not use Claude or OpenAI or ChatGPT.
And we have to keep forcing them to be honest, because they’re not going to do it on their own. But at least when somebody figures something out that they’re doing it, at least be honest about it instead of going in circles.
Karen: Yeah, that kind of authenticity and integrity is kind of hard to come by. I’m trying to keep an eye out for companies that are trying to do the right thing and try to shine a spotlight on them when I can find them. So yeah, if you know of any?
It sounds like Asaura AI is trying to do things the right way, and so I commend you as a founder for that. You know, your values are going to permeate the company for the remainder of its life. You’ve got them off onto a good start.
Hodman: Thank you. I’m trying.
Karen: Yes. I really appreciate you making time for the interview today. Is there anything else that you would like to share with our audience? Anything that’s upcoming?
Hodman: Asaura is out of beta, and it’s available for everyone to use at www.Asauraai.com. Asaura is a productivity system that’s built to help your brain when you can’t get started on something, whether it’s because you have ADHD and you’re dealing with executive dysfunction or you’re neurodivergent, or you can’t think linearly for that moment for some reason, if it’s 4:00 PM, it’s like the end of your workday. If any of that sounds like a problem you recognize in your own life, I’d love for you to try it. You get three free task breakdowns a day. I’m currently shipping updates twice a week based on user behavior, so the more you use Asaura, the better I can make it for everyone.
You can find me on Substack on Between Thinking and Doing, where I write a lot about behavioral science, building AI for good, AI ethics. And then The Data Letters, where I write about AI failures, data quality, and what it takes to build a model you can actually trust. So those are all the different ways that you can connect with me.
Karen: What’s the best way for someone to contact you if they want to give Asaura a try? Just sign up on the website?
Hodman: You can sign up on the website, at asauraai.com, A-S-A-U-R-A A-I dot com, and just sign up for free. I don’t take any credit cards, unless you want to upgrade. You get three free task breakdowns every day.
Karen: All right. Well, that sounds great. And did you want to say anything about your book chapter? The book is by SheWritesAI, and it’s called AI Everywhere. It’ll be Volume 2.
Hodman: Volume 2. I’m writing a chapter on designing AI for neurodivergent brains. And I’m so excited, and the deadline is so soon, and I’m going to make the deadline.
Karen: I’m curious, when you’ve been working on writing the book, has Asaura helped you in any way with getting unstuck on parts of the work of the writing? That would be an interesting story, I think.
Hodman: Yeah. A lot of it is based on decisions that I made, but also, from a lot of what I learned from the feedback that I got from the beta testers, the feedback that I got from the people who I interviewed during the validation stage last year. So yeah, there’s going to be a lot of it in there.
Karen: Yeah. I’m really looking forward to that. I know we’ve got a good community of people who are neurodivergent and who are using AI who write about this on Substack and have different perspectives on how it helps them with some of the things that you’re talking about, some of the problems that Asaura AI is addressing, so it’s really great to see the momentum and the recognition around that.
I also saw something interesting. I don’t know if this is also true in Canada, but in the US there’s the ADA, the Americans with Disabilities Act.
Hodman: Mm-hmm.
Karen: And there’s some talk that, as an accommodation for a disability that some places where people are not currently allowed to use AI tools, that the ADA actually could be interpreted to say that companies are required to allow people to use them because it’s a disability accommodation. And I can kind of foresee some legal fights coming up around that. A lot of companies limit what people can do. Some of them tell people they have to use it. Some people tell people they can’t use it. And this seems like maybe a path, that maybe neurodivergent people could use the ADA to help them get the accommodations that they need to be able to perform.
Hodman: Absolutely. Absolutely. I actually kind of wrote about that in one of my Substack articles on how the Anthropic and Department of War argument. Their conflict was a disability rights issue. But I just, I think AI has so much great potential to make the lives of regular people easier. Everyday people. And I say the things that I do and write about the things that I write because I want it to be done well. Because it has so much potential out there that we can’t let gigantic tech conglomerates taint it for us.
Karen: Let’s get the link to that article into the interview. We’ll include that when we publish it. I think that’ll be nice for people to see.
Hodman: Okay. Yeah, I’ll send it to you. Yeah.
Karen: Well, great. Anything else that you wanted to add?
Hodman: No, I think that’s good.
Karen: All right.
Hodman: Visit the Asaura AI site and use it.
Karen: Great! Well, thank you so much. I appreciate you making time for the interview today. Thank you, Hodman.
Hodman: Thank you.
Interview References and Links
Asaura AI
Hodman Murad on LinkedIn
Hodman Murad on Substack (Between Thinking and Doing, The Data Letter)
About this interview series and newsletter
This post is part of our AI6P interview series on “AI, Software, and Wetware”. It showcases how real people around the world are using their wetware (brains and human intelligence) with AI-based software tools, or are being affected by AI.
And we’re all being affected by AI nowadays in our daily lives, perhaps more than we realize. For some examples, see post “But I Don’t Use AI”:
We want to hear from a diverse pool of people worldwide in a variety of roles. (No technical experience with AI is required.) If you’re interested in being a featured interview guest, anonymous or with credit, please check out our guest FAQ and get in touch!
6 'P's in AI Pods (AI6P) is a 100% human-authored, 100% reader-supported publication. (No ads, no affiliate links, no paywalls on new posts). All new posts are FREE to read and listen to. To automatically receive new AI6P posts and support our work, consider becoming a subscriber:
Series Credits and References
Disclaimer: This content is for informational purposes only and does not and should not be considered professional advice. Information is believed to be current at the time of publication but may become outdated. Please verify details before relying on it. All works, downloads, and services provided through 6 'P's in AI Pods (AI6P) publication are subject to the Publisher Terms available here. By using this content you agree to the Publisher Terms.
Audio Sound Effect from Pixabay
Microphone photo by Michal Czyz on Unsplash (contact Michal Czyz on LinkedIn)
Credit to CIPRI (Cultural Intellectual Property Rights Initiative®) for their “3Cs' Rule: Consent. Credit. Compensation©.”
Credit to Beth Spencer for the “Created With Human Intelligence” badge we use to reflect our commitment that content in these interviews will be human-created:
If you enjoyed this interview, my guest and I would love to have your support via a heart, share, restack, or Note! (One-time tips or voluntary donations via paid subscription are always welcome and appreciated, too 😊)
