[MUSIC PLAYING]

MOLLY BLAKEMORE: Hello, and welcome to Free Speech with CR. I'm Molly Blakemore, and I'm here with my co-host, Dr. Keith Flamer.

KEITH FLAMER: Molly, it's good to see you again, and it's good to be here with our guests. And would you please introduce yourself for us?

MARIE DOCE: Yeah, my name is Marie Doce. I'm the Founder of Grounded AI. I have a local implementation firm here in Humboldt County. I help businesses here implement AI in their day-to-day workflows. I'm excited to be here. Thank you for having me.

KEITH FLAMER: We often wondered what brings people to Humboldt County? Now talk to us a little bit about yourself and why you're here, and why is it so important that we talk about AI these days?

MARIE DOCE: Yeah, thanks. So I grew up here in Humboldt County. I went to Jacoby Creek in Bayside and NPA for high school, took classes at HSU back when it was HSU as a high school student. Just really loved the area. And in 2007, I moved away for college on the East Coast and lived in Boston for seven years. That's where I started working in tech.

I moved to San Diego during that time frame. I was working at Tripadvisor, working in a remote role. Then I stayed in San Diego for another seven years, and then in 2020, I moved to Austin, Texas and I was there for five years. And moved back home last summer after 18 years away with my two kids. They're one and two now.

And I was ready to be closer to family. My parents are here. I have extended family out in Willow Creek, Hoopa area, lots of cousins in all over California. And I just missed California, and I wanted my kids to have access to the same things that I loved about growing up in Humboldt, and just farmers markets and events and--

MOLLY BLAKEMORE: Great, so you're local, really?

MARIE DOCE: Yes.

MOLLY BLAKEMORE: Welcome back.

MARIE DOCE: Thank you.

KEITH FLAMER: Welcome back home.

MARIE DOCE: Thank you.

KEITH FLAMER: So you went to HSU, correct?

MARIE DOCE: No. I took classes at HSU as a high school student, so I was in the Humboldt Symphony Orchestra. I played the viola, and--

KEITH FLAMER: So do I.

MARIE DOCE: Oh, really?

KEITH FLAMER: That's wonderful.

MARIE DOCE: You play the viola?

KEITH FLAMER: Yes, and I'm going to go back and--

MARIE DOCE: Oh, my gosh.

KEITH FLAMER: So once I retire, I'm going to pick it back up again.

MARIE DOCE: Oh, that's amazing. I hope too someday as well when I retire.

KEITH FLAMER: I get it. I know. When you retire, yeah.

MARIE DOCE: Yeah, oh, that's great. So I played viola under Carol Jacobsen, and then I took German classes, actually at Humboldt State.

KEITH FLAMER: [SPEAKING GERMAN]

[LAUGHTER]

MOLLY BLAKEMORE: You have a lot in common.

KEITH FLAMER: I know a little bit of German.

MOLLY BLAKEMORE: He also is German scholar.

MARIE DOCE: Wow! [SPEAKING GERMAN]

KEITH FLAMER: [SPEAKING GERMAN]

MARIE DOCE: [SPEAKING GERMAN]. Yeah, I lived there my junior year in high school.

KEITH FLAMER: In Germany?

MARIE DOCE: In Düsseldorf. Yeah. There was a scholarship through the US Congress and the German Bundestag. And I applied when I was 15, and got accepted and was 16 and living in Germany. And it was a wild--

MOLLY BLAKEMORE: Wow. That's young.

MARIE DOCE: Yeah, my mom's like, I can't believe we let you just go to Germany as a 16-year-old. I was very adamant. And it was 2005, so it was before technology as it is now. We had Skype, and it worked terribly, and Gmail. I'd just opened my first Gmail account, so I could communicate with my parents while I was there, but it was crazy.

MOLLY BLAKEMORE: We all remember those days.

KEITH FLAMER: I remember Skype.

MOLLY BLAKEMORE: I mean, but you have to really think about the days before technology, when you couldn't just pick up your phone and text from wherever you were.

MARIE DOCE: Video call anyone in the world instantly. Yeah, it's wild.

KEITH FLAMER: So how did you get into AI?

MARIE DOCE: So short answer is I had this career in tech, started in 2010. And end of 2022, I was looking for a new opportunity. I had some friends from my previous startup. I was in the startup world in Austin, and a friend had recently started working at a company called Jasper AI. And they're a generative AI tool for marketing teams.

No one knew what generative AI was in middle to end of 2022, not till the very end. It was a very new concept for me, but I thought it sounded interesting. And I loved the folks who'd gone over to Jasper, and so I signed the offer letter. And then the next day, ChatGPT launched, November 30, 2022.

MOLLY BLAKEMORE: Wow!

MARIE DOCE: I had to look back. I'm like, wow. I signed the offer letter November 29, and then the world changed as we knew it. And January 3, 2023 was my start date and everyone was just so new to this concept. ChatGPT had been out a little over a month. People had tried it. It was just those early days of, what can this thing do? What is Gen AI? Everyone called it Gen AI back then.

MOLLY BLAKEMORE: And now you have a company here in Humboldt, and it sounds like some of those same things that you were doing internally, you're trying to do for your clients here, kind of explain what's happening in AI, how can you use AI, so can you tell us about your company now?

MARIE DOCE: Yeah. Grounded AI, started it in January, and really came out of this inspiration point that I had last summer, moving back, and realizing that things that I took for granted living in cities for 18 years just simply didn't function the same here. Just Wi-Fi calls getting dropped.

I had Google Fiber in Austin, Texas, and it was amazing. And long delivery times for Amazon and food delivery apps not being very reliable, and Google Business pages not having consistent hours. That's just a thing here, and I never experienced that in the city. So just seeing the technology gap and food delivery app is fine. I don't need a food delivery app. I'm happy driving five minutes and picking up dinner.

That's not what raised alarms for me. What did was the rate of change in AI, how quickly things are changing, and seeing this technology gap between our area and cities and seeing that become exponentially wider with how quickly things are changing in AI. So that's how I started Grounded AI. Wanted it to be focused on our local market, our local businesses, and their pain points.

And I just wanted to be a very local first company that would really deeply understand business pain points and operationalize things that could be operationalized using AI. That's a lot of the work I did in tech.

I worked in product management, which is essentially you're the person that decides what changes are going to happen to the software, so you talk to customers, you look at quantitative data, and you use all that to build a product roadmap, and then you communicate that customer value to engineering teams. So you need to have enough technical knowledge to do that, so a lot of what I'm doing now utilizes the things I was doing for as long as I was.

KEITH FLAMER: How do you figure out how a business would implement AI because your job description is awfully broad. And we have a lot of businesses here that have heard about AI, or watch movies about AI, and have no idea how to implement it or merge it into their businesses. What do you tell them? What do you look for to help them understand how they can jump ahead?

MARIE DOCE: I make sure to spend a lot of time just talking to members in their organization, deeply understanding their day-to-day pain points. I don't just rush with a solution before I really understand their exact industry. There are a lot of industries I don't have experience in, so being open and having that humility of hey, I don't know anything about manufacturing, so can you walk me through what tools you use and what those pain points are?

And the customers I work with have been happy to answer that because that's their area of expertise. It's like help teach me and then I will take that and then teach you where AI could fit in, and so it involves more than just a 30-minute phone call. It involves a series of meetings and discovery sessions.

Once I understand the problems or the aspects of their work that are taking the most time, then I provide an implementation plan and a set of recommendations, and then we talk through it together. And so it's a very iterative and structured process, where I'm not just coming in and saying, here, just one size fits all solution. I try to be very custom because every company is different.

MOLLY BLAKEMORE: I would imagine that the companies that reach out to you already know their pain points and have an idea that maybe AI could help them, but they don't really know how. But what about companies that maybe don't think about AI at all? And do you think that they could be helped by AI? What kind of things do you think people don't think about when they think about how could AI help us?

MARIE DOCE: The blanket term AI is applied to so many different things. It's applied to AI generated art, AI generated books. It's used as a punchline. A lot of comedians say, oh, it just writes your emails, and that's the thing that you hear over and over. I don't need something to write my emails. I can write my own emails.

And so I think so much of what I do is just education, and that's why I say yes, you can write an email, but I think-- and then also being honest about my personal stances. If I have a personal email, I'm writing a personal note, I don't use AI for that. I want it to sound like me and come from me.

There are different things that I would use AI for, like automating certain tasks that I have to do every day, manually updating a spreadsheet, using AI for that. And so just giving those examples and talking to folks about it and listening to their concerns helps with a lot of that.

KEITH FLAMER: We are in a really deep conversation at CR about AI. Not just how we use it in the educational realm, but really, how do we prepare students for an AI world, even though if you ask me what I stands for, what it could do, I can't answer that question. But I do know that our students need it not just for their schoolwork, apparently, but also they have to be prepared for the workforce that you're training, so how do we do that? How do we prep students to work in an AI environment in a business?

MARIE DOCE: If I were to think about this as a problem, and the problem I see is AI being used for cases, where it shouldn't be used--

KEITH FLAMER: Such as?

MARIE DOCE: Replacing critical thinking, writing a college essay, summarizing a book that you should be reading as part of a class. There's a whole so many things that I had to do in college that I'm so grateful for learning how to read a novel and then write a critical essay about it. And it didn't sound perfect, but there would need to be a clear set of guidelines around where do we not use AI and why is that important? Where can we use AI?

Oh, I get an update to my email list. I need to add a new email address to this list in this database where it's being stored. Do I need to do that? Do I need to copy paste the email address and to do this rudimentary task? No. Is that taking away my creativity? No.

So there are certain questions we can ask ourselves before deciding, does AI make sense here or not? Is it saving me meaningful time? Is it not replacing my creativity by giving me more time to be creative? Yes, and so that's kind of where I would start with that overarching picture and then go into more details around the things where it could help and collectively decide, OK, we're not going to use AI to replace our critical thinking. Here's why we need critical thinking, here's why we should think for ourselves and not just ask ChatGPT what I think about a book I just read, for example.

KEITH FLAMER: Is that possible? Does that really happen?

MARIE DOCE: I read the--

MOLLY BLAKEMORE: What should I think, maybe--

KEITH FLAMER: What should I think, versus--

MARIE DOCE: I was reading an article by a student who just graduated from Stanford saying that they had moved back to in-person exams with a--

KEITH FLAMER: A blue book?

MARIE DOCE: Yeah, they moved back to that.

KEITH FLAMER: Oh, back to the old days. I remember that.

MARIE DOCE: Yeah, because that's how they had to make sure.

MOLLY BLAKEMORE: Classrooms are doing that, yeah, because we have that problem. We're trying to prepare students for the workforce, and then also prevent them from not learning what they need to learn because they rely too much on these new technologies. So that's what we're struggling with. And what is the proper policy to have--

KEITH FLAMER: Or guidelines around it.

MOLLY BLAKEMORE: --and guidelines.

KEITH FLAMER: I often struggle too about what are the pros and cons of AI? We talk about yes, it can make your life easier. OK, granted, but what are the real pros and what are the real cons? What should one be careful of as we implement AI and businesses in higher Ed, those type of things. What things should we say? Those are things that the third rail never touch it.

MARIE DOCE: I think to start with the pros, I don't think I could run my business single handedly without AI automations. Just the entire process of starting my business, getting it off the ground, filing everything. I get daily updates that synthesize what's in my email, calendar, to do list, and summarize a note for me that I can then quickly read rather than jumping between five different systems.

I have so many customers I'm working with, so it helps me retain more context and not less, but that's because I know how to work within these systems in a way that makes me more efficient. And so that's where I would think as is the vast pros of just allowing people to build and conceive of things that they otherwise wouldn't have the capacity for as just one individual person.

And then it's become exponential. I've been so successful getting to this point, where now I'm finally at the point of like, OK, now I need to hire, but I'm in a good enough position where I can do that. And again, I just don't think I would have gotten there without these tools and utilizing them.

MOLLY BLAKEMORE: Which is interesting because we hear a lot about AI taking people's jobs. And the first part of that is I can do all this stuff without any help. So that does feel like one person doing the job of several people with the help of AI, but that allowed you to then become someone who can offer jobs to people, which--

MARIE DOCE: Exactly, because starting a business from day one, I wouldn't have had the money in January to pay people to do all of the work that I needed to get done.

MOLLY BLAKEMORE: So you become a job creator?

MARIE DOCE: Mm-hmm.

MOLLY BLAKEMORE: And that's beneficial to especially this county here.

MARIE DOCE: And I am very hell bent on hiring locally. I posted my jobs online that I was hiring for and heard from people all over the country, all over the world even. And I would get back like, thank you for your interest. I'm only hiring people who are--

MOLLY BLAKEMORE: How many of those were AI generated responses?

[LAUGHTER]

MARIE DOCE: I know. That's a problem too. And then, yeah, on the cons list, there are so many. And as far as how to deal with them, it's just one hard conversation at a time and being open and talking to folks. There's a lot of discourse online around AI generated art.

And in May, I went to an event to talk about it. It was like a town hall discussion on AI generated art, and I was so afraid to go to this event. I was--

KEITH FLAMER: Why were you afraid?

MARIE DOCE: Because as someone who works in AI, I was afraid they would see me as the enemy. Get out of here. We don't want to hear what you have to say. And I show up to this event. I introduce myself, and everyone's so nice. And you have a great background. I'd love to understand how you see AI and asking me questions.

And these are folks who are like, I will never use AI to draft a social post, and they're very anti-ever using it. And we were able to come together in the same room and talk to each other and there wasn't anything-- nothing bad happened, and it was great, actually.

KEITH FLAMER: I am one of those people, who are anti-AI, and I just realized that it may not be because I don't like the software. I just don't know how to use it. And so I'm wondering how many people are out there like me, who would say, well, I get it, but how do I get it so I understand?

And, I'm going back to the old days, when I bought my first computer. And we had dial up internet. You hear the dial tone, and that was a long time ago. I know that, where I went to college and they had to have these courses that teach you how to use a computer, so now I'm thinking, well, maybe we have to teach people how to use AI.

MARIE DOCE: I had those too, the typing classes.

KEITH FLAMER: Yeah, I remember that.

MARIE DOCE: Mavis Beacon and all that.

KEITH FLAMER: Exactly.

MOLLY BLAKEMORE: Well, if you think about the time frame, 2022 is not that long ago, so the thing is that people don't really understand it. And they've seen a lot of it online, and they don't like it because they've seen a lot of the slop and what it's doing to social media.

MARIE DOCE: And there's just too much content out there.

KEITH FLAMER: Or the mobies like the Terminator.

MOLLY BLAKEMORE: You're right.

MARIE DOCE: There's all how to videos, but it gets overwhelming if you try to go online and say, well, tell me how to use this system, and you find a million videos all about the same thing. What do I trust? What do I believe?

MOLLY BLAKEMORE: And that's only a small lane of AI too, so I think there is that big information gap between what people see and think of as AI and what you do or what you help businesses do.

MARIE DOCE: Most people think-- the amount of the number of times I've heard, I use ChatGPT and you go back and forth and I don't really save time using it because I just go back and forth with ChatGPT all day. And when I finally get the output, well, I could have just done that in a fraction of the time, going back and forth with ChatGPT.

So yeah, next week I'm hosting a workshop with the Eureka Chamber of Commerce, June 24, where I talk about a tool called Codex. Codex, you download onto your computer, it sits on your desktop, and it talks to the files on your computer. It works within a folder, and then it connects with your email, calendar, to do list.

Even if your to do list is an Apple Notes, which mine is, it'll check everything for you. And that removes, I mean, just the copy pasting and re-giving it context. You don't have to do because it already has everything, so if you say, hey, pull up my meeting notes that I had earlier today and let me know what my action items are. It just does it. You don't have to copy paste.

KEITH FLAMER: Really?

MARIE DOCE: Yeah.

KEITH FLAMER: I'm going to have to take that class. That's wonderful.

MOLLY BLAKEMORE: Some people listening to that, and myself included, I must admit, might be concerned about security in something like that, and who is then getting that information? And so that's also a very important point--

KEITH FLAMER: Very much.

MOLLY BLAKEMORE: --that people talk about with AI because they're all privately--

MARIE DOCE: The guardrails. What do you not upload to AI? What do you not give it access to? Don't upload your Social Security Numbers, or a client list of Social Security Numbers to AI.

MOLLY BLAKEMORE: Student information of any kind.

MARIE DOCE: Exactly.

KEITH FLAMER: But now, could AI access those files anyway, even though you don't--

MOLLY BLAKEMORE: If it's on your computer interacting with all.

MARIE DOCE: If you give it access to one specific folder that it can work within and save files in, and then you put files in that folder, so it'd be like a Codex folder that you can reference this. There's an ability to save which folder you're working in when you use Codex, and so that's what you do.

KEITH FLAMER: I'm going to sign up for that class, because I do have that email. And there's something that you said a while ago, and where you used AI when you hire. In terms as you are hiring employees, you wanted to look at certain things, so I made a jump to how do you use AI and human resources as people apply for jobs?

I could see some of the advantages, but I always think of what the disadvantages. Could you talk about what some of those cons with using AI in HR?

MARIE DOCE: Yeah, I think a good example is like Amazon years ago built a proprietary large language model that was meant to scan resumes to make the hiring process quicker, and it rejected all the female applicants. So it would flag certain terms like women in tech and rank them lower, and they tried to fix it--

KEITH FLAMER: Than white men in tech. Is that correct?

MARIE DOCE: Yeah, and certain names would get ranked higher, so there's so much bias. They tried to fix it, they couldn't, so they just had to retire the tool. So automating the human element of looking at a resume and assessing a resume, that shouldn't get outsourced to AI. A lot of tools do that already.

KEITH FLAMER: Do what?

MARIE DOCE: That's what--

KEITH FLAMER: They do what?

MARIE DOCE: There are a lot of HR, AI systems right now. I don't know the specifics of all of them, but I do know the prevailing thing I see online is you apply for a job and you just never hear back, and the thought is that. And I don't know if this is proven, but that resumes are just being filtered out using AI assisted software.

How I would use it in hiring, I would use it to gain insights from an interviewing call, so I could understand how it went. And here are my thoughts, now let me disseminate them. I would use it to craft very carefully worded follow-up emails. I'll usually draft something, and say, here's my email. Can you give me any suggestions on how to make it work better?

That's more of the things I would use it for and not for the actual screening candidates, if that makes sense. I make sure to go through every single resume, and then make a decision on if I'm going to talk to them or not.

KEITH FLAMER: So how can you ensure that the resume that you're reading is human generated versus AI generated?

MARIE DOCE: You can tell sometimes. It's harder to tell nowadays now that people are getting better at using AI.

KEITH FLAMER: Because they're getting much better.

MARIE DOCE: But there are certain tells, and you can see it in emails. Sometimes they're just certain words that are used or turns of phrases. I talked about this on another podcast, but this sentence structure of it's x, not y.

AI likes to summarize things and then make connections where there aren't any, and it doesn't really make a lot of sense, but it sounds like it makes sense structurally. And then you read it and you're like, wait, those two concepts don't have anything to do with each other, but it likes to say, it is this and not this other thing. And when you see it in that framework, you know it's AI, because you're like, no human would make that jump. This doesn't make any sense.

MOLLY BLAKEMORE: And I could see writing a resume using AI because the actual resume part is just about a bunch of data. But what you've done and what your skills are, but the more of the cover letter, the part that's really supposed to talk about you and your essay and what are the points you want someone to see in your resume and that kind of expository writing--

MARIE DOCE: Exactly.

MOLLY BLAKEMORE: --having AI do that, I would have a bigger problem with.

MARIE DOCE: It dampens the actual person behind it, I found. If it's used too much, it kind of genericize content, where it's like, this could be about anyone or anything. I read a bunch of cover letters recently, and the ones that stuck out to me, where it's like, I want to talk to them, are the ones that they weren't perfect phrasing. They were the ones that told a real life story about their life in a way that really hit me emotionally.

And this is why I want to do this because of this that happened, and that takes a real person writing that. And maybe they used AI to clean up the phrasing, and that's fine. There's a big difference between using AI from the get go to create everything, versus I have this draft, can you review it for me and edit it for clarity and phrasing, which is how I tend to use it more.

MOLLY BLAKEMORE: The interesting thing, I just was thinking about this when you were talking about the bias in the Amazon hiring model, is what I read was that most large language models are like that because of her historical writings, because there is so much bias in all of the things that humans have written over the centuries. That's what it's learned is that white men are better than--

MARIE DOCE: Exactly, it's trained on the internet. It's all of the internet came in. Think about all of the opinions--

KEITH FLAMER: Nice points. Good point.

MARIE DOCE: --Reddit threads. It's a really good point, and it's so true.

MOLLY BLAKEMORE: It comes from us. I mean, it really comes from us.

MARIE DOCE: We made it biased.

MOLLY BLAKEMORE: We made it biased.

KEITH FLAMER: So how can we unbias? How can we make it more equitable in terms-- how can we do it right?

MARIE DOCE: How can we unbias--

MOLLY BLAKEMORE: Re-train it.

MARIE DOCE: --when humans are inherently biased, because it doesn't take out that there is still bias in hiring regardless of AI because the bias, to your point, started with humans, so how do you fix that?

KEITH FLAMER: So are you saying that we should take the human element completely out of the process and just--

MARIE DOCE: Create a magical, unbiased model.

[LAUGHTER]

MOLLY BLAKEMORE: Yes, it's a thorny issue.

KEITH FLAMER: It is.

MOLLY BLAKEMORE: It's a wicked problem, as they say.

MARIE DOCE: And so something we were talking about, this was a little over a year ago. I was on LinkedIn, and all of these women in tech started posting AI generated images of white men saying, I gave ChatGPT my resume, took out my name, and said, create an image of me based off my resume. And so I did that, and it generated a generic looking white guy--

MOLLY BLAKEMORE: Wow.

MARIE DOCE: --because of my background in tech. And obviously, it wouldn't do that for every field, but because tech is so male dominated, that's what it's gathering. So if you're in an industry that's more weighted female, if you're a male nurse, and you're like, draft an image of my resume, it'll most likely draft a woman.

And so again, it's bringing in all the data from the internet. Every organization in tech I worked at was minority women, especially in more of the technical areas like product management, engineering. It was anywhere from 10% to 20% women compared to, so many times, was the only woman in the room.

KEITH FLAMER: It reminds me of a bit of our conversation about the Unmasking AI. Talk to us about that book, and what you learned about that book.

MARIE DOCE: I read a book called Unmasking AI by this incredible scientist named Dr. Joy Buolamwini. She has her PhD from MIT, and it was while she was there that she was testing this video sensing technology. And she saw that as a Black woman, it didn't even recognize that she was there.

MOLLY BLAKEMORE: So video, what kind of technology?

MARIE DOCE: It was like not motion detector technology. It was like facial recognition technology. Yeah, so when you go on your phone and it instantly knows it's you. It was that technology, didn't recognize that she was even there. And then she had to literally get a white mask and put it on, and then it recognized her face.

And so just that erasure of entire groups of people by technology because, again, going back to that point, the people building these systems aren't trying to be biased, but say you build a facial recognition software and everyone in the room is a white person. They're only going to test it on themselves. They're not going to look at other use cases, and that's how you build a biased system.

KEITH FLAMER: So what does that tell you or us about our society? Honestly, because I have my opinion, but I'm curious about what does that lesson mean to you? What do you learn from that?

MARIE DOCE: It means that we still have a very long way to go in that technology is-- there's a risk that technology deepens biases rather than equalizing because everyone talks about technology as being the great equalizer.

We'll all have access to the same technology, and we'll all be up leveled and upskilled, but the picture isn't as rosy as the tech industry would like you to believe, and so I think people do one of two things. They either say, well, I don't want any part of this. This is terrible, and I'm just never going to use this technology because I'm not represented.

[LAUGHTER]

KEITH FLAMER: I got you, which means--

MARIE DOCE: And I've spoken to so many people who feel that way.

KEITH FLAMER: So what you're saying is that I have to change my attitude.

MARIE DOCE: Well, not necessarily.

KEITH FLAMER: It's OK.

MARIE DOCE: The education piece, just learning about the technology is important for everyone, whether they use it day-to-day or not. You can learn about it, and say, OK, it's not for me. I'm not going to use it, but, at least, you're opting out from a place of knowledge, and not I just unilaterally hate something that I may not understand fully. And so yeah, that's one way to go about it, and I see that a lot.

And then the other places, I want to be a force for change in the industry, and that may be overly ambitious, but I just think that the more underrepresented groups involve themselves and bring a chair to the table. And if there isn't a seat at the table for me, I bring my own chair, and I sit down anyway. And I make myself a part of the conversation.

And that's what I did with Grounded AI because every founder I worked with, I worked in startups. The first startup I worked at was in 2019. Every company before that, every CEO and founder was a male, and so I never thought I could be a founder. Honestly, I didn't think that. I just didn't fit the bill of the typical founder archetype that you see in greater mass media.

I'm not a tall, white guy. I don't have unbridled confidence in myself. You can't really as a woman. You have to be questioning yourself. If you're too confident, it's a little bit like--

KEITH FLAMER: Oh, I see.

MARIE DOCE: --she needs to tone it back a little. And so just I didn't see myself in that. I didn't have the right founder personality as my own self-imposed bias. And then going through this experience with Grounded AI, I learned I can take the parts that I really admired in these male founders I worked with, because so many of them were so great, and they had so much passion for what they were doing.

And the best ones I worked with had a clear way to articulate their vision in a way that could emotionally resonate with people, and that's something that I like to think I can do as well. Even if I'm more introverted, I feel very passionately about this, and I like to talk about it.

And then the other things. I don't need to be unbridled confident 100% of the time. I can say, hey, I don't understand this. And I do think that that can in its own way, be an asset. When I'm going to a business that I don't understand, I'm not going to walk in and say, oh, yeah, I know everything about manufacturing in your industry. I'll say, I don't know, teach me. And there's no one size fits all way to be a startup founder, I've learned.

MOLLY BLAKEMORE: I would say that AI has a real image problem right now because of those unbridled egos that are running all of the AI businesses.

KEITH FLAMER: I've never heard that before.

[LAUGHTER]

Just kidding.

MOLLY BLAKEMORE: And you see it at these college graduations, where people bring up AI and they're getting booed.

KEITH FLAMER: Just recently with Google.

MARIE DOCE: Yeah, they walked out. I saw that. I know.

MOLLY BLAKEMORE: It happened at several of them, where these young people are seeing their futures almost, or what they imagine is their futures being stolen away by AI. And same with these huge data centers that are going up everywhere, and the big protests against those data centers and the environmental issues around AI.

And so I would say that, yeah, AI, as much as we use it every day and people don't realize they're interacting with AI pretty much every day, it still does have a big image problem, so how can you get people past that to see the benefit? I mean, I think we're kind of asking the same questions a lot, but how are you working with people to try to get them past that idea of it?

MARIE DOCE: Yeah, listening is such a big part of it. Understanding where the controversies primarily are in being open and clear about my own stances as a person and as a company. I know AI generated art is very controversial, and it's not a primary facet of my business because I feel very passionately about local artists and I want to support the work they do.

And yeah, we had open studios a couple weekends ago, and I only have my work Instagram account, and all of the artists here locally are on Instagram. And I'm talking to them. I'm buying their art. They're like should follow me on Instagram. I'm like, yes, I will. And it's this awkward moment of like, please don't hate me as soon as you see my Instagram is like, it's my business account.

So these kinds of interactions will happen, and I understand why artists feel so strongly about AI because of the way it impacts their world. Just like college students, I understand why they feel so strongly against AI because of the way they've seen it impact their day-to-day for the negative, and there's just such a huge difference in AI in those situations, artists, college students, writers compared to businesses that need their operations streamlined.

MOLLY BLAKEMORE: Anytime I'm listening to the radio or NPR, I'm hearing all these ads for agentic AI, and people might not know what that means. I mean, we have the LLMs and Gen AI, and now we have this agentic AI, and that's actually a lot of what you're talking about. And can you explain what that is.

MARIE DOCE: Yeah, so an agent is basically a concept in technology, where you have a task, you have guidelines around that task, and then the agent completes a set of tasks in the background, so you're not seeing every single thing that it's doing.

And so to give an example of an agentic workflow, I would say, well, there's smaller tasks like I have to generate a newsletter every two weeks, and I have a specific format that I want my newsletters to be in, certain structures that need to happen. And I will give AI here's all of my ideas, brain dump. This is what I want it to talk about. These are the articles that I think are really important, and then it'll format it in a way for me.

I'm not looking at it doing all of that. I'm just saying, here, use your newsletter skill. A skill is basically a task that an agent does, and then it completes it. And it creates a Google Drive file, it links to it. It's like, here it is for you to review.

Another more complicated one would be like, say there's a complicated estimation process for a physical manufacturer, and there's a lot of steps that you have to go through to estimate what a new project is going to be. An agent would go through the steps and retain all of that knowledge and data, and then complete it in that way, then you review it.

And I say the review part because that is so important to me. When people talk about agentic workflows, they're like, oh, it's just going to start sending all your emails and it's just going to start scheduling things for you. And, at least, the way I operate as a business is that human in the loop, and that human reviewer piece is very important.

MOLLY BLAKEMORE: And so the agent would be for people who are familiar with Excel. Would that be like when you create an Excel function? But you can send the agent out to do it.

MARIE DOCE: You could have the agent update Excel for you. Like say, oh, I have these new-- I closed business for this client and this client, this client this week. Can you update my Excel prospect tracker for me with these numbers? And if there's any additional data you need, come back to me for that.

And then it'll say, OK. Yes, I just updated these rows. It looks like this one's incorrect, which it does. It's flagged things for me before. You listed 3,000 here and it's 2,500 here. Which is it? And so it'll do audits for you of spreadsheets, documents.

MOLLY BLAKEMORE: But in macros is what I was thinking about, so you can create macros. An agent is something you create, so you tell it all these things that you want it to do, and then you save that agent. And then you run the agent whenever you need that task to be done.

MARIE DOCE: Usually, the process is I have to do something. OK, I did it now. And this is within Codex. You're in the chat. OK, I just did x. Now can you create a skill for me? And it'll create the skill for you, which is a .MD, a markdown file. It saves it in that folder on the desktop, and it's a certain skill of like, write a newsletter, or draft a social media post, audit this spreadsheet, and then you can have it run on a schedule.

I have a Business Health checkup every Tuesday. It runs. It looks at all of these data sources, and then tells me where I'm tracking my business based on my KPIs close business. And so it'll do all that on a schedule.

KEITH FLAMER: I'm going to switch gears a little bit because there's something that you said, and I want to explore this a little bit. And I'm going to paraphrase, is that when you first got into the AI business and tech business, you were often the only female in the room, and you self-regulated what you say.

So I'm wondering if you were to have a group of young girls, young women, who's going to look to you for advice, for mentorship. What would you tell them?

MARIE DOCE: My first thing is, if something doesn't feel right, speak up. It's risky in corporate environments, but necessary to say, if I'm not comfortable with something that was said, or something that happened, or if I witnessed bias happening against me, or another person, I say it.

And that's scary, and that's a concept that a lot of people have a hard time with, but as far as the way I operate, not saying anything is more harmful because then the systems just persist. And a lot of times, as women, or any other underrepresented group, the onus is put on you as the underrepresented group to be the one to speak up, to be the one to make the change happen.

And so I'm not saying that's the answer either. It shouldn't be all on women. You need to find people that support you, who aren't biased because they're not all bad, and build a community and a network of people that are like-minded. And I like to think that that's what I'm doing here in Humboldt, and I found some really incredible people through StartUp Humboldt, SBDC, the Eureka Chamber.

The fact that community is so important here, you're not just an anonymous person, your personal and your professional reputation matters so much here, puts us on a different level as far as our expectations of one another, and that's a good thing.

MOLLY BLAKEMORE: Do you feel like it's getting better in tech for women and minorities?

[LAUGHTER]

KEITH FLAMER: That's a very good question, I'm wondering. No?

MOLLY BLAKEMORE: Sounds like no. That's unfortunate. And what can we do, maybe in our capacity as a higher ed in higher ed to combat that, to make it better for people?

MARIE DOCE: Yeah, and keep in mind, I'm speaking as just one person and one. There are good organizations out there. There are other female founders, who hire women and support women. There are male founders, who support women, and all of that exists. And the good people out there exist.

And just giving students as much visibility into those companies and those people and showing the range of founders and startups that can exist, because for me, going back to that, I only saw male founders, so I just didn't see myself in ever being able to have a role like this. So the more examples you give to show that diversity--

KEITH FLAMER: So you are an example? And so you're making a difference because you're now a founder.

MARIE DOCE: Yeah, I'd like to think so. Yeah, I'm still in the early stages.

KEITH FLAMER: You are.

MARIE DOCE: Thank you. I appreciate that. No.

KEITH FLAMER: That's wonderful.

MARIE DOCE: It means a lot. And I have a daughter and a son, and I want them both to grow up with me as an example. And I try to lead in that way with everything I do so that they can see what women are capable of, and it's important.

KEITH FLAMER: Thank you.

MARIE DOCE: Thank you, both.

KEITH FLAMER: I really enjoyed our conversation today.

MARIE DOCE: Me too.

KEITH FLAMER: OK, now I'm becoming more of a fan of AI than I was prior to talking to you, so thank you.

MOLLY BLAKEMORE: Keith's going to have you look at his computer and put all the agents on there now.

KEITH FLAMER: I'm going to have to first figure out what an agent is in AI--

MOLLY BLAKEMORE: And then we'll have to get that--

KEITH FLAMER: But once I figure that out--

MOLLY BLAKEMORE: --by our firewall.

KEITH FLAMER: There we go, so I'll give that a shot. So thank you so much for this.

MOLLY BLAKEMORE: Yes, thank you for--

KEITH FLAMER: We appreciate it.

MOLLY BLAKEMORE: --talking with us.

[AUDIO PLAYBACK]

- One small step for man.

[MUSIC PLAYING]

[END PLAYBACK]