Ep 66: From Copilot to Confidant - Exploring AI's Role in Emotional Support

Watch the YouTube video version above or listen to the podcast below!

Full Episode Summary

In episode 66 of Enterprising Minds, the trio—Dave Dougherty, Ruthi Corcoran, and Alex Porkorny—explores an unexpectedly human side of artificial intelligence. Ruthi opens up about a deeply discouraging day at work, including failed demos and feelings of professional isolation. Lacking peers to vent to, she turns to AI tools for emotional processing and problem reframing.

This sparks a rich dialogue about how generative AI (ChatGPT, Claude, Gemini, Copilot) is evolving beyond task-based usage into a kind of reflective journaling and emotional support tool. Dave shares his experience using AI after a difficult meeting, leveraging voice mode to rant and receive constructive feedback. At the same time, Alex discusses how different modes of interaction (voice vs. text) shape his use of AI for emotional or technical support.

The conversation also dives into how prompting AI forces clarity in one’s own thinking—a kind of digital journaling that helps users work through professional and emotional roadblocks. Each host shares how different tools serve distinct roles: ChatGPT for creative tasks, Gemini for deep research and image generation, Copilot for productivity, and Claude for nuanced advice. There’s even a comparison of AI tools in terms of their research integrity, with some tools fabricating sources and others being more precise.

On the technical side, they touch on AI memory, personalization through projects, prompt engineering, and even AI’s limitations in source citation and visual generation. The group critiques platforms like Perplexity for their scraping practices and praises tools like Adobe Firefly for ethical and high-quality visual generation.

Closing on a lighter note, Ruthie updates the team on finally installing Gemini and experimenting with it for household tasks. Dave teases a past story involving AI and toilet repair, reminding listeners to subscribe to the newsletter for more.

This episode is a thoughtful reflection on how AI is becoming not just a tool for work but a companion in how we navigate our lives, careers, and emotions.

Ep 66: From Copilot to Confidant - Exploring AI's Role in Emotional Support Podcast and Video Transcript

[Disclaimer: This transcription was written by AI using a tool called Descript, and has not been edited for content.]

Dave Dougherty: Though welcome to the latest episode of Enterprising Minds happy Thanksgiving for anybody in the US and anybody who wants to celebrate Thanksgiving back at you. Ruthi has two separate ideas for. So today I'm just gonna let her start and we're gonna have the discussion. So Ruthi, what's up?

Ruthi Corcoran: All right, we'll kick us up with the first, and if we've got time for the second one, we've got one in our back pocket, so

Dave Dougherty: Sounds good.

Ruthi's Rough Day and AI Support

Ruthi Corcoran: I had a pretty rough day yesterday. Multiple things just either blew up in my face or projects that I had been working on for months just. Failed really rough, like two rough demo meetings in a row. If you've ever had a call where you're presenting and then there's just silence and it's not the good kind.

It's not wow, I just blew this out of the park. We're doing awesome. This is the like. No one is speaking because. They have either checked out or they have very negative thoughts and they're side chatting. It was these types of things. This doesn't happen so often, it happens every once in a while.

It happens to all of us. And boy, it was a rough day. And at the moment I find myself in a curious situation where. I don't have straight peers. I have, I'm in a mix where I'm leading a group of folks. And I play that sort of leader role in, in many places. And I'm on a team though, where I'm new and I haven't built rapport with my equivalent peers in the hierarchy.

And so that's a weird spot to be in. And it'll evolve over time. It always does. And I was thinking to myself, gosh. Wish Alex and Dave, like I could just teams chat them, right? Because I'd have a lot, I want to bounce ideas off of somebody. I need a place where there's, yes, Dave is waving, can call his phone at me.

And and so I was thinking, gosh, this seems like a perfect situation for.

Using AI for Emotional Support

Ruthi Corcoran: Being able to talk through with chat, GPT or Claude or Gemini or copilot in the case where I gotta talk about like real world situations just to think it through. And it's work related, so it's proprietary. We gotta keep it in house and.

And I was, I sent you guys a message just to say, Hey what, which is the tool that does this best? And what's funny, I'll just add this too. What's funny is I think we had talked it previous in the show how we were surprised how many people were using Chas UPT in a therapy situation. And here I found myself in a, this isn't really something that I, I.

I'm gonna take the time out of Dave or Alex's day and part of it might be I've got internal stuff that I can't talk about. But then it's, is one of these aI tools better than others at helping you think through the problems. And I'm not talking, trying to solve like how do I build a dashboard problems, but help me reframe, help me tease out what's happened and get to a different spot, which is a thing that I have found useful in other contexts of help me reframe a problem I'm working on.

But this was in a more human emotional context. And so two, two thoughts and then I'll open it up. One is I was surprised to find myself thinking this. because I was like, oh, this is a shift in the way that I've thought about ai. It's mostly been task-based that I've used this.

And so that was a surprise to me. I guess that was, there was only one point, not two points. What do you guys think?

Personal Experiences with AI

Dave Dougherty: Okay, so first question Alex, have you used an AI as an emotional support buddy? Whether personal work,

Alex Porkorny: maybe in some way. I depends on the depth that you're talking about. I've definitely used them for like random advice at times. Did something recently of using voice mode on chat, GBT. And then having it like quiz me with questions into a kind of a q and a response kind of thing.

So I definitely have done some stuff like that. Probably more technical to be honest.

Dave Dougherty: Okay. That's very on brand for you, so that's fine. Yeah.

I will say for me, I have used. Not in the full therapy kind of way, but like I did have a bad meeting where we were talking about it was related to like career progression and like how things are supposed to happen, supposed to being key word there. And I found myself like, I can't focus on this.

I'm too. Too worked up. So I went for a walk and I went way out of the way to find a meeting room where it's like, all right, nobody's gonna know me. And luckily it was late in the day, so a lot of people already left. And it, I did have the weird experience of having to, in the very prompt and I was using voice mode.

So this is the other thing, I'm talking to the air. But then prompting it. I just had a bad meeting. Let me rant for a little bit and then pick apart the pieces that actually matter.

Okay. Yeah.

Whereas, because like I know I'm too worked up about this to actually see. The real things.

Alex Porkorny: Yeah.

Dave Dougherty: Right.

But then in having to prompt, it forced me to then work through, okay, here's what my expectations were. Here's how this meeting did not match up with my expectations. What are the go forward things that we can then do? Also don't come back with me with overly positive things. I'm not in the mood right now.

Those are excellent questions. No, screw you gt. I'm not in the mood. Like

Ruthi Corcoran: you bring up just two really important things. So one is, I think what you described with the rant and tell, talk back to me. What is the importance things is what I was looking for, which is I need a mirror that helps me see this in a different light or more objectively or helps me move forward.

Dave Dougherty: Yeah,

Ruthi Corcoran: where are my opportunities for moving forward? Because the scene was, I see those, I can latch onto them. Like I don't it, I'm not having, I'm not wanting to stew in the trouble. I just, I can't quite figure it out on my own. And so that's a really cool, and the voice to text piece, I think probably makes a big difference because it you have a different way of thinking through.

AI Tools for Work and Personal Use

Ruthi Corcoran: But then that brings me to my second piece, which is I have noticed especially in the last few weeks, the types of projects that I have been using copilot to tackle. The writing of the prompt forces me to be much more clear in what it is I'm trying to accomplish and i'm sure we've talked about this in the, in previous episodes, but that's such a powerful tool.

And if you're. If you don't have the habit and the practice of journaling, particularly about work items, this is a mechanism through which you're almost doing that right? But in a very productive towards what you're trying to do way at work

Alex Porkorny: there, there's that quota. A problem written down is 80% salt.

That's, that does get to it. It's actually interesting. Between the two of you of Dave, I totally would. I agree with that of voice kind of speaking out what's going on. But I would want to read the response actually. I would want to shift really, and that would be a faster way for me to digest it and skim through it and get to the next point, which would then maybe go back to a voice prompting or just typing in and kind of moving on from there.

So it's interesting too, of a lot of that between the two of you. That, that's framework. You're able to take yourself enough out of the situation to understand what you need to be able to break it down to actually be able to ask for that, versus just being like grabbing a random coworker and being like, we need to take a walk.

I need five minutes. Like I, I don't even know what on earth just happened. And you're still in the processing moment of that didn't go well, it didn't feel good, and I'm stuck on those feelings right now. Like I haven't gotten anywhere past, that moment, you're still at that, first, first stage, right?

Yeah. There is.

Dave Dougherty: There is a certain underlying self-awareness that is necessary.

Alex Porkorny: Yeah. In order for these things to then into context.

Dave Dougherty: Yeah. So that was the thing for me where I was just like, look, I know myself, I'm gonna be upset about this, but gimme like 20 minutes to just rant on it and then that steam will be gone.

Sure I might be tired and I don't want to deal with it. And that's fine. That's what tomorrow's for, but yeah. But then also like with that conversation around, not just about Hey, talk me off the ledge kind of thing. Which that's extreme. But also, like even with jobs, I've found, this is my perception of my skills and what I bring to the table.

This is how I talk about myself when I am going through job interviews or like face-to-face things or, those sorts of things. These are the roles that I want to go after. Am I a fit already or do I need to change my narrative around how I'm presenting myself, my skills, my whatever. Are there things I should bring more forward?

Should I bring thi, quell things down? And it was an interesting process of doing that over time because I realized. You know what? I haven't been giving myself enough credit for the music background that I have and the amount of work and the amount of entrepreneurship that is inherent in that that is applicable to other things.

And then it's my job to actually sell that idea and give myself credit for it so that other people do as well. I know that's a side tangent piece, but I think it's definitely related to the the other thing,

Ruthi Corcoran: and I think what I what I noticed too is it is really valuable if you have a friend who can give you honest feedback.

But that's hard. Like that is a very rare, difficult thing to find. And even if your friend does give you honest, true feedback sometimes there's a limit to how much another person can say and retain the relationship. Whereas what's interesting about this scenario is you're saying, Hey, give it to me.

You're not friends with the ai, so there's not so you might be able to. Get a larger list or potentially more poignant feedback as a result. Now, of course, it's gonna be missing a fair amount of context in the day to day and how you interact with people, so you're not gonna get everything right.

And the second thought I had is one of the values I've seen in working with copilot on different projects is just getting a very different perspective than anyone on my team or that I bring to the table. And sometimes it's super helpful to get something that you wouldn't. Normally run into to just balance out what kind of perspective you have.

because it could be, you're sitting in a bias bubble over here.

Dave Dougherty: And the idea of AI being the average of everyone's knowledge. I've never been big on crowdsourcing information ever. But if you take an average of everything that's out there, that's at least. Halfway decent floor to test yourself against.

At least for bias. because there are some very strong cultures For sure that you want to be like, all right. This company is mostly engineers. I'm trying to do something creative for the love of God. Help me describe this in engineer speak.

Ruthi Corcoran: Yeah.

Alex Porkorny: Yeah. Yeah, there's some interesting points there also of AI's use.

There was this is part of my Minnesota nice background, but there's an Ethan Mullick article that really helped me because he needed a, like a blog post title. So he asked for 25 and then asked for another 10. And I like used to feel bad being like, oh man, I should ask for this much, of an AI tool.

Gotta get over that to just keep on hammering at it. But that's a really interesting point here too is also basically is it has unlimited patience, it has the understanding that you give it. So context is everything, like a therapist in the first session. I wouldn't expect a whole lot out of it.

Maybe after a few sessions, I would expect something out of it. The first one's just an intro, so similar thing of the first chat is gonna just be an intro and then eventually we'll get into something that's, probably a little bit more, better than the average.

Speaking to your point of the average response. Yeah, you'll get that the first time, but it's not that great. But also trying to be able to ping this thing constantly whenever you need it. I found to be very helpful too, because taking like a work example, like the number of different random projects.

I spend a little bit of time on that. I'm in a meeting, then I have a little bit of time for this and then this question on this one, and then whatever else I'm able to keep going through those different chats threads. So just okay, here's the next step requesting this, that step, okay, that didn't work.

Try this now and then go to a different chat and try something else with that one. So then I got a couple different ones going and that's able to keep that frame of reference. A lot longer, which is helpful. That

Ruthi Corcoran: is a superpower, like

Alex Porkorny: Being able to like, have those different ones run or being able to have that access to that or what as aspect do you mean?

Ruthi Corcoran: Access to that and having it retain the context to be able to pick it up? Or help you re-pick it up.

Dave Dougherty: Right.

Ruthi Corcoran: I, there's something there that's even a step deeper, and I don't know exactly what it is, but that's, that is something that an a, a, an extra teammate can't do.

Alex Porkorny: Yeah. And then there's also projects in chat GPT if you use those.

I use it because I have a personal content ended up using at work. So I have kind of split up my projects personal and works of different projects. And so it has memory across both, but it basically focuses more on that particular project and the information that you've given in that project.

So that way, at least, like when I have the kind of personal discussions, which. It might be random stuff like, I don't know, woodworking question or some random thing that I might Google, but instead I use chat GBT for it. Like all that kind of stuff goes into there. So the context there, even kind of the personality of that response is a little bit different than the one that I get from my work project.

So it starts to like also deviate a bit. So one tool, slightly different response types based on the context that's been given and the history there.

Comparing AI Tools: ChatGPT, Gemini, and More

Alex Porkorny: So Dave, you mentioned though, when you were speaking and you got this response, which one, which tool were you using?

Dave Dougherty: So I was using chat GT for that one.

Okay. I, it's interesting, I don't, I. My experience with the Google Gemini live setting. I think because I'm so familiar with Google and I'm used to searching in a particular way. Yep. Using Gemini in other ways feels like a stretch to me. But I also find. This will be an interesting discussion.

How are you guys using the various ais? Because I find myself leveraging chat GPT much more for this podcast for general like creative things because I've built custom GPT for that. But then I'm also testing. Basically, anything I do in chat GBT I'm also doing in Gemini, but I find myself going to Gemini for more basic knowledge stuff.

I trust it more with deep research like it. I. Just like the answers, it provides better, I think it goes to better sources on the internet than GPT does. Like GBT is, feels like the type of person that would just give you a list of Wikipedia references. And as somebody who grew up in a family of educators, that is unacceptable.

I get it. Yep. Yeah, I've been leveraging those things. And then I use Google Gemini for for the images for the newsletter, the header images for that. And I cannot get SOA to do it as well as Gemini. And I think part of that is because it was founded as a multimodal model from the get go.

It wasn't LLM plus this other thing. Okay, now stitch them together, right?

Yeah. How are you guys finding yourself using them? Which ones are you using?

Ruthi Corcoran: I Useche, GBT for all of my random. Questions like, give me the cook time for this thing on my air fryer, or translate this size chart of how big the shirt is and what the measurements of the shirt are.

Translate that into what actual size I really need, like all those types of things. JDBT. I have only just recently started using Gemini because I actively know I want to explore it and see how it's working. So I don't have an answer for you yet. But that's a start. But then you're using

Dave Dougherty: copilot at work,

Ruthi Corcoran: and then copilot is, that is the bread and butter of my day to day.

Dave Dougherty: Interesting.

Ruthi Corcoran: And I don't even use chat bt at work particularly anymore. If I have a, how can I frame this better with copilot, occasionally I'll use chat BT for that. because I find it helps me create better copilot prompts. But the recent upgrades within copilot have been sufficient to be able to do what I need to do.

Dave Dougherty: So for your personal ones, are you using a free or a paid version?

Ruthi Corcoran: Paid.

Dave Dougherty: Okay. Paid. I'm assuming none of us is paying $200 a month for the top tiers?

Not yet. Okay.

Ruthi Corcoran: But go fund us and we'll tell you how it goes. Yeah,

Dave Dougherty: absolutely. Exactly.

Alex Porkorny: So I personally paid license to chat Chi pd. And that one is used extensively on myself with my phone at work I built for my other employees GBTs that create certain types of content that's templated out, brand standards, that kind of stuff. Got a couple of those now that are being shared in the company.

So those are kind of private to me, but shared by a link, so that's able to be shared out. So that one's pretty extensive. A lot of technical questions, like all day long chat, GBT basically. There Claude, I've been testing out recently just because I've been hearing a lot of podcasts with some of the creators of it.

Basically long story short, basically is a group of individuals who were at Open AI and thought that they weren't really taking security and safety seriously. So they split off and made their own company philanthropic. So they made clo. And some of their mentality behind it of having a pre-trained model with a set of rules and then giving it the information versus just having a model that's just given The information definitely ends up being, there's some quality standards and safety aspects there that it definitely does a lot better.

So I've been liking it. Just been playing around with it just recently a little bit more and found some of the responses. Speaking of your Wikipedia kind of thing, you mentioned it a group chat with you guys, but had the same response to Claude and Chad GT and Chatt gave me basically like a standard, solid kind of friends advice kind of piece, which was good, but it wasn't very specific and I gave it to Claude.

And Claude was like, by week three do this and you're pronouncing results in week five, do this. And I was like, oh. That's a whole different level. That was a lot better. However, I had to keep checking it because it keep kept making mistakes even within a single response, it would, conflict itself well.

And Joe and I have used a lot for deep research and it has been fantastic for that. Speaking of that tested those exact same prompts against Claude and Chat, GPT and Perplexity and Gemini a hundred percent. If I was gonna put something together that I wanted to stand behind at work, that would be Gemini.

Chat, GBT, even when I really push it for citations, I follow those links and the content is not there.

Dave Dougherty: Yeah,

Alex Porkorny: it's making it up, like the links are legit links, but it by no means cites that particular, phrase that it just used. There's no way that it's backing up any of this.

Dave Dougherty: I had something like that where it was citing, company YouTube video. So I clicked into the YouTube video for that particular phrase. And it wasn't, it was a company overview video that they made for like CES or something. That's, and I'm

Alex Porkorny: like,

Dave Dougherty: what?

Alex Porkorny: Yeah, no it a little bit better recently. Actually this probably in the last week or two, modified one of my GPTs.

And that a now cite sources within our websites, public facing files. And so a list of set of citations for all the different, like things that comes up. There's a landing page builder, blog, post builder, some other ones that are like that. Having things that are factual, can have claims that are non-factual.

So that's actually been really good. So lately Decents the history has been really bad. Gemini also can take a really long time with it. Steep research.

Dave Dougherty: Yeah.

Alex Porkorny: If you use the thinking model with chat, GBT there's a selector of different models. You're commonly thrown into chat GT five, which is a filtration model.

So it tries to look at your career and tries to guess join. You spend a lot of resources on this or a little, most of the time it goes little. Which is why there was a lot of backlash when it first launched. Then if you switch it though, you can have it just always be thinking, but then you get always a five to ten second, maybe even 22nd pause before you get a response for anything.

So can be a little frustrating. I'm always using thinking work. What's that? I'm always using thinking. Yeah, same. I said it to that just because I use it a lot at work with like technical tutorial kind of things of I need to do this, show me how to get to this part of this particular platform, right?

A MarTech stack questions and stuff like that. And the straight, regular one. I kept jumping around and missing things, missing labeling things and when you're trying to like, literally look for something on a page kind of has to be accurate. It's the tutorial. So that's been a little frustrating to play with.

So I can't say any of them is perfect for everything. That's kind of the problem is like, to get to Ruth of your kind of original question of which AI tool would you use to decompress from a day? I guess the one that has the most content most context, that's the one that I would use, but I don't know if any one of them actually can do it.

Perfectly well. And then one random note, just so way we can have this discussion.

Image and Video Generation with AI

Alex Porkorny: because I'm curious, Dave, your thoughts on this is perplexity and image generation? I did a test, you guys know about this like a few months ago and my gosh, I got some like AI spaghetti soup kind of mess of like imagery back from the exact same prompt and then perplexity came back with something that was.

Almost perfect. Like it was crazy how different, Claude was like, it looked like a child's drawing of like squares and circles. There was something that somebody created with the kind of the basic word shapes and PowerPoint or something, or in paint, and then, google had something, but it made no sense.

There was elements that were there, but it made absolutely no sense. And then perplexity was amazing. Yeah. But there's a lot of concerns around perplexity and I can say it definitely from a scraping standpoint. They scrape sites like crazy and they used third parties to get around rules. Yep. They were definitely not playing by the rules.

That's why I don't use them. Yeah. And then I just had this thing with Reddit and chat GBT, because I. Saw a design of a thing that I want to build and I was like, I like some of the elements of this particular build. I want to incorporate that in our current plan. So I posted the link and said, oh, due to rate controls, I can't access that.

And then it continued and then it said, okay, looking online, it uses these particular materials and it's everything that was listed within that post. So somehow still live, Chachi t gets around the Reddit rate controls as well, so there's. I guess I have questions on a lot of these things. I know Reddit has straight con contracts with Google, at least I think one other, so I have some concerns with perplexity of using that, especially for anything.

I don't know. Image wise,

Dave Dougherty: I stopped using perplexity when I heard a number of interviews with the CEO. I haven't seen anything with them. It was clear to me, and this is my opinion, go do your own research. But in listening to him talk, it was very clear that I was like, okay, we're gonna build this.

I'm gonna sell it, and then I'm gonna sip, my ties on the beach, and it was just the I don't care about the laws because if we get big enough, we'll pay the fines and. I'll still get what I want anyway, which is like such a nasty part of San Francisco culture that I just hate. And so I'm like, if you're that outward about it, screw you, I'm not gonna support you.

I'm not gonna use your product. I'm not gonna I won't recommend you to any friends that are gonna be doing that. So I am not the one to talk about. Perplexity, I'm not surprised that you got a better image generation from them because of their correct. And stole everything. Laissez faire. Yeah.

Yeah. Around copyright. I will say for me, image generation is a lot better on Gemini. Now granted, I'm doing, I'm not doing very specific things. With it, right? It's, here's this newsletter. Give me a couple of options that fit the themes of said newsletter, right? Yeah. So instead of finding a stock photo image, I now get, an AI one that's a little more tailored.

And that's just what I need it for. The way the prompting changes, I find is interesting, right? Because with the media piece, you need to have very specific references of I like this style of a camera, I want it at this angle. I want, yeah. These elements to be captured. That's very different from, I have this thought, help me with this text, which, we end up talking about a lot more.

So yeah.

Alex Porkorny: Yeah, I need to do a lot more video generation stuff too. I've not even really gotten into that and at least for a work purpose of even like a few slider kind of gifts kind of things for just banner ads. I haven't even done that. Like I've done a couple random tests, but not I'm in need of a music video.

If you want to. How project Whole music. Yeah, I can do it six seconds at a time. Yeah. I hope it's a really short song. No,

Dave Dougherty: No. I tried a little bit, I did a little bit of a trailer or something for a song release, and that was my last attempt with video generation. But I was leveraging runway runways.

Really good. That was before Gemini released their new video editing model, I forget what it's called, like flow or VO or whatever it is. But yeah, and so it becomes interesting because if you only get a certain amount of time per generation, I find myself going okay. I'm gonna go to chat, or I'm gonna go to Gemini.

I'm gonna have it help me create the storyboard. Of the kind of shots to match the theme, to match like the story of the song, to then create the various scenes that we can maybe jump cut or reuse some of that. Or like, all right, maybe I should get a couple of shots of, the artist playing the guitar or instrument or whatever.

And then go back to these sort of, display things that like reinforce the imagery of the song. Then you go to the image generation and say, okay, here's the overall project. Here's the individual scene I need you to create. And then you see what you get. And that really hard part with that is making sure that you have characters stay the same across the scenes.

Sure.

Alex Porkorny: Yeah. S gotten better recently, so Good. Forward to trying that. The other one I was gonna just throw a quick shout out to is Adobe Firefly.

Dave Dougherty: Oh, yeah.

Alex Porkorny: Work related kind of thing. Just because that you get kind of more legally protected access to your imagery that you've created. Versus some of the other ones it's a little bit more questionable.

Dave Dougherty: And what it's doing with backgrounds or like expanding from a narrow shot to a wide shot. That's just awesome.

Ruthi Corcoran: Yeah. As well.

Dave Dougherty: Yeah. Cool. I don't know. Ruthi, did we answer your question?

Ruthi Corcoran: I think we had a lovely conversation about my question, which is exactly what we should be doing.

Dave Dougherty: Okay. So what's your, that's wonderful.

What's your next step? Are you gonna download Gemini and try a live thing? Because I have fixed a toilet with Gemini Live. It is possible. I have the app

Ruthi Corcoran: on my phone now. Okay. It was, it took me over a month after you told us about. You're a handy thing. Yes. And I have used it to figure out how the heck to use our irrigation system.

So far so good. I'll keep you updated. Okay.

Dave Dougherty: Yeah. For those of you who don't know the toilet story, go subscribe to the newsletter. It is on there. Pathways by Enterprising Minds. Thank you for sticking around. I hope you enjoy. Your time with your family if you're in the States, for all of you German listeners.

Have a great weekend and yeah. We'll see you in the next episode. Take care.