Sep 21, 2026 · 36 min · 15 segments
**About the Episode** In this episode of Inspired Changemakers, Julia Healey sits down with Svetlana Makarova for a timely conversation about artificial intelligence, human expertise, and leading…
Svetlana MakarovaGuest
Julia HealeyHost
And I think that's where we really connected is because I was like, hey, listen, I'm building all this technology right now for United Charitable, right? And I know that I need to like think about AI, but how do I think about it? Like, how do I think about it in this sense, in that sense? And you kind of broke it down to me.
So let's talk about maybe AI from two different concepts, from maybe two different buckets.
And so this is where a lot of the companies that are currently creating a lot of, I would say, excitement in the space is how do you put... your existing data to use, and you have big data science teams who are actually developing these AI solutions from scratch.
Most of the market, when you hear AI actually speak, they're using already deployed models, so things that you don't really necessarily have control over how these models are trained.
Companies that are using it inside their organizations, they're using these black box models because they're licensing the models, and they're using them on top of their own data.
and so and there's we kind of spoke about this a little bit before we started so because you don't really control the model you're licensing it from another vendor you're putting in your data how do you know are they using it to like improve these models so that's a big other question and um there's ways around it but and i think you have to kind of understand this are you talking about building these these tools you have more control what goes in how you train how you maintain there's more transparency in these models But a lot of the market are kind of using these black box models.
You talk about Perplexity, Gemini, Grok, you have ChatGPT, you have Claude.
It's biased towards pleasing you and telling you exactly what you want to hear.
And one of the more common use cases over time, you think it's, you know, there's a report, I think it's by Stanford that they've published every year, is like, what are the most common use cases for these chatbots? And the growing leads become therapy.
And so in 2026, or I think 2025, most common use case for these AI chatbots is therapy.
And because you're building, it's easier to build trust and you don't you're not going to get judged by this agent.
And it's going to actually tell you exactly what you want to hear, be constructive in its output.
And I think that's where we really connected is because I was like, hey, listen, I'm building all this technology right now for United Charitable, right? And I know that I need to like think about AI, but how do I think about it? Like, how do I think about it in this sense, in that sense? And you kind of broke it down to me.
So let's talk about maybe AI from two different concepts, from maybe two different buckets.
And so this is where a lot of the companies that are currently creating a lot of, I would say, excitement in the space is how do you put... your existing data to use, and you have big data science teams who are actually developing these AI solutions from scratch.
Most of the market, when you hear AI actually speak, they're using already deployed models, so things that you don't really necessarily have control over how these models are trained.
Companies that are using it inside their organizations, they're using these black box models because they're licensing the models, and they're using them on top of their own data.
and so and there's we kind of spoke about this a little bit before we started so because you don't really control the model you're licensing it from another vendor you're putting in your data how do you know are they using it to like improve these models so that's a big other question and um there's ways around it but and i think you have to kind of understand this are you talking about building these these tools you have more control what goes in how you train how you maintain there's more transparency in these models But a lot of the market are kind of using these black box models.
You talk about Perplexity, Gemini, Grok, you have ChatGPT, you have Claude.
It's biased towards pleasing you and telling you exactly what you want to hear.
And one of the more common use cases over time, you think it's, you know, there's a report, I think it's by Stanford that they've published every year, is like, what are the most common use cases for these chatbots? And the growing leads become therapy.
And so in 2026, or I think 2025, most common use case for these AI chatbots is therapy.
And because you're building, it's easier to build trust and you don't you're not going to get judged by this agent.
And it's going to actually tell you exactly what you want to hear, be constructive in its output.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.