Artificial Intelligence: Real Talk

jethro_1_07-23-2025_101746: Welcome to Transformative Principal.

I am very excited to have on the program today, Derek DC Krieger.

Uh, he's a passionate advocate for making artificial intelligence approachable and accessible.

Particularly for individuals within the Neurodiverse community.

As someone who is autistic, Derrick has dedicated himself to fostering a supportive environment where people with autism, A DHD, and dyslexia, can leverage AI to enhance their skills and income.

Through his work at Practical ai, he emphasizes clarity and support, ensuring that users can navigate AI without being overwhelmed by its complexities.

He's got a great history and this wonderful community practical AI that you, uh, can read a little bit more about on the show notes at Transformative Principal dot org.

Uh, but Derek, DC you prefer DC So we'll go with dc Excuse me, I keep saying Derek.

Uh, welcome to Transformative Princip.

derek-crager---practical-ai_1_07-23-2025_131818: Well, uh, thank you Jethro, and honestly, is, uh, fine.

Uh, Derek in a crowded room, people.

Hear Daryl and Eric and anything else.

So, uh, but yeah, my name's Derek, uh, dc either one.

I, I, I respond to, so thank you.

jethro_1_07-23-2025_101746: Excellent.

Well, great to have you here.

Talk to me a little bit about this idea of using ai, particularly with Neurodiverse people.

What.

What's the difference for someone who's neurodiverse as opposed to just any regular Joe off the street?

derek-crager---practical-ai_1_07-23-2025_131818: Well, I, uh, I think the, uh, the experience is different because our histories are different.

Um, if we, uh, to throw a, a couple terms out here, you know, neurodiversity on one side and neurotypical on the other, uh, the neurotypical is, uh, is that category.

We throw the,

jethro_1_07-23-2025_101746: I.

derek-crager---practical-ai_1_07-23-2025_131818: uh, traditionally average people to, you know, air quotes here, quote unquote, average.

Um, but normative, I guess is, uh, is, uh, a better way to put it.

Uh, a average always sounds like an insult, so, uh, when it's applied to humans.

Um, but yeah, the, uh, the experience coming in with AI is that, uh, uh, it gives us someone to talk to.

And, um, connecting through a human level is important.

And I'm even talking, uh, beyond, you know, uh, keystrokes and clicks.

It's, uh, if we can get to that human, human conversation with ai that's, um, uh, one thing I do with, with my company, both on the profit and the non-profit side, is I built a voice interface.

To emulate this, uh, patient supportive anthropomorphized ai, that, that sounds like a human.

So I think that's important to get through the psychological barrier.

either way, um, those of us that, uh, that are on the spectrum, uh, or considered neurodivergent, um, we have, uh.

Honestly a, a rough history.

Um, we are the, uh, lowest employment rate when it comes to, uh, to employment statistics.

And a lot of times we do good things, but, uh, we can't explain 'em well.

So that means we, we don't connect to the social.

So, um, AI allows us to speak with ourselves in a manner.

Um, there's.

Multiple platforms out there, whether it be Notebook, LM or uh, Chad, GPT project or perplexity spaces.

Those are just three off the top of my head that allow us to, uh, create a diary for ourselves and be able to talk to ourselves.

So, uh, in that manner, it's um, uh, those of us that are a DHD, we're either at 50,000 feet or we're deep in the weeds and nowhere in between.

So I know it allows me, uh, to.

Uh, what I'm deep in the weeds to talk to myself through the, uh, AI interface and say, Hey, gimme a broad view.

'cause I'm, I, I'm stuck on something.

Or even just the opposite.

If I'm at the 50,000 viewpoint and my mind's going everywhere, racing and I can't get into the weeds, or I can't identify the top five or the top 10, or what are the highlights?

Then I can ask myself through, um, this AI.

Front end diary, you know, to, uh, uh, to elicit those, those items.

So it's, uh, it's a way to, um, get advice from ourself and, uh, not, and especially we, we won't have the fear of asking a human.

I mean, in school, what's one of the first things we always hear at the beginning of the year?

You know, there are no dumb questions type of things.

But even still, um, those of us that are on the spectrum of some sort, we've.

We've heard that and, and we still get experience that our questions are dumb, so we refrain even deeper and go on our shell.

jethro_1_07-23-2025_101746: Yeah, that makes total sense.

And in, in working with my own daughter on this kind of stuff, she, she sometimes doesn't have the capacity to articulate the question that she's asking.

And so, for example.

This morning at breakfast, she was, she was asking a question and we couldn't understand what she was asking.

And she was trying to refer to something in the past and instead of saying Last Christmas, she said last week.

And so then we were more confused about that situation.

And there's, there's this paper that came out recently called Your Brain on Chat, GPT, cognitive Debt.

And this idea of cognitive debt has really been.

Weighing heavily on me in that the paper proposed that essentially people who use chat GPT to do things that they are taking on cognitive debt.

But in reading that paper and starting to understand some aspects of it, I believe that chat, GPT actually provides cognitive equity for some people because it gives them an ability to, to do the things that.

Typically get in trouble, uh, cause them trouble in their day-to-day communication.

So, for example, for my daughter to put together complex words and phrases can be very challenging, but she can get the base idea down.

And so using chat, GPT actually enables her to have a voice where she didn't have a voice before because it helps her explain things where she's just getting the highlights and it helps explain them and fill in the gaps.

Does that make sense?

What do you see with that kind of an approach?

derek-crager---practical-ai_1_07-23-2025_131818: Oh, that's, uh, we're definitely in alignment on this.

It's, uh, what is there, depending on, uh, the research, there's 30,000 or 60,000, uh, words in the, uh, English language and, um, a typical human, you know, we're walking around with five, maybe 6,000 words.

So your word for something may be different.

my word for something, and we write this off sometimes as cultural, and yeah, the culture has a certain, um, uh, connection to it and the reason why somebody might call a
couch at one place, but a Davenport and someplace else, it, it's, if we don't know there's that if translation, then yeah, we're like, wait a minute, what is a Davenport?

And, uh, so yeah, the, uh, the, the way your family, uh, speaks it, it's the same way for.

For your, your daughter.

I, I picture that.

Uh, and first of all, I wanna preface this by saying that anybody on the neurodiverse, uh, spectrum, um, in that neurodiversity group, uh, just because we have diagnosed with the same professional diagnosis with whatever that might be.

That does not mean that we're the same as somebody else sitting next to us with that same diagnosis.

I think that's very important to understand.

That's something that I didn't even understand until, uh, 10 years ago, and so.

there's something that, uh, uh, Dr. Nancy Doyle came up with.

Uh, she's a doctor out of the UK and she built this, uh, community called Genius Within.

And she talks about what is neurodiversity, and she identifies, uh, the spiky profile.

And to her, the spiky profile is that, um, it's like an XY graph and we plot out our skill sets and the neurotypical.

People in the world, they're like a plus one to a minus one or a zero all the way across, but they come out to be that average.

You know, there's that word again, average.

Um, but that's normal.

And so for the, uh, those of us that are, you know, neurodiverse, we might, um, be a 10 neurotypicals are a one.

But to balance that, we might be a minus 10.

Where a neurotypical is a one, and that plus 10 might be in our creativity.

It could be building something.

It could be our analytical mind, and then maybe we come up with something great for a company or a school or a project or, or for mom and dad.

And then we try to explain it.

so like your daughter is trying to explain, maybe we're a minus 10.

And because we lack the control over the words and to some of that, it really, that lack of control over the words is the interface with the risk of humanity or humankind.

And if, if that's our only method of communication, then it sounds like whoever's trying to communicate is ignorant or they lack the knowledge where sometimes it's just the opposite.

Sometimes there are so many things going on in that mind.

That, uh, they might know 10 words for that topic and they're just gonna throw one out at you because their gears are turning so fast.

They're just spitting to put the, uh, language together.

And in their mind, their mind is running so fast that they're thinking about 10 things ahead of the conversation.

So to put them in a scenario where they have to stop, you know, smell the roses type of phrase.

There are all kinds of phrasings in life that we pick up.

Um, you know.

Take a minute, take a heartbeat type of things.

Um, live in the present.

Uh, for those of us with our minds where the gears are going, it's very easy to use words that are conversation partners don't understand.

So it is very much like a, uh, interpretation needed and chat.

GPT does offer that.

Uh, definitely.

jethro_1_07-23-2025_101746: Well, yeah, and that's something that's really interesting.

I, I like that genius within approach because in my daughter's situation, she may be a minus 10 in a whole bunch of areas, and then a plus 10 in only one area, and.

You know, it it, when we talk about this, we often talk like we're really good in one area that makes up for the deficit in another area.

But it may be that we are.

Negative in, in a lot of areas, uh, as it relates to what is typical and what is normal.

And that makes it really difficult to assimilate and participate in our modern economy.

And the, the things that we do on a day-to-day basis in our society that can make it really challenging.

And one of those things is being able to read.

For example, um, even, you know, with all this stuff around ai, so much of it is text-based and so.

It, you mentioned having a voice interface that helps.

And then you also mentioned the idea of this patient, uh, infinitely patient thing that can deal with the millions of questions that are asked.

And, and that is exactly something that, um, that I've seen with, with neurodiverse.

Diverse folks that I've worked with, that, that they have a lot of questions and ask them repeatedly and constantly, even when we, we already have determined that these things are, are the case and happening.

They still come back to these questions often.

And so having something that is infinitely patient can be a really powerful support for someone.

And so my question then with, with these contexts in place, how, how do you do it so that.

The, the AI is actually beneficial for the person and that it is attuned to what it is that they need and that it, it can help them in their specific use case.

derek-crager---practical-ai_1_07-23-2025_131818: Oh, certainly it's, uh, that, that's, an important distinction because we hear, and we read the, of the headlines that talk about AI being inaccurate or hallucinating, et cetera.

So we have our special sauce on the back.

End of our model.

And, uh, our model is, um, is set up where we take the knowledge domain, whatever that knowledge domain is.

So we, we compartmentalize a, a specific knowledge.

For example, let's say seventh grade science.

Maybe we take the seventh grade science book and we digest that into a database.

You know, vector database is a cool term to throw around.

Um, but, uh, we digest that into a database.

We have an AI on the back end that interprets the data, and then we have that transition point.

We have a, a different, uh, um, large language model on the front end to interface between the human the, uh, the back end.

And, uh, so it's really about can.

Putting the knowledge in a container and allowing our AI to speak and from that knowledge domain.

Nowhere else.

So we put guardrails on it.

And so the conversation when we're talking to it is very much like speaking to somebody that, uh, is on the spectrum because we set this up as a subject matter expert, this SME in a box, and it talks ad infinitum on that subject.

But if you ask it about, you know, curly fries at Arby's, it's probably not gonna be the answer.

Yeah.

And it, uh, versus the public domain, uh, tools that are out there.

You're just gonna get the first answer that comes up.

That's really Chad GT's model is to just return an answer.

So, yeah, that's, uh, it's, it's important that we stick to the quality of the data and we increase the accuracy of the conversation by limiting our interface with just that specific.

jethro_1_07-23-2025_101746: So how does this work then in a situation where there is, you know, let's say we need this to be a helper for navigating the community, for example,
and uh, someone doesn't have the language skills to get out there, but they can text and then can the AI then be their voice and say, I want to go to this place and it will.

Communicate that for them.

What, what kinds of situations where exist, where, where you can do more than just teach a subject, but can also help them navigate an environment or a conversation or a relationship.

derek-crager---practical-ai_1_07-23-2025_131818: Yeah.

Well that, that's ideal and that's really our forte.

That's, that's where we thrive is, is going beyond just the steps because.

Uh, we can have a, a list of ingredients on a recipe, and it tells us how to bake muffins, but that doesn't mean that that index card can tell us what to do if our oven is oversized or it's outta whack, or we we're missing an ingredient.

Um, and then in the, uh, in the.

How about this?

Um, if we have, uh, our earbuds in and we're listening to our map, let's say Google Map, that's walking around town, so that would be similar to the scenario you're, you're talking about in a geographical sense.

Um, the, uh, the, uh, Google Maps is going to respond on just straight how to get there, but you can't ask Google Mass.

Um, you know, the road's blocked or the, the bridge is washed out.

Um, I mean.

got some workarounds in there, but it's not conversational.

Whereas if they were, if, if your daughter was speaking to you the telephone, she has her backpack on, her earbuds in and she's talking to dad and she says, Hey dad, I got down here and I, I can see the McDonald's across the way, but the bridge is washed out.

Um, then dad would say, oh, yeah, yeah, yeah, just go down to First Street, take a right.

And that's the workaround.

And that's a very simplified version of.

Of what our AI is doing right now, conversationally.

So any place that a human can be and offer mentorship, that is, um, that's the domain we're covering right now.

We're not trying to be all things to all people, but in that scenario, uh, we take the topic and we are working on.

Um, business side.

We have school side on specific topics, but also, um, I learned when I, uh, I'm not a certified teacher.

I'll, I'll, I'll say that to the crowd here, but I know that there's a shortage of teachers.

So in, uh, in what time I have to spare, I volunteer at the local, uh, as a, uh, substitute teacher.

And, and I found out that, uh, depending on the school system, you have, uh, adults that chaperone or mentor.

In individual students, one-on-one throughout the day.

They walk around with them, they go with them, they help interpret, uh, like we discussed a little bit ago, and they help interface and they help, uh, um, uh, that they go better for that student so everybody can learn.

In the school system without disrupting the class.

So a lot of money.

If we had one adult for every student, I mean, we'd have, we wouldn't have 25 to one teacher to student ratios, right?

So if we had that economic, uh, capability, we'd be supporting the teachers much better.

Um, so with this, uh, with this, and I saw this a couple years ago and it started my mind turning on that.

your scenario, can we give somebody this, this virtual mentor that they can carry around with them and it's, it's not even a, uh, a visual handicap at this point.

So it doesn't separate them and call them out for being different.

Oh, you got the human walking with you.

Um, it's in their ear and they can say, Hey, I'm going to math class and I really have, I really have trouble with math and I don't know where to start.

I don't know how to talk to the teacher about it.

Can you talk me through this?

in a very supporting, caring way, um, it will say, Hey, you know, try this.

And then depending on the knowledge that is contained, um, there's different scenarios that it will walk the student through on how to connect it.

Uh, we could have that knowledge programmed with, um, you know, what's my locker combination?

I forgot.

and then I. I was listening to your episode last week with, uh, uh, you and Damon talking about software and I think, uh, I think it was
Damon that was going into, um, we have this knowledge loss when a teacher leads and the teacher takes all that knowledge of the student.

And then we have a certain amount of records, but those records are really cold and, and they really don't tell us about the student other than really their attendance and, and very, a very high level view.

And I know a large part of your discussion was on.

do we share a spreadsheet and what happens when we fat finger the spreadsheet and who has access to this knowledge?

And, and, uh, so I envision taking the tool that, that we built and going beyond just supporting one person at a time, but in, in a scenario where, you know, we're not.

the privacy concerns are taken care of, and it's all within the school system.

How cool would it be for, you know, little Johnny or Jill to come into a class in, in, in second term and the teacher doesn't pull up a record and look at this, you know, black and white ones and zeros
about the record, but what if they could just dial a phone number and said, hey, um, you know, I've got, uh, Johnny and his, his twin sister Jill in here in from, uh, you know, the next borough over.

Um, can you tell me about 'em?

And then conversationally, they're gonna be told about the highs, the lows, what they're having trouble with, what they really excel at.

And I, I tell people, Jethro, that our AI technology is 50,000 years old back in the day before, you know, Gutenberg press and before writing instruments, um, we told stories.

And it was human to human.

That's how we learned.

So we're going back to the basics, the true basics of having that conversation and we're scaling it, and we're giving the user this one-on-one time with this dynamic knowledge base that they can use when it suits them.

They don't have to set an appointment, they don't have to call in at a certain time.

Um, but all that information is there when they need it.

jethro_1_07-23-2025_101746: Yeah, I just, I love what you're talking about DC because that's exactly where I want to get to, where the support is there.

But the human still remains in in the conversation.

Now you talked about going into the math class and saying, I really don't get this, and one of the real problems we have with students is that they.

They don't articulate that.

They act out with poor behavior.

And, and that's typically how we find out that kids are struggling in school, is that they have poor behavior to start.

Now, if someone cannot articulate that they're struggling with math, this is an area where maybe that, that phone call the teacher makes can help communicate that.

Maybe not, but if there's a way for them to.

Be able to get the help that they need, uh, without acting out behaviorally, then that would be ideal because that then lets everybody focus on learning
and not focus on dealing with these negative behaviors that crop up when kids are doing things that to get out of the work or to avoid the difficult work.

And, and this is a really important thing to all teachers and school leaders because.

That's usually what happens is that it becomes a behavior issue, especially if the kid can't communicate effectively what's really going on.

And this is, this is a huge problem in education that if we can find the solution to it will be, uh, it'll be a lifesaver for everybody, the kids, the adults, everybody involved in the process.

derek-crager---practical-ai_1_07-23-2025_131818: I agreed a hundred percent teachers, gosh, put in 12 hour days for a seven hour class today, and, uh, they're doing work before, during, and after.

And you would think it translates into a one-on-one time with a student.

But, uh, I think on one of your episodes, uh, uh, saying I think, uh, Alaska has like five different state tests and they're on different platforms.

So, uh, so much time we teach to the test just out of, you know.

We're, we're kind of offhand as force uh, that we don't have time to actually teach.

So, um, yeah, absolutely is, uh, with a, with a conversational, uh, tone, um, we can iterate back and forth through conversation much quicker than an email or, uh, records or maybe I make a phone call.

And are these records correct or do you have more records about this student?

And, um, there's, you know, there's students with parents.

There's students without parents.

There's students that, uh, have, uh, helicopter parents.

There's those, uh, students with, uh, parents that, uh, work all the time.

So we, we have no idea what the student, support system is back home.

And, uh, absolutely if, uh, we need to work and offer every possible avenue of communication, um, because those, uh, sometimes we just need to be heard.

And I, I point to.

You know, Maslow's hierarchy, you know, it, uh, um, we have certain needs that need to be met before we can even exist.

And I believe that, uh, um, being heard is, is a very strong need in, in human civilization as an individual.

And if we're just heard that dispels and deflates and depressurizes, a lot of those volatile situations.

jethro_1_07-23-2025_101746: It, it definitely doesn't, and people need to be heard no matter what, and a lot of times.

Kids in schools seek to be heard by their behavior when the other things they're trying to do aren't working or they don't have the capacity to do that.

So, um, this has been, this has been a great conversation.

How can people, uh, get connected with you and with what you're doing with practical ai?

derek-crager---practical-ai_1_07-23-2025_131818: Oh, certainly.

So I have, uh, a nonprofit side, uh, it's called AI for Autistics, all spelled out AI for autistics.com that tells about a community that we build.

Um, autistics have the, uh, the lowest employment rate we, we may have touched on earlier.

So we guide and we teach on how to use, uh.

AI to build software.

And I know that came up in, in your conversation last week too.

Um, and, uh, I think you said you use repli to build, but there's several other ones out there.

So, uh, this a great tool for, uh, for those analytical, lean minded, which a lot of us fall into.

Um, so we teach 'em how to build things, and that's a no cost.

It just, it just is there.

Um, on the for-profit side, the bigger realm of practical ai.

Uh, app.

So just like it sounds phonetically practical, AI apps, apple pair, pair.

And if you, uh, find my name on LinkedIn, please reach out, say hello and I'd love to connect and, and help uh, everyone in your audience.

jethro_1_07-23-2025_101746: Yeah.

Very good.

Well, this has been awesome and I know we just barely scratched the surface, so thank you so much for taking the time to chat with me today.

And, uh, please, everybody go check out.

Practical AI app, uh, or uh, AI for autistics.com.

Thanks again so much, Stacy, for being part of Transformative Principal.

derek-crager---practical-ai_1_07-23-2025_131818: Well thank you and uh, thank you for having me here.

Have a great day.