Reporter Daniel Walters joins Val to talk about AI and his recent articles on the use of LLMs in Spokane politics and media, including at RANGE.

. This is Free Range, a co-production

of KYRS and Range Media.

2

I'm Valerie Ozier, editor at worker-owned

news outlet, Range Media, and I'm here

3

with one of our reporters, Daniel Walters.

4

Hello.

5

Yay.

6

Who for the last few weeks has been

shoveling more than a million words into

7

an AI detection machine to figure out

what the state of AI slop in Spokane

8

is, and so we're gonna talk about that.

9

Daniel, can you tell me why did

you start out on this journey?

10

Yeah, I don't know.

11

It's always so

12

I think AI is hitting different

institutions and different careers and

13

occupations in radically different ways.

14

Some people, like coders, it's like,

"Wow, I can just do everything now

15

and I can just, enter in things,"

and, for mathematicians it's oh,

16

they're doing all the work for you.

17

For writers- … it's a not just a

… It's like almost a philosophical threat

18

beyond- … just an actual threat to jobs.

19

So it's not the- … it's like hitting

the funding, it's hitting the purpose.

20

Yeah.

21

It's hitting the thing that we've been

trained all our lives to become good

22

at, to create meaning through this

space that is pretty sacred for us.

23

So we … as writers, we tend to think

about a lot of this idea of oh, your

24

words are your words, and if you're

s- using somebody else's words that's

25

not on your staff it's plagiarism,

and it's the kind of thing that

26

could destroy a journalist's career.

27

But more and more with the advent of

AI, it seemed to have increasingly

28

subtly kind of snuck in- increasingly

to a variety of locations.

29

You beg- you read something and be like,

"Ah, that kind of looks like AI," but

30

you didn't have- … any way to prove it.

31

Now we have a way to prove it effectively.

32

Or at least pr- give very

strong evidence there.

33

And it's using AI.

34

So we learned

35

I was reading a lot of coverage about

some of this stuff, and used to be

36

that basically there's, there … Oh,

there … For a couple years there's

37

been AI detector software- … and

it's always been pretty bad.

38

In the sense of- Yeah … like sometimes

you'd have a an em dash punctuation

39

mark that- a lot of writers like to use.

40

Including Range a lot Or you'd, yeah,

you'd use word likes delve and, some

41

of these words that, that AI writing

like ChatGPT would rely upon, and

42

it would just trigger basically.

43

And so it was really … effectively

they were- Pretty inaccurate.

44

It was like the, 25% of the time

there'd be a false positive-

45

effectively for some of

these- … these older apps.

46

Pangram was different.

47

Oh, wait, and yeah I'll chime in there

with the older apps 'cause, … I saw in

48

your story you mentioned one that they

used when I was in high school, which

49

was it was to check for plagiarism.

50

Now I'm blanking.

51

Was it Turnitin?

52

Turnitin, yeah.

53

Yeah.

54

And so I'm like, wow, that

thing is still running.

55

Yeah.

56

And so it's it's basically a lot

of them were like- … and what

57

Pangram was like a relatively recent-

… addition to this, really really

58

got going extensively n- last year.

59

And what they ended up doing

was they ended up having this

60

big data set of human writing.

61

And then they effectively mirrored

it by creating essentially AI

62

simulations of that writing with the

same kind of prompt "Talk about this

63

topic for the same number of words."

64

And it's called basically they

were mirroring effectively.

65

So basically they were- … like

taking this and creating an AI

66

version of the human writing.

67

Okay.

68

And then they're basically analyzing it

using AI and computer software and some

69

of this large learning so like kind of

some of this deep learning- … aspects

70

to figure out what are the differences,

even the really subtle differences-

71

between human writing and AI writing.

72

Yeah.

73

And so sometimes there's all these little

tendencies where if it's just oh, AI is

74

7%- … more likely to use that word.

75

So that's gonna…

76

If they're using that word, it's gonna up

the possibility it's AI just a little bit.

77

But in aggregate, all these little

whispers of signals become this

78

roar of- obviousness once you end

up combining them all together.

79

And so that's what it ended up

doing, and they've tested it

80

in their own internal things.

81

The older model was like, it only

said that human writing was AI writing

82

once in a fully human document was

an AI document, once in 10,000 times.

83

The last ver- the latest version,

Pangram 4, that's come out, it's

84

basically only gets it wrong, at least

according to them, once in 24,000 times.

85

So it's really rare that it, something

will be untouched by AI- … and it'll

86

suggest that it's an AI w- written thing.

87

If it's a something of a

l- large enough length.

88

So if it, we're talking about you

know- … 200, 500 1,000 words, then

89

we're talking about enough of a data

set- … enough words that it can

90

really do a very good job of telling

you, has AI been mucking around here?

91

Interesting.

92

Yeah.

93

And it, this tool is- It's just

solely looking at patterns and

94

s- kind of statistical values?

95

I- is it, or is it also utilizing…

96

I know that there's some type of like

digital watermark type things that can be-

97

it's not using the digital watermark- Okay

… aspect for, I don't think it is for,

98

Like it does that for images sometimes.

99

Yeah.

100

So like there's an image software

too, so I put like- a, a candidate's

101

illustration, little kind of AI through

that, and I'm like thinking it's

102

gonna give me this percent chance, and

it didn't give me a percent chance.

103

Oh, it's AI 'cause there's an actual

watermark that, Oh … OpenAI Chat-

104

Yeah … the ChatGPT software just

added there to let you know- Okay

105

It's AI.

106

Yeah.

107

Some software like some tools like

Anthropic, which I think creates

108

Claude- … is like there's basically

been some attempts that we're

109

gonna add some little subtleties.

110

Yeah.

111

Like intentionally have some writing

ticks in here to give you a watermark-

112

this is what this is what Anthropic

does- … and this is what Claude does.

113

But in this case, it's not.

114

Yeah.

115

It's just looking at how does the, the way

that an AI talks and writes effectively-

116

… how is the way that AI writes

different from the way a human writes?

117

And I, I'd say it's really impressively…

118

like it's not…

119

I'll say it, it doesn't, if it's,

if it shows, something shows up as

120

100% AI, it doesn't mean that they

just typed it in and it's they typed

121

in "Write me an essay"- … and that

they just copied and pasted that.

122

Sometimes it like, sometimes they might

be messing around with it, sometimes

123

they might go with- different versions.

124

I think there's some real dispute

about how deeply AI-ridden

125

the 100%- … mark is from it.

126

But basically it just means that like

according to Pangram- Like all the

127

story has, like throughout the entire

story, there's like- … symbols

128

and s- and signifiers of AI writing.

129

And so we tested it on our own writing-

… that we knew wasn't AI 'cause we wrote

130

it, and- … we all came up clean.

131

And then we ended up testing it by-

It's like- Yeah … like a drug test.

132

Yeah, exactly, and we just wanted

to see is this gonna like- … with

133

our habits and the way we

write, is this gonna like- Yeah

134

Show up?

135

Especially- And then- … like we're

prolific em dash users since before

136

AI became the, the em dash user.

137

Yeah, and not only that, but not only

that, but I am a person who writes

138

things like, "Not only that," as a way…

139

Like I would say

like- … "Not this, but this."

140

Yeah.

141

That was like a habit that I had developed

a kinda little writing tick that I had.

142

So these are like both things that

the AI can use, overuse sometimes.

143

So the question was is like, are we

gonna get flagged for just, that-

144

Yeah … aspect of our writing?

145

And the answer is no.

146

And in fact, a really fascinating

thing was is that we ended up

147

taking I ended up testing it out.

148

I'm basically like I had for

like I think one of our writers'

149

columns, I ended up, Lauren, one of

her- … columns on I think, biking.

150

It was probably something on biking.

151

Yeah.

152

I basically had it like I said, "Hey,

ChatGPT, write me up a paragraph

153

summarizing this story," and then

I- put that paragraph in the middle

154

of the story- … and ran it again.

155

And then it basically detected,

oh, this is all human except

156

this little paragraph here.

157

Oh, yeah.

158

This like paragraph, this

like, 200-word little piece.

159

That looks like it reads like it's AI.

160

That's so cool.

161

And so it actually like and maybe there's

a couple like lines before- and after,

162

but basically it like detected it.

163

It was a similar thing where

we ended up running a column

164

that Aaron Hedge had written.

165

Not a column, but a s-s- a news story

Aaron Hedge had written about Spokane

166

Police Department officers wearing

masks, and it came up as like there's

167

just a little bit of AI at the end.

168

I'm like, "Oh, what did Aaron do?"

169

And then I looked- … oh, it

was 'cause we like we copied a

170

full statement that we'd found,

like record requested an email.

171

We got an email- … from the police

chief, and that was what came up as AI.

172

And that to me showed that okay,

this is a, this does a pretty

173

good job of detecting this.

174

And even when we're like saying "Hey,

Going through, do a mild co- asking

175

AI, do a mild copy edit to this.

176

It didn't really end up

shifting it very much.

177

Even when I had…

178

I was writing an intro for this, that

I was like halfway done with writing

179

the intro for our section- and I asked

ChatGPT "Hey, finish up this intro."

180

So it was like, half-written.

181

There's a bunch of garbage notes

that the- Yeah … coming in.

182

And then it wrote it doubled

the word count- … and it

183

wrote something about it.

184

It did a lot of changing, but it ended up

saying it was 60% AI as opposed to all AI.

185

So it's just, it was impressive

the way that it wasn't really,

186

it wasn't something that was very

sensitive to just, oh, there's

187

just one phrase- … here or there.

188

It seemed pretty robust to not suggesting

that this was just, Yeah … the hu-

189

human writing was fully AI writing.

190

Interesting.

191

Yeah.

192

And when you talk about those those

specific tests that you did with

193

the inserting little- … things

and changing it and trying to have

194

it copy edit where you were i- in

your stories, you were we had we, we

195

published three different stories.

196

One was our methodology.

197

So basically running through

what Daniel just said.

198

Another one was about Spokane

politicians who are running for office

199

and their use of AI within their

statements within their advertising

200

and campaigning and all that stuff.

201

And then the other one, the other

story was about AI use in the Spokesman

202

Revi- Spokesman-Review's op-ed pages-

which is where, community members

203

and leaders in the community submit

stories f- or articles, columns.

204

Not articles, columns.

205

To, comment on current events, right?

206

For the, the younger listeners out there.

207

So when you were finding out or

when you were searching, what was…

208

Or sorry, what was the first story

that you decided to put into Pangram,

209

and what made you put it in there?

210

The first…

211

I'm trying to think.

212

What was the first story

that I was looking at?

213

I'm trying to remember.

214

I'm trying to remember

what, which one I tested.

215

Which one I tested first, and it, I

think it might have been a, I don't know

216

if it was, I don't know what drew me

to testing the Spokesman-Review- site.

217

I know that I was just curious

about, certain things- … how they

218

show up, and I am sometimes curious

about some of the different…

219

i, I had talked about- … when we had

done our questionnaires and I had talked

220

about even before our questionnaires-

… came out, so we basically asked

221

all these politicians- "hey, we're

gonna ask you 10 to 20 questions."

222

"We want you to give us an

answer in less than 200 words.

223

Give us all these answers

to all these questions."

224

I was joking or s- half joking "Hey,

we should all run this through Pangram.

225

We should see if these- yeah

226

are AI."

227

And so then I think I ended up start,

then I started, actually did that.

228

So I subscribed did that.

229

I was like, oh, I put some

voter guide stuff through there.

230

I put some of our some of that stuff

through there, and I was like, oh wow,

231

this is showing up as quite a bit of AI.

232

Yeah.

233

And there are some, but there

are some people who didn't end up

234

responding to our questionnaires.

235

And I didn't wanna punish the people

who did respond to our questionnaires-

236

Yeah … because some of them are,

like, like a Baumgartner and Tony Keep.

237

Some of these people

that actually, like…

238

not Tony Keep, but some of these

people actually, like, that were, like,

239

conservatives, I didn't wanna punish

the people that, I didn't wanna reward

240

the people that refused to respond.

241

Yeah.

242

And so I tried to seek

out other writing by them.

243

Isaiah Payne, I looked, okay,

Isaiah Payne, and then I looked

244

for his something he wrote.

245

And I saw, oh, this, he wrote

something for the Spokesman-Review-

246

… so I'm gonna test this.

247

And that turned out to be like

predominantly, m- full of AI- … writing.

248

And so then I just started going

with these different, going through

249

some of the different columns-

and I started just thinking okay,

250

what if I tested a whole year?

251

What if I tested- … two years?

252

What if I tested three years?

253

And I wanted to ramp up the scoreboard.

254

And then by doing- … we actually and I

haven't actually seen this I don't know

255

if this has been done to a full degree

we actually get a sense for, like- Like

256

an actual curve- for what's happening.

257

And so we could see two years ago

258

I only found basically two or three

stories that had- … significant

259

amounts of AI at all.

260

And then 2025, quite a bit more.

261

It was, like, in the mid-30s.

262

And there was, like, 58 this year,

and we're only three-quarters

263

of the way through the year.

264

Yeah.

265

So it had just…

266

And it actually had come to a point where

the last two months, as of September

267

20th, the last two months- … previous

two months had been about half of the

268

Spokes Review's op-eds had significant

amounts of AI writing in them.

269

And so that was, like, a, a real s-sense

of okay, we're actually like getting

270

to the point where the inhuman writing

is maybe eclipsing the human writing.

271

And yeah.

272

Yeah, and then that…

273

It's so funny 'cause as a human editor,

Yeah … like it's pretty easy for me to

274

spot AI writing in a lot of cases, or if

it's spec- mostly if it's a longer story.

275

Like yesterday I was looking up I was

trying to find out some information

276

about Facebook is gonna be charging

business pages to post links.

277

Obviously, we post links all the time.

278

And so I was trying to find out more

information about that, and then I clicked

279

on a story th- through Google, and I'm,

like, reading it, and one hallmark that

280

I always see pick up on in AI-written

stories is they repeat information-

281

… throughout the story a bazillion times.

282

And I'm like, "Yes, I got that.

283

I understand."

284

Yeah.

285

And it's like and it'll just, And-

… there's also things with the structure,

286

like, when they're like, "Okay, this

is what it means, blah blah blah."

287

And it kinda has these same subheads.

288

That's pretty easy, but detecting the,

like, small amounts that or to the

289

level that this tool is detecting-

is definitely unmatched from a human.

290

But let's talk about

what people were saying.

291

So you, you're- … you put all these

things through Pangram- … finding

292

out, it's a certain percentage of AI.

293

Some of them are 100%, some of

them are- … 75 or whatever.

294

You call these people.

295

So you called- … or emailed, reached out

to the the candidates for office- … on

296

both sides that were- … using AI.

297

What did they say?

298

First of all, no one I talked to said that

they- … had just typed in the prompt,

299

"Write me something on this-" … "chat."

300

They're all like, "These are my ideas."

301

And just it's just the, But then

also they would say "I'd also

302

use this for brainstorming, using

this for see the pros and cons."

303

So I'm like, "Wait a minute.

304

Is this actually…

305

To what degree are these your ideas?"

306

But it, for most of them were saying

that, that- Yeah, I used AI, but only

307

for minor things- minor grammatical

edits trying to double-check maybe s-

308

checking some sentences- … trying to

make sure that, it's I got it under word

309

count- … or trying to- And for the

record- … I don't think any of the

310

candidates actually got it under our

word count for our questionnaires- the-

311

which bothers me … the fourth strict,

the, the ones that I saw that actually

312

were all the ones that were all AI'd

actually did hit, they're actually- Oh,

313

that's good … all under word, yeah.

314

For the most part, most of the

ones- Okay … that were fully

315

AI were, at least it can count.

316

Yeah, no one, which is, So was

it the ones that were, like, part

317

human, part AI that were- Yeah, some

of the ones that were part human,

318

part AI were a little bit longer.

319

Like Michael Cathcart- That's funny

… who he has, some of it was completely

320

human, some of it was a little AI,

but it was like- Interesting … so

321

I would say that basically no one

suggests they didn't use- … it at all.

322

Some of them said they just use Grammarly,

which is a tool that had been- a

323

spell-checker, grammar checker that has

since become a much more active AI tool

324

that does- … a lot more suggesting

rewriting sentences and things like that.

325

But no one was, but mo- people

were downplaying what that meant.

326

And I'm skeptical- … just because

we had tested, With our own writing

327

we had tested with our own writing-

… and and maybe it's different when this

328

is such a s- short- … they're short

sections, but we just weren't getting- And

329

I even- Yeah … tested it with ours where

I- … I asked Claude to write something

330

and I inputted it into- a document that

was all human written, and it still came

331

up as hun- 100% AI, or sorry, 100% human.

332

And so I was like, okay it wasn't a big…

333

It was, like, one sentence that I added.

334

And so it's interesting to see, like, how

what, how much was, AI changing their work

335

if it still showed up so significantly

in our, when we looked at it.

336

Yeah, it was, that's what I'm saying.

337

It was like, it was

actually very difficult.

338

I don't think I was able to, by

taking like some, taking human

339

writing and trying- … to asking

AI to just alter the fully human-

340

… writing, I wasn't getting to 100%.

341

It was…

342

And maybe it would've been different

if we charged with a very short amount.

343

But or if, but I just

wasn't getting there.

344

Sometimes it would say it was AI assisted,

which is a little bit different- … where

345

it says that but in part because

the AI is also looking at things

346

like structure, formulaic structure.

347

The pangram is.

348

Looking at things like formulaic

structure- … looking at where

349

your line breaks are, all these kind

of things that are- … more, so

350

it's, I think it's a lot, it's…

351

If you start with a completely AI

thing, it's hard to make it, it's

352

hard to tell pan- it's hard to

fool pangram into thinking it's

353

human by- kind of modifying it.

354

But I think it's also

the other way around.

355

If you start with something fully human,

I think it's actually pretty hard to,

356

in most cases, to turn it into 100% AI.

357

Yeah.

358

It's not like it, it can't ever happen,

but- … i'm just sort of- and none

359

of the, we, I didn't ask everyone,

but the people we did ask, no one

360

was willing to show their Google Docs

history- or their writing draft history

361

or kind of the, the one person who

did was, who was Paul Dillon showed

362

his- … Mic- his Microsoft Word document.

363

So he was this council member who's

very anti-AI- … philosophically,

364

and it's like this is destroying

what it means to be human.

365

He was a col- used to

be columnist himself.

366

He is really sad about what AI is doing-

… to our sense of human connection.

367

So he's philosophically morally

opposed to using AI to write, and his

368

thing comes back as AI from Pangram.

369

He's like, "What is this?"

370

So he gi- shows me the history, and he

is making some, some small modifications,

371

but he's making small modifications to

what his co-writer had submitted him.

372

And I tell Paul, "Hey, Paul, why don't

you go into Microsoft Word, look at

373

the version history of what this…

374

Not the version history.

375

Look at the the document

properties for the thing- … he

376

sent you for go down to info."

377

And I'm like, "Tell me when it

was, when was the document created

378

and when was it last edited?"

379

And he's like, "Oh, there

was a six-minute timeline."

380

So I suggested either he'd made a new

doc, he save it as a new document- … or

381

he'd only spent six minutes writing

this entire thing, which- … I can

382

tell you that writing- … something

500 words does not take six minutes.

383

No.

384

I wish it did, but it does not.

385

Yeah.

386

And sometimes I'll draft something

in my, my Notes app- … or whatever,

387

and then I'll paste it over to a doc.

388

Yeah.

389

But even then I'm futzing-

… around with it for a while.

390

And Paul said he reached out to

his co-author, Brian Henning, and

391

Henning- … had admitted what he'd done.

392

He'd basically tried to summarize

and condense some of his previous

393

work, combine it with basically tried

to pull in stuff he'd previously

394

written with the new context and

like- … he had that, done that.

395

And he apologized to Paul.

396

The frustrating thing is

he never called me back.

397

I tried to call him multiple times.

398

Oh.

399

And so it's I think it's yeah,

you're you're betraying your trust

400

with Paul, but you're also somewhat

betraying the trust with the public-

401

Yeah … where you're acting like

this is something you're writing.

402

It's your name is on it.

403

And you may have, it may be,

like, drawn from some of your

404

writing- … but the actual process

of putting things together and putting

405

the arguments together isn't there.

406

Yeah.

407

And for some people in some

professions, that's like who cares?

408

Who cares- Yeah … what the actual the

style is, how the sentence is constructed.

409

I think as writers- … journalists,

we really do care about that because

410

that's how we build our reputations.

411

That's how we make meaning.

412

And we really- Like struggle

with- … with that.

413

And like it's become like- Yeah … for

me it's become harder and harder- … to

414

write in a timely fashion, and it's

because my, throughout my career

415

it's become harder and harder to

like actually get anything done.

416

And some of that's because technology

has been, like social media has

417

really- … eaten away at my brain and my

attention span- … and just the ability.

418

And I also just this really exacerbated

my existing ADHD- … until I like see

419

all these possible ways that a story

could be, and I'm torn between them.

420

There's a paradox of choices.

421

There's a whole long kind of weeping

and gnashing of teeth- … dark

422

night of the soul, trying to write

something relatively simple sometimes.

423

And there's a little bit of and so in

the midst of that, I feel like a lot of

424

people are struggling with that right now,

including- really experienced writers.

425

And then here comes AI- … in, in the

dark alley, opening the trench coat

426

and showing you like this easy way out.

427

This is like this quick and easy path

that will forever dominate your destiny.

428

Kind of here's this way, all

you have to do is like ask, just

429

tell me to write something, yeah.

430

And so I'll tell you like I, I was,

this was Range, it was a previous

431

place I was working, and I had an

editor that submitted like suggested,

432

some suggested paragraphs to my story

for kind of the- … the nut graph,

433

which is the central kind of summary

piece that comes at the beginning.

434

And it was really long, it was

not like his writing at all.

435

It read like significantly like AI.

436

Yeah.

437

And so like I don't wanna call him out

on that, that's not something, if we were

438

in a situation where I could do that.

439

And I'm like, "I'm going to like just

rewrite this entirely and try to change

440

it to get with what he's trying to make me

do, but try to do it all in my own words."

441

And when did that- … it took me a f- a

fair amount of time to try to do that kind

442

of and then I ran that through Pangram and

it turned out it was all AI except for his

443

little introduction, which is like- Wow

… feel free to edit this or whatever.

444

And so that's like I'll never

work , with that person- Yeah

445

again because it's like those are

the kind of things where like even

446

if you're trying to not- … take

shortcuts, there are other people, your

447

collaborators, your other people who will.

448

I have a, a, Luke, our publisher-

… quote tweeted you earlier- … about

449

Paul Dylan, accidentally or,

having a piece of writing attached

450

to his name with AI in it.

451

And then Luke said or tweeted

"Even if you aren't using AI,

452

your writing buddy might be.

453

Loving partners disclose when

they've gotten in bed with Claude."

454

Yeah.

455

It's so get tested.

456

Go to the- Yeah … go to the health

get, go get Pangram and get tested

457

is what I recommend before you enter

into a relationship with a co-author.

458

And and I think right now we're

at the point where we haven't

459

really figured out the- … ethics

around this, the universal ethics.

460

And I think it's like it's begun like

ethics are- Ethics are a system of norms-

461

… that people the people agree on.

462

And we haven't figured out is this a

betrayal if something- … is AI written?

463

What if it's mostly AI written, and

what if you wrote a little bit of it,

464

but it's like how many people, like

how many people if they knew that

465

it was like mostly AI written or AI

assisted would they read that column?

466

How many people would be upset at their

partner if they wrote you a love note and

467

it was- Oh my gosh, I just- … written

in, with AI, like AI thing wrote it.

468

I have a tangent about this.

469

There was- Do it … like a Reddit Am I

the A-Hole- … post where or I- it was

470

one of those types of like Reddit threads.

471

But this woman or this guy wrote his

wedding vows- … with AI, and during the

472

ceremony, allegedly during the ceremony

according to this post his fiancee

473

left him at the altar and walked away.

474

And she was like, "Are you kidding me?

475

Did you use AI to write

your wedding vows?"

476

And then she just walked off.

477

That's what the post said.

478

Is this an urban legend?

479

This feels like that.

480

No.

481

It's one of those Am I the- And then

she's eaten by a sewer crocodile.

482

Yeah.

483

Yeah.

484

Okay.

485

It was one of those posts.

486

Okay.

487

But I do like it, it is crazy.

488

It is weird to think like of all of our

different like philosophies and- and

489

ethics or scruples- … with regard to AI

because we were, through- throughout doing

490

this story we were like, "Oh, we need to

have a newsroom AI policy like yesterday."

491

Yeah.

492

Like we don't really have, we've

just been going on vibes basically.

493

And so part of this series was

I had to create our AI policy.

494

And I was working off of other

newsrooms have done AI policies.

495

And I had, lists of examples from

other newsrooms with varying levels

496

of comfortability with AI, … and how

they were using it in their workflow.

497

But I couldn't really find a

good example of one that was

498

like- Comprehensive and hum- like

human, for lack of a better word.

499

Yeah.

500

But 'cause like our policy was cr- we're

worker-owned, and so- … everything

501

we do is made by, i- is, agreed upon

or we try to come to a consensus,

502

and so with our AI policy, we have

different people on staff with different

503

levels of comfortability with AI.

504

Some Aaron Sellars is like, "I'd rather

chop off my left hand than even have

505

AI write, write an email for me."

506

You're like, "I don't ever want it

to touch anything that I write."

507

But I'll use, you'll use AI for

spellcheck and- And I will use it

508

also- … to full disclosure, I'll use

it to make - pictures of myself and

509

my wife as Hollow Knight characters.

510

So like cartoon characters.

511

Yeah.

512

So I will do that, like-

Yeah … for Valentine's and such.

513

So that's an important use case as

well- … to be honest about that.

514

We need that for the-

Yeah … for the world.

515

So there's, that's actually a good

example though because when I'm doing

516

that, it's like it would mean, yeah,

it would mean a lot more if I had-

517

hand drawn it if I was an artist- Yeah

518

who was able to hand draw it.

519

And so it means something 'cause

I was coming, come up with the

520

idea and I'd try to like- … put

it all together into the work.

521

But just like buying a- Buying a holiday

card … a card that like matches.

522

It's kinda like that.

523

Yeah.

524

And so like there is, but I also think

that yeah, and so I think we're just

525

beginning to, we're all beginning to-

… try to figure out what does this mean?

526

What does this- Yeah … and and I

talked to a lot of the people that were,

527

a lot of people, some of the people I

talked to were basically like, "Should

528

I discl- " They were kinda like the

George Costanzas "Was that wrong?"

529

Yeah.

530

"If anybody had told me

that this was an issue…"

531

They were like, "Should I have disclosed

this to- … the spokes review?"

532

There wasn't like, like they

kinda wished wished there had

533

been like a checkbox or something.

534

Is it new to them?

535

Not really.

536

It was like- Yeah … and I think for

a lot of these people, it just becomes

537

something that's a part of the workflow.

538

A, a lot of the people we're seeing

weren't, they weren't like the

539

people, these weren't like just like

the students or the kinda the blue

540

collar workers or like a lot of these

people- … that like, they were

541

the people that were like using AI.

542

They were, a lot of times,

were the most powerful people.

543

These were the CEOs, these

were the college presidents.

544

The educated and the, yeah.

545

These were the, yeah, the PhDs

and the doctors and the, and many

546

people who had like- … excuse

me, comms teams around them.

547

One of the people that was using,

like putting out AI stuff, was

548

like a politician who was like

a champion college debate- who'd

549

been a champion college debater.

550

So no one is good at, like including

like in the impromptu- … the impromptu.

551

So she was like great at rhetoric, right?

552

But she was relying on her staff and

giving her notes to the staff, and

553

her staff- … was using to turn those

notes into AI, and she was looking

554

them over, so like- Yeah … most

of these people don't need to.

555

But in some cases it's because

like they're corporate and

556

they wanna be cautious.

557

They don't, they actually wanna

remove some of the humanity from it.

558

They want it to be like safe

and match their policies.

559

They don't wanna be, go beyond,

and like there's, I use AI to and

560

I wrote this in our AI policy.

561

It's like we're not gonna use,

like I, when writing our AI policy,

562

I kinda had to like separate.

563

It was a really big like kind of thought

exercise too because I'm- So much of our

564

internet is based off of AI, no matter

565

Like, things that you don't think.

566

Spell check- … is technically AI.

567

Grammarly is AI.

568

Even if you don't use the

generative AI features of it.

569

Different people, there, there's

not really a lot of agreed upon

570

definitions of what each thing

is- … and how they'll use it.

571

So I think that's the first step is before

we figure out our, widely used ethics

572

is we kinda have to define these things.

573

But I did separate our editorial and

content creation from operations- … in

574

our policy because as sometimes we

need … me and Luke use Claude to help

575

us with planning and, like- budgeting

and things like that where- … or if

576

I'm trying to draft an internal policy

or an internal training thing and I just

577

have a bunch of notes and I just don't

have the mental energy- … to write

578

it all out, and I'm like, "Hey, Claude.

579

Take my- … crazy notes and

draft it into a co- somewhat-

580

cohe- like, cohesive, document."

581

And it'll do it, and then I'll take

that and then work off of that.

582

And that, but that is for- … internal

things that are not going to- … be

583

under my byline or under the

Range- … masthead or whatever.

584

So it's like we have- varying views on

AI and safety Yeah that's the guarantee.

585

If we're putting out something on

our website- … if we're putting

586

out something published into the

world- … that, it's going to be,

587

it's going to be human written.

588

Yeah.

589

And I will say that one of the, the

dangers sometimes of farming out some

590

of the internal emails- … and things

like that to AI is that sometimes

591

you don't really know, if you're

not careful, you don't know what you

592

said, and you don't know- Oh, yeah

593

what you communicated.

594

You don't remember what- … you were

creating or the policies- Yeah … you

595

were creating or the structure

you were creating because the act

596

of- … the grind, the donkey work as

one of my sources- Yeah … had said.

597

The donkey work of putting together a

story, that's how you really remember it.

598

Yeah.

599

That's how you really, and I, when I'm in

the, and for the record, even the internal

600

things are, like, heavily edited by me.

601

Like they are.

602

Yeah.

603

Every line I go through- … because

I'm like, I don't want, I'm not just…

604

we're not some corporation where I have

to poop out a report and- give it to

605

the CEO, and the CEO- … may or may

not read it, like- Yeah … this is

606

my real coworkers that I care about.

607

I'm not just giving them

busy work to read this thing,

608

definitely it does not leave…

609

I'm c- I have some documents I wanna

put through Pangram now actually.

610

You should.

611

And this is…

612

one of the good things that, I don't know.

613

I'm hopeful that Pangram will

continue to be able to provide, 'cause

614

obviously there's new models that

come out- … and there's the risk

615

that people will begin to, start to

write more and more like AI because

616

that's what they think is good writing.

617

But I would hope it would continue

to provide a distinction because

618

I think in this world, like,

how do you distinguish yourself?

619

The challenge was that AI

allows people to just create-

620

… phenomenal amounts of content.

621

Yeah.

622

And most of the times,

like, fairly empty content.

623

Some of it will have information,

but it'll be a lot of sound and fury

624

signifying nothing or just very little.

625

And so it'll be all out there.

626

It'd be really hard to separate

the wheat from the chaff.

627

Yeah.

628

And but with Range our guarantee is

w- it's gonna be all wheat, baby.

629

We're just-

630

it's gonna be, like, it's gonna be our

mistakes are our mistakes- Yeah … and

631

our our writing is our writing, and I

think people genuinely care about that.

632

The challenge is are they going to be…

633

Do they have the energy to try

to suss out the difference, yeah.

634

And I think m- the, I think the advantage

of Pangram is that you can, if you're a

635

subscriber you can put it as a embed it

in your LinkedIn and your Twitter feed-

636

and it'll automatically, like- kind of

look through the post and say, "Hey, does

637

this look like AI or look like human?"

638

A lot of the posts are so short that you

like, you wanna take it with a grain of

639

salt- … because a, a pangram gets weird

if you're like under 100 words basically.

640

But yeah, I would say that like that's

a, that's an advantage of knowing

641

that, okay, I c- we can actually prove

that this is human and we are- Yeah

642

writing with this, and we're like

sometimes struggling like- Well-

643

Yeah … and like for us in particular,

like it's, We are member supported.

644

And so- Yeah … what are the

members supporting if not- … the

645

human writing we put out there?

646

So- … for us, I feel like it's obvious,

like the lines are pretty obvious at least

647

when it comes to like editorial content.

648

We all share- … like that same line.

649

I think with the wider society it's

gonna be interesting because a lot of

650

people don't give a crap about writing.

651

A lot of people, Yeah, and then that's

been something that's been striking

652

me about the, the reaction to this

series- is that it's a lot more tempered

653

than I thought maybe it would be.

654

Like I, I didn't think that

like- … like Spokane really cared

655

about what the Spokesman's Review

is putting in their op-ed pages.

656

But I, expected a little bit more like-

Outrage on one side or the other, I always

657

assume when I put an article out, it's

gonna basically, me putting an article

658

out to the world's gonna be like Prince

Ali entering the palace in Aladdin-

659

… with the big, like, elephants and all

the monkeys dancing and everything.

660

I feel like that's gonna be the

dramatic "It's out world finally,"

661

and everyone's gonna be clamoring and

everything, and it's sometimes it's not.

662

And I think it's like- … I think,

it's maybe it's we didn't sell

663

it, but also it's, it could be the

situation that either people, their

664

friends are called out in the story.

665

'Cause we really call- we were not as, we

were not afraid to call out Democrats in

666

this- this piece as well as Republicans.

667

We were like, we went after both

sides- … hashtag both sides,

668

pretty - … as I'm want to do.

669

And then we also, like I think a lot

of people use AI to some degree in

670

their everyday life, and so they're

like- … huh, but I don't know.

671

Sometimes as a journalist it's not…

672

It's my job to put out things that

are worthy of being cared about.

673

It's not my job to force people to

care or force- … the reaction.

674

I wanna put out something

that's worthy and and it's

675

there if people want it , yeah.

676

So I think it's, yeah,

it's something that I'm…

677

I don't know.

678

It took me a long time.

679

It's something I'm- Yeah … I'm proud of.

680

And I don't know.

681

I thought, I talked to, I was talking

with my colleague Erin Sellers

682

the other day, and she was, you

know- … just tr- struggling to

683

like these things like, "Daniel,"

'cause I'm a lot older, I'm like 40.

684

And she said, "Daniel, does

this ever get any easier?"

685

And so I told her the same thing

that, that Michael Boa and this

686

wizened, the wizened old Inlander

687

was like, "No, it gets harder."

688

And so it gets, the older you

get, the more difficult it gets.

689

But there is a beauty and

a meaning in the struggle.

690

And I wrote a piece about, a lot

of my last pieces I wrote for

691

the Inlander was a piece about

AI and whether AI can replace me.

692

And what I eventually came up to was

is that, Maybe eventually AI will be

693

able to do everything that humans,

simulate everything humans can do.

694

But the act of creation- will

always be meaningful for humans.

695

And the struggle itself is

often what gives the art and

696

the creation true meaning.

697

Damn.

698

And so that's the kind of, that's the

conclusion I came to- … with all

699

struggling, and it's like this, maybe

we're in the, maybe it's gonna be

700

like, you can now take a digital photo

of something, but there is meaning

701

still in, in painting, there is still

meaning in doing something yourself.

702

My brother made me a a chessboard,

and I c- I could've been like,

703

"Josh, I could get a much better

chessboard for $5 at Walmart."

704

"What are you doing?"

705

But, but I don't because

there's craft in there.

706

Yeah.

707

And I say I think you will have

somebody that says artisanal sentences.

708

And I didn't- Oh my gosh, that, I'm

gonna put that in the merch idea channel.

709

Yeah I use some AI when to assist

me with creating some of these b-

710

customized book covers for a wedding.

711

But when I proposed to my girlfriend at

the time- … I sought out a local artist.

712

I'm like, "Hey, I wanna take all this,

this writing that we've done together.

713

I want you to take it on, put it on yellow

paper and make it into daffodils," right?

714

And it's like those are something

that I couldn't, those, I

715

couldn't do the shortcut there.

716

I wanted to pay a local artisan there

because I thought this is more meaningful.

717

And and that wa- it was.

718

And and the, her sister helped

do the, the, the invitations

719

create the invitations for our

wedding, who's a graphic designer.

720

And that was a lot more

meaningful than- Yeah

721

just, the AI doing it.

722

So I think we'll still have that,

but- … but I think it's also important

723

to be able to actually have a mark of

authenticity because right now I think

724

there's a lot of people that pr- pretend.

725

They'll put their byline, they'll

say by, by, John Johnson, and John

726

Johnson did not do this alone.

727

Yeah.

728

And it's also I'm- … as an editor

this has also caused me to reflect

729

on my role in the writing process.

730

And I know it's been a somewhat subtle

debate in journalism of should editors

731

also have a byline on writers'- stories.

732

I don't think it needs

to go to that extent.

733

But thinking about my role in editing-

… your words or- … or adding a sentence

734

or a paragraph or things like that.

735

Butchering, I think it's called.

736

Butchering.

737

Yeah.

738

Yeah.

739

But, should- I don't know.

740

Do we need to disclose that too?

741

That's just a question that I'm-

Yeah … kind of thinking of.

742

And I always thought about, I

think sometimes it's and the

743

editors should their byline

is on they're on the masthead.

744

They're, like, they're there in the story.

745

You can look for someone.

746

Yeah.

747

And you know that there's this

is something that's supported

748

and like- … and they're being,

they're also being paid too.

749

And so it's like a, this is a part of

the- Yeah … this is a part of it, and

750

they're humans, and they're, like, helping

shape it, and like we are- And there's

751

a lot of human judgment calls that we

have to make in editing and, like- Yeah

752

the way we phrase things and- Yeah.

753

No Yeah … and it's important.

754

And I don't think that's, yeah, and

I think that the process of making

755

those judgment calls can be very

different than- … than with AI.

756

And I've seen editors start to,

some editors start to rely on AI

757

and become much, I think, much

weaker editors because of it.

758

Because they're like, they're shot,

they're farming out their expertise.

759

Yeah.

760

Yeah, and I don't know.

761

I think it is like a, nothing

is ever completely pure.

762

Your writing is not, there's a reason

why you're trying to … And I do

763

have ChatGPT check my stories for

typos, and it flags a bunch of things.

764

And it flags a bunch of things wrong-

… or the ways I don't wanna change it.

765

But it and it, but I just do it for

typos, and I change them individually.

766

And my writing never shows up on Pantagram

as AI because I'm- Yeah … I'm just using

767

it as a spell-check or grammar check,

and that's, I think that's okay to do.

768

It's okay to use it as an advanced,

I think it's okay to use it as an

769

advanced Google tool to, like- … as

long as you're actually seeing that

770

it matches what they're claiming,

'cause a lot of times it doesn't.

771

Yeah.

772

But yeah, no I, I, but I think we're

we're all still trying to figure that out.

773

And the challenge was, is that

journalism is in such a financial

774

crisis right now- Yeah … that we are

so vulnerable- … to companies like

775

McClatchy- … wanting to find the

quick and easy route to "Oh, maybe we

776

can-" Maybe we can replace journalists

with- … AI and churn out, you know- Yeah

777

10 times the amount of copy, and it's

interesting because before AI ki- and

778

data centers, 'cause they- … tho- those

two controversies go pretty hand-in-hand.

779

Data centers do, have

existed without AI before.

780

That's- … how we do cloud computing.

781

But before it, it got into the

m- mainstream- probably back in

782

2024- … we were part of Documenters.

783

The Documenters program, which is-

… based out of Chicago, where they train and

784

pay regular people to go to city council-

… meetings and government meetings and take

785

notes on them, and then an editor from

their- partner organization will edit

786

the notes, and then they get published.

787

And so the- Which is awesome.

788

Yeah … which is awesome, yeah.

789

And we were part of that

program for a while.

790

… We re- just did not have

the bandwidth to do it.

791

Yeah.

792

It was a lot.

793

And so we haven't been

able to get back to it.

794

But I remember when we went to Chicago

to go talk to Documenters for one of

795

their summits there were some newsrooms

that were already implementing AI

796

note-taking in- … city meetings.

797

And that kind of, was an uncomfy,

like- … feeling because it's on one

798

hand these meetings are generally not

getting covered by anyone- journalist

799

or regular person or whatever, or AI.

800

So having Otter taking notes or

whatever is not terrible, yeah.

801

It is is better than nothing, … and

on the other hand, it also had this

802

kind of direct result of even this

program that is meant to train and

803

pay regular people to do this was,

like, starting to get outsourced or

804

it had the threat of being outsourced.

805

And so there's that really- … very

real tension of- … and it's the

806

tension that, comes up, I think, in

our newsroom a lot when we talk about

807

how we how all of us have different

feelings about using it personally.

808

Because for some people it's like,

yeah, it makes doing this task that

809

is impossible otherwise for me to do

makes me able- to do it, and otherwise

810

I would not be able to do it, and

then the result would be X, Y, Z.

811

So it's interesting to think now,

where we're like, okay- would it

812

be good for us to stick an AI Otter

AI notetaker in public meetings?

813

Or does that just defeat the whole

purpose of being a watchdog if

814

nobody's actually even watching it?

815

There actually are sites that create-

Yeah … like I've come across, that

816

cre- are creating summarize, summaries-

… of public meetings using AI.

817

And a lot of the summaries are really bad.

818

Yeah.

819

It was like, "And then Zix Sapone,"

as opposed to- … Saxophone.

820

It was like, what?

821

And but the, but it can be useful for

a journalist who can find the- and go

822

to the right, the point in the meeting

and try to, like- … confirm it.

823

But that's the thing is that a lot

of stuff is being put out as, the

824

v- you know, first just this raw-

… messy, flawed, error-ridden stuff.

825

And it's it's part of our job to

like comb through it and figure

826

out what's real and what's not.

827

But I don't know.

828

Like I think our kind of radical

supposition at Range- … that we're

829

crossing our fingers and praying

that ends up working- … is like

830

that people pay for inefficiency.

831

They're like, we are not- … we

are not the most efficient newsroom.

832

We're not this, we are sometimes- Yes

833

a newsroom that's a little bit slower.

834

We're like, I was, my byline was not in

Range stories for a month because I was

835

working on this big- … Range AI stories,

and I was working, working like heck- Yeah

836

… 'cause this is the radio, we say heck.

837

To heck to try to get it done.

838

But I wasn't, I was like I just…

839

And I was just thinking the other day as

this was coming out, I was proud of it,

840

and we had all these fun- … graphics

that were made, and there was some- Yeah

841

… humor, and there was some interest.

842

I could not do- … this

project, I could not have done

843

at any other publication- Huh

844

in the Inland Northwest.

845

I couldn't have done it at the Inlander.

846

I couldn't have done it at the Spokesman.

847

I couldn't have done InvestigateWest or

Seattle Times or like- … or the, or,

848

or the Coeur d'Alene Press, wherever.

849

Anything in the- … like

I couldn't have done that.

850

And it was like- … it was just there

was enough time and enough everyone coming

851

together, and the team that was coming.

852

This was something that was unique

to Range- … and the big leash the

853

long slack in the leash they gave, and

the, … the idea to the optimism that

854

we have here- to try to create something.

855

Let's pull it all together.

856

Let's actually- … make this happen.

857

And like at that, at a time when so

many other publications are trying

858

to make ends meet by just cutting to

the bone or forcing people to just

859

churn out incredible amounts of po-

copy or, withering, like whittling

860

down page counts to- paper thin.

861

We coulda, I coulda done a, like a

version of this, but I couldn't have

862

done something as fulsome or as-

… dramatic at any other, any place else.

863

Yeah.

864

And that's one of the reasons I

was really glad to be at Range.

865

Aw.

866

Is that this is I think this is a

beautiful thing we're doing, and we're

867

hoping- that people end up supporting us.

868

'Cause that's the only way that I think

this kind of media will thrive- … is

869

by, through the kindness of strangers.

870

Yeah.

871

Yeah, it's that makes me really happy to,

to- not to hear that nobody else would

872

have done this, but like that this is a,

a special place that can uniquely do this.

873

I just put in a document that is a fully

internal document for actually our co-op.

874

And we got some new board

members joining- … our co-op.

875

They're from the bakery,

brand new to boards.

876

And we did not have an

orientation guide at all for them.

877

And I was like, "Okay, I have three

years or four years of meeting minutes

878

that I wrote myself as a human- 'cause

I'm the, the secretary of the board.

879

And then I was like, "They need

to know all of this history."

880

And so I put those notes in Claude,

and I was like, "Can you come up

881

with a brief history to orient

like- … new board members?"

882

And then I also had I was like, "Okay,

what else should I put in this g- guide?"

883

Because I never got an orientation.

884

I don't know what I'm supposed

to be orienting these people to.

885

And so I came up with this, this

pretty good comprehensive guide

886

for these board members- … and

I edited every single word of it.

887

And so I just put it through Pangram,

and it came up as 47% of this text is AI.

888

And I would say- That sounds

about right, doesn't it?

889

That sounds pretty accurate.

890

I'm a little- Yeah … like some

of the sections that they're

891

highlighting, I'm "No way, I, that

is actually a section I did write."

892

Or, "Oh no, that's actually

a section Claude wrote."

893

But like the percentage- is completely

correct 'cause it is about like-

894

Yeah … and then again, that is for

an internal thing that couldn't really

895

have been created easily otherwise.

896

But it's also amazing

that you've done that.

897

And partly 'cause I think

it's- … a long enough document.

898

Yeah.

899

But like that it's that…

900

'cause like it's not 100%.

901

Yeah.

902

'Cause a lot of the other people

that were, that I was talking to

903

were saying, "Oh, I just had it

like tweak a couple things- … and

904

it's turning out to be, 100%."

905

And so it's yeah, I think, 'cause

you're starting with something

906

that's human written and then

trying to go from that too.

907

And I, It's funny because…

908

And I actually used two different models.

909

So I used Claude for, to summarize

the like the, the meeting minutes into

910

a kind of like a timeline of these

important decisions that have been

911

made- … and the context behind them.

912

And then I, And for the more like how

boards work, like this is how you vote and

913

you motion to vote and blah, blah, blah.

914

I used actually NotebookLM, which is,

I think it's called Gemini Notebook

915

now or something, but it's Google.

916

And that is more of a closed loop system

where- … you can put in like documents.

917

So I had like our bylaws in it.

918

And like our, like all of our

like legal stuff and like rules

919

and things like that in it.

920

And then I was like, "Okay, pull…

921

explain how to be on a board

based off of these things."

922

And so it's really interesting to see

what Pangram highlighted 'cause it, it's

923

a pretty big mix of like which ones, but

I would say the percentage is about right.

924

Yeah.

925

No.

926

It's just- And I think it's…

927

Yeah.

928

It, I think that this

does a pretty good job.

929

I think the percentages are

not always totally right.

930

Like sometimes I think

it might, it just…

931

Yeah.

932

And so sometimes it's a little bit, but

I think it's like basically what I could

933

feel confident is that if it says there's

a significant amount of AI- and it's a

934

lo- it's a fairly long document, there,

there has been AI- … used in writing it.

935

And no- Yeah.

936

And this was a 6,000-word document, yeah.

937

That's a long document.

938

Yeah.

939

You just wasted all those tokens.

940

Sorry.

941

Yeah.

942

I had to see.

943

I had to see what it would turn out.

944

… But yeah, it's, that's

pretty accurate, I would say.

945

But I don't know.

946

I think it's a good…

947

I don't know.

948

I think there's different tools

where people are gonna use it for.

949

Sometimes it's useful for finding

things or for, I've used it before

950

for medical documents that I'm trying

to understand- … what they meant

951

or just trying to double check.

952

I have too.

953

Like I've, I did some story where

I was talking about Yeah "Hey,

954

does this make sense to you?

955

What is this person saying?"

956

Trying to understand that.

957

I think it can be good for that.

958

Yeah.

959

Trying to understand that stuff.

960

And actually makes me…

961

I am positive about people going to AI

for medical information- … because the

962

ecosystem for Google was so bad before.

963

Yeah.

964

You would Google something- … and

then it would come up there'd be, like,

965

13 there'd be, like, maybe the Mayo

Clinic, but then 13 New Age blogs.

966

And there would be, like, alt medicine

and all these other things that are-

967

Yeah … not very scientifically based.

968

And then, but then the the auto- now

the Gemini, the Google AI, would come up

969

and say, "Hey, this is what is happening

here, and this is what's happening here.

970

It's probably like this,

and you should this.

971

Here's this study."

972

And if he's So it's, that's

something- It's- … that's

973

probably a little bit better.

974

Yeah.

975

And I have a lot of health problems

myself, and so I've- … I've been using

976

Claude to help sort through that and- Yeah

977

And help me advocate for myself.

978

I am like- … "Okay, how do I get my

doctor that is in this specific insurance

979

system to order this test for me?"

980

Yeah.

981

And have insurance cover it.

982

What, how should I word this?

983

And it'll tell me how I should word

something, and then I can- … and I

984

can do that, and it takes a fraction

of the energy- I would have to, to

985

r- 'cause otherwise I'm scrolling

on my phone until 2:00 AM on Reddit

986

threads being like, "Do you have this?

987

What do you have?"

988

And it's- … it's crazy.

989

So there are uses for

it, and I don't know.

990

I think the, I think- A big

thing right now to get to is

991

just disclosure - probably.

992

That's a good thing, too, is like-

… if we were saying AI, we used…

993

if these journalists or these op-ed

writers- … or they're, like,

994

saying, "We used op-," "We used…

995

This was written with conjunction of

ChatGPT or conjunction of Claude,"

996

would people- … want to read it?

997

Yeah.

998

And I think that's a good question.

999

If they don't- … maybe they're

wrong to not wanna do that.

00:46:23,358 --> 00:46:24,683

But I think that also

says that maybe don't.

00:46:24,690 --> 00:46:26,447

I think people wanna hear from humans.

00:46:26,457 --> 00:46:28,445

They don't wanna, for the most part.

00:46:28,447 --> 00:46:31,599

There's a lot of AI, AI art or AI things

coming out, and sometimes it was, like,

00:46:31,602 --> 00:46:33,024

popular and people don't know it's AI.

00:46:33,040 --> 00:46:38,056

But part of the whole point of art

is engaging with, communicating

00:46:38,056 --> 00:46:39,974

with other, with fellow humans.

00:46:40,024 --> 00:46:41,411

And that's what's powerful about art.

00:46:41,446 --> 00:46:44,904

And that's, I went to, I was in

Spain and I went to, one of these

00:46:44,938 --> 00:46:49,396

the museums where they have paintings

from Bosch and from these other these

00:46:49,415 --> 00:46:50,511

artists that are, like, these…

00:46:50,917 --> 00:46:53,429

Some of these artists that are like,

here's these, these triptychs of

00:46:53,429 --> 00:46:54,882

heaven and hell- … from the…

00:46:54,896 --> 00:46:56,333

W- I'm trying to remember what the name.

00:46:56,361 --> 00:46:57,579

Oh, yeah, Hieronymus Bosch.

00:46:57,579 --> 00:47:01,859

And, like, all of these is r- oh,

this is coming across, this is time

00:47:01,859 --> 00:47:09,252

traveling across, like- … 1,000 years

of human history to get to me, and

00:47:10,539 --> 00:47:14,034

all their assumptions and their quirks

and their- fears and their personality

00:47:14,039 --> 00:47:18,676

communicating with me, and it's powerful

because it's, there's a human behind it.

00:47:18,755 --> 00:47:20,111

And I think that's gonna become…

00:47:21,265 --> 00:47:23,372

I think there's a lot of that's

gonna become something sacred.

00:47:23,426 --> 00:47:25,830

And, I've had people, so many

people responding to this story

00:47:25,830 --> 00:47:31,199

like- This is gonna feel so

quaint in two or three years.

00:47:31,217 --> 00:47:32,901

Everyone's gonna use AI for everything.

00:47:32,901 --> 00:47:36,487

It's gonna be like- Yeah … you're

a person who's just refusing to use

00:47:36,488 --> 00:47:38,366

a calculator when trying to do math.

00:47:38,382 --> 00:47:40,094

And I'm like, yeah, maybe.

00:47:40,607 --> 00:47:43,791

But there's a beauty in, I

don't know, maybe this is my

00:47:43,826 --> 00:47:45,913

latent conservatism coming out.

00:47:45,920 --> 00:47:47,871

My Tolkien style conservatism, right?

00:47:47,921 --> 00:47:49,287

There is a beauty in the old things.

00:47:49,287 --> 00:47:52,698

There's a beauty in the old ways,

and there's a beauty in inefficiency

00:47:52,698 --> 00:47:53,671

and grappling with things.

00:47:53,673 --> 00:47:56,349

And there's a beauty in just

trying to understand what other

00:47:56,349 --> 00:47:58,653

people are thinking and saying.

00:47:58,653 --> 00:48:02,500

And that, those slow moments, those,

those dark nights of trying to figure out,

00:48:02,502 --> 00:48:04,559

hey, am I being fair to Jonathan Bingle?

00:48:04,915 --> 00:48:07,875

Am I not being, like, tough

enough on Jonathan Bingle?

00:48:07,925 --> 00:48:09,562

Am I, is this encompassing it?

00:48:09,562 --> 00:48:10,015

Is it accurate?

00:48:10,015 --> 00:48:10,534

Is it readable?

00:48:10,548 --> 00:48:12,960

All these- … like these struggles

that I have, like you could shop

00:48:12,960 --> 00:48:16,119

that out to AI, but there is

something- … deeply lost there.

00:48:16,227 --> 00:48:19,396

And and so maybe it's the

thing, like maybe we are the

00:48:19,396 --> 00:48:20,962

last of the Mohicans here.

00:48:21,200 --> 00:48:23,172

And the, and trying to be like this is…

00:48:23,202 --> 00:48:24,745

But I don't know.

00:48:24,777 --> 00:48:27,470

Maybe this is, if this is, if

this is- … if this is it,

00:48:27,471 --> 00:48:29,105

we'll make such an end, right?

00:48:29,314 --> 00:48:29,669

Yeah.

00:48:30,212 --> 00:48:34,794

Until then you'll get your artisanal

words- Yes … at rangemedia.co.

00:48:35,066 --> 00:48:38,853

And a time may come when the courage of

men fails- … but it's not this day.

00:48:38,999 --> 00:48:39,456

Thank you.

00:48:39,631 --> 00:48:40,917

So that's our time this week.

00:48:40,999 --> 00:48:44,683

Free Range is a weekly news and

public affairs program presented

00:48:44,683 --> 00:48:48,516

by Range Media and produced by

Range Media and KYRS Community

00:48:48,516 --> 00:48:51,250

Radio, KYRS Medical Lake Spokane.

00:48:51,616 --> 00:48:53,382

Thanks for listening, y'all.