Catherine Gregory | Ada Lovelace Institute

Speaker: 00:00:01

In a complex and rapidly moving

2

: 00:00:04

area like AI and data,

you have to reevaluate

3

: 00:00:07

your comms and messaging and definitions

periodically.

4

: 00:00:10

It's not going to be something

that's evergreen.

5

: 00:00:12

It's not going to be something

that's static.

6

: 00:00:16

Welcome to Research

7

: 00:00:18

Unraveled, the podcast, where we take

a deep dive into the somewhat niche

8

: 00:00:22

but hugely impactful world of research,

communications.

9

: 00:00:26

In particular,

we're exploring the idea of complexity,

10

: 00:00:29

which lies at the heart of this field,

and hearing from communications experts

11

: 00:00:33

about how they navigate

and overcome or unravel that complexity.

12

: 00:00:38

Research unraveled is brought to you

by me, Peter Barker,

13

: 00:00:41

owner of Orinoco Communications,

14

: 00:00:43

an agency where we specialize in working

with research based organizations.

15

: 00:00:47

This month, we're talking about a topic

that has had a fair bit of coverage

16

: 00:00:51

over the past couple of years.

17

: 00:00:53

Artificial intelligence.

18

: 00:00:55

I don't know about you, but

19

: 00:00:57

even though this emerging technology

has dominated the headlines

20

: 00:01:01

and social media feeds during that period,

I still feel like there are huge gaps

21

: 00:01:06

in my understanding

of how AI is being used

22

: 00:01:09

in ways that affect me and other members

of society on a daily basis.

23

: 00:01:13

We hear about the apocalyptic visions

of what

24

: 00:01:16

might happen if we lose control of AI,

and about the potential

25

: 00:01:19

for whole professions

to be replaced by artificial intelligence.

26

: 00:01:23

And personally, I'm becoming

quite comfortable with using platforms

27

: 00:01:27

like ChatGPT on a daily basis

28

: 00:01:30

to help me with tasks both at home

and in my professional life.

29

: 00:01:34

But both those dramatic visions

of the future and ChatGPT,

30

: 00:01:38

which sucks up

quite a lot of the conversation around AI.

31

: 00:01:42

I really just the tip of the iceberg

when it comes to how

32

: 00:01:45

artificial intelligence can

and is impacting our lives.

33

: 00:01:48

Thankfully,

there's a research organization

34

: 00:01:51

that's dedicated to investigating

that very question.

35

: 00:01:54

The Ada Lovelace

Institute is an independent, UK based

36

: 00:01:57

research institute

that was set up in: 2018

37

: 00:02:00

In my conversation today,

we will hear about its mission to ensure

38

: 00:02:04

that the opportunities

and benefits of data and AI are shared

39

: 00:02:07

by all, and the ways in which

the Institute communicates that mission.

40

: 00:02:11

From this month's guest,

Ada is Head of Communications and Content.

41

: 00:02:16

Katherine Gregory,

in what I hope you'll agree is a lively,

42

: 00:02:20

fun and most, well,

fascinating conversation with Katherine.

43

: 00:02:23

We cover how to communicate

with policymakers,

44

: 00:02:27

to reframe narratives and some techniques

to increase the likelihood

45

: 00:02:31

that your message will stick.

46

: 00:02:33

We talk about, in particular,

how the Ada Lovelace Institute

47

: 00:02:37

attempted to influence

the conversations being had about

48

: 00:02:41

AI at the AI Safety

mit held at Bletchley Park in: 2023

49

: 00:02:47

We also talk about the thorny issue of how

50

: 00:02:49

to pin down a definition of a constantly

evolving technology like AI,

51

: 00:02:54

and about the benefits of bringing

researchers and communicators together

52

: 00:02:57

early on in a project to generate the

highest quality communications content.

53

: 00:03:02

Of course, we'll finish by hearing

Katherine's favorite bit of comms advice

54

: 00:03:05

that she's ever received

and her book recommendation,

55

: 00:03:09

which actually this time

is a documentary recommendation.

56

: 00:03:13

Here it is.

57

: 00:03:14

Katherine,

thank you so much for being here today.

58

: 00:03:16

I'm really looking forward

to our conversation.

59

: 00:03:19

And I just wanted to start

by hearing a bit from you about the Ada

60

: 00:03:24

Lovelace Institute and when it was set

up, why, what its mission is.

61

: 00:03:29

Brilliant.

62

: 00:03:30

Well, first of all,

thank you so much for having me, Peter.

63

: 00:03:32

So Ada is an independent

research institute with the mission

64

: 00:03:36

to ensure that data

and AI work for people and society.

65

: 00:03:40

So we occupy a really interesting place

in the AI and data ecosystem.

66

: 00:03:46

We work in this space,

but we don't actually focus on the types

67

: 00:03:50

of technologies we want to build,

but we focus on the types of societies

68

: 00:03:54

we want to build

and how AI and data fit into that.

69

: 00:03:58

We are part of the Nuffield Foundation and

we are a relatively young organization.

70

: 00:04:04

We are set up about six years ago.

71

: 00:04:06

Wonderful, fantastic.

72

: 00:04:08

And so how did you what's your sort

of connection, your story with Ada?

73

: 00:04:12

How did you get involved

in the organization?

74

: 00:04:16

So first of all, I definitely don't have

75

: 00:04:19

a technical background, and I hope

that's helpful to some of your listeners

76

: 00:04:23

who may want to get into this space, but

might not be techie people, so to speak.

77

: 00:04:28

But one of my passions,

and one of the things

78

: 00:04:31

that really attracted me to this job,

is that I love to translate complexity

79

: 00:04:37

into easy to digest, compelling messages

that resonate with audiences.

80

: 00:04:42

And my my first comms job ever was working

in the Obama administration.

81

: 00:04:47

I mean,

if you want a place for storytelling,

82

: 00:04:50

that was definitely the place to be.

83

: 00:04:52

I was a speechwriter

at the Treasury Department over the summer

84

: 00:04:55

while I was getting my master's degree.

85

: 00:04:57

And I would say that

learning the discipline of speech writing

86

: 00:05:00

is one of the things

that's really impacted my career the most.

87

: 00:05:04

So I learned how to translate topics

like the debt ceiling,

88

: 00:05:08

which is something you hear,

but no one really gets what it is

89

: 00:05:11

into punchy messages

that would really resonate and land

90

: 00:05:15

with the different audiences

that we were speaking to.

91

: 00:05:18

And I think so much of those skills

92

: 00:05:19

are transferable

into other areas of comms as well.

93

: 00:05:23

Generally,

I've spent my career in the public

94

: 00:05:26

and third sectors, so purpose

driven work is really important to me.

95

: 00:05:29

I've overseen columns in Europe

for the Urban Land Institute,

96

: 00:05:32

and then just before Ada,

I was head of comms at Working

97

: 00:05:36

Families, the national charity

for working parents and carers.

98

: 00:05:39

So yeah.

99

: 00:05:40

And then and then Ada happened.

100

: 00:05:42

And as I said, I've never worked

in the tech space before.

101

: 00:05:45

I'm a digital native,

but I definitely walked into Ada

102

: 00:05:48

with very little knowledge

on the AI ecosystem.

103

: 00:05:51

And the complex dynamics around it.

104

: 00:05:54

Luckily, I work with experts,

and I would say I probably spent

105

: 00:05:58

the first three months in the role

just asking my colleagues

106

: 00:06:02

to explain things to me

like I'm five years old.

107

: 00:06:05

I think that was my constant refrain,

just boiling things

108

: 00:06:08

down to their simplest explanations,

simplest parts,

109

: 00:06:12

so I could really get my brain around

what we were talking about.

110

: 00:06:16

That is, great, great advice,

I think, isn't it?

111

: 00:06:19

I mean, I think often when we're plunged

into an unfamiliar environment

112

: 00:06:22

and we feel like we're supposed to, you

know, if it's a professional environment

113

: 00:06:26

where we're working within it,

this temptation is sometimes to sort of,

114

: 00:06:30

you know, pretend that you know more than

you do, but sort of just asking questions

115

: 00:06:34

and being honest about what you don't know

is always the best way to learn.

116

: 00:06:38

So that that sounds, yeah, great advice.

117

: 00:06:40

And it's funnily enough, actually,

118

: 00:06:41

I think asking asking for things

to be explained,

119

: 00:06:45

like I'm a five year old is pretty much

all I use ChatGPT for at the moment.

120

: 00:06:50

So it's quite, quite interesting.

121

: 00:06:52

That would go.

122

: 00:06:52

I mean, it's always yeah,

123

: 00:06:53

especially in the pretend

I'm doing it for the benefit of my kids,

124

: 00:06:57

but only because that's the level stuff

needs to be pitched.

125

: 00:07:00

Bit as well.

126

: 00:07:00

But yeah, so that's amazing.

127

: 00:07:02

So you, you sort of now fully immersed

in this world, I'm sure, still asking

128

: 00:07:06

lots of questions, but presumably by now,

you know, you sort of

129

: 00:07:10

have a much stronger grasp of the field

than you did when you started.

130

: 00:07:13

And in the time that you've been working

there,

131

: 00:07:18

this podcast is all about complexity

132

: 00:07:21

with, you know, associated

with the communication of research.

133

: 00:07:25

So is there one

134

: 00:07:28

area of what you do,

one aspect of your work

135

: 00:07:31

that has come across

as particularly complex? Ada.

136

: 00:07:36

Yes, I,

137

: 00:07:38

I think we can really go back to basics

with this

138

: 00:07:41

and just talk about the definition

of AI itself.

139

: 00:07:44

Right?

140

: 00:07:45

I think that a definition

141

: 00:07:48

of what you work on seemingly

should be this very straightforward thing.

142

: 00:07:52

Right?

143

: 00:07:52

But with, AI is so rapidly evolving, it's

hard to pin down.

144

: 00:07:58

You know, I think if we had to define

AI a year and a half ago, pre ChatGPT,

145

: 00:08:03

the definition would have been

entirely different than it is today.

146

: 00:08:07

And so I think that complexity, it

it introduces

147

: 00:08:11

this tension and kind of discomfort

from a comms perspective.

148

: 00:08:14

Because if you think about, for example,

an elevator pitch,

149

: 00:08:19

you really want to just have that.

150

: 00:08:20

Now have that be evergreen,

leave it alone.

151

: 00:08:24

That's the end of it.

152

: 00:08:25

But with something like AI

and also data driven

153

: 00:08:28

technologies as well, it's

not that simple and straightforward.

154

: 00:08:31

And so if we've really had to reckon

with this tension,

155

: 00:08:37

and I think what it really comes down to

is the fact that there's not one

156

: 00:08:42

definition that is going to satisfy

every context and satisfy every audience.

157

: 00:08:48

So I think it's about, for us

158

: 00:08:49

at least at Ada, it's

about going back to our evidence base.

159

: 00:08:53

It's about acknowledging the limitations

of the definitions upfront,

160

: 00:08:57

actually

saying that as part of defining AI,

161

: 00:09:00

that it's a rapidly evolving area, there's

not a universally accepted definition.

162

: 00:09:05

And then looking to our evidence base,

163

: 00:09:07

ensuring that we're approaching

the definition from a very Ada

164

: 00:09:10

perspective, which for us

is looking at people and society,

165

: 00:09:13

the interaction between these technologies

and people in society.

166

: 00:09:17

There's a word

for that called sociotechnical.

167

: 00:09:20

And then and then reevaluating

the definition periodically.

168

: 00:09:23

Right.

169

: 00:09:23

So it's in a complex and rapidly

170

: 00:09:26

moving area like AI and data.

171

: 00:09:29

You have to reevaluate your comms and

messaging and definitions periodically.

172

: 00:09:33

It's not going to be something

that's evergreen.

173

: 00:09:35

It's not going to be something

that's static.

174

: 00:09:36

And so that can be uncomfortable

for Collins.

175

: 00:09:39

But I think you just sort of

have to embrace

176

: 00:09:41

the awkwardness and discomfort

and just reevaluate.

177

: 00:09:45

It's messy. Right.

178

: 00:09:47

That's that's so interesting.

179

: 00:09:48

You said, you know, I think one way

180

: 00:09:52

to to deal with that complexity

is to to remain constantly focused

181

: 00:09:56

on, on Ada's mission

of considering these emerging technologies

182

: 00:10:00

and AI in the context of it

impact on society.

183

: 00:10:05

That makes me think about,

you know, your audience, but, well,

184

: 00:10:09

the mission is, is making AI and data

185

: 00:10:12

work for people in society,

but actually the audience itself.

186

: 00:10:15

Interestingly, your audience from a comms

point of view is not society's,

187

: 00:10:19

such as the policymakers

who work on behalf of that society.

188

: 00:10:24

How are you considering that audience?

189

: 00:10:26

There's another way of sort of

cutting through the complexity.

190

: 00:10:29

Talk a bit about who they are, because

I think that's one of the things as well.

191

: 00:10:32

Our policymakers are often considered

to be sort of a bit of a homogenous

192

: 00:10:36

mass, as it were, but very, you know,

there's a lot of different elements

193

: 00:10:40

within policymaking and different roles

and different values and needs and so on.

194

: 00:10:44

So can you talk to us

195

: 00:10:45

about a bit about that audience

and how you take them into consideration?

196

: 00:10:49

Yeah, absolutely.

197

: 00:10:50

So, so one of our key audiences

and one of our key routes to impact

198

: 00:10:55

when you're thinking about legislation,

best practice

199

: 00:10:59

when it comes to

AI and data is policymakers.

200

: 00:11:02

And I think,

201

: 00:11:04

my personal

202

: 00:11:05

definition of policymakers

would extend beyond Parliament.

203

: 00:11:08

Right.

204

: 00:11:09

It would be, decision

makers and local authorities.

205

: 00:11:12

It would be decision makers in the NHS,

anywhere where these technologies

206

: 00:11:16

are being deployed and have the potential

to impact the lives of everyday people.

207

: 00:11:21

And so, like as you said, first of all,

policymakers are not a monolith.

208

: 00:11:26

This encompasses

a lot of different people.

209

: 00:11:29

But if you boil it down

well, I think it's easy to think

210

: 00:11:33

that policymakers might be subject matter

experts.

211

: 00:11:36

Right.

212

: 00:11:37

But in reality,

they really need that complexity

213

: 00:11:40

distilled down to the question

of how will this thing,

214

: 00:11:44

how will this technology impact

the daily lives of the people I represent,

215

: 00:11:49

of the people who use the service

that I'm in charge of, etc.?

216

: 00:11:53

And I think for us,

if we can always keep that question

217

: 00:11:57

in the back of our minds, that will really

help us communicate with policymakers.

218

: 00:12:01

I also think that just

in very practical terms, and this is a

219

: 00:12:05

I think this is probably common knowledge.

220

: 00:12:06

Policymakers are incredibly time poor.

221

: 00:12:10

And so anything that is overly long,

222

: 00:12:14

overly complex and sort of riddled with

jargon is not going to get read.

223

: 00:12:19

And so I think it's about really it's

that translation aspect coming in.

224

: 00:12:23

Right?

225

: 00:12:23

So if you have a very dense

research report,

226

: 00:12:27

you also need a translation

that will work for this audience.

227

: 00:12:29

And I

228

: 00:12:30

there's some research

and I wish I could cite it because it's

229

: 00:12:33

really interesting that someone on my team

brought it up the other day,

230

: 00:12:36

which says that policymakers like to read

a short briefing like the two pager,

231

: 00:12:42

but they like to know

232

: 00:12:43

that there's a 100 page paper

that is backing up that two page paper,

233

: 00:12:47

even if they don't really want to read

that that longer output.

234

: 00:12:50

So that's a really interesting dynamic

as well.

235

: 00:12:53

Yeah. That's fascinating.

236

: 00:12:54

I interviewed Caroline Wood

from the University of Oxford recently.

237

: 00:12:58

She's done

she did a big piece of research,

238

: 00:13:02

a report into communicating evidence

to policymakers.

239

: 00:13:05

And one of the other things

that she mentioned, in addition

240

: 00:13:08

to the sort of the overwhelm

that lots of them feel about

241

: 00:13:11

just the quantity of information they're

having to process on a daily basis.

242

: 00:13:15

And again, as you said,

you know, they're they're not all expert.

243

: 00:13:20

They can't all be experts

in every field that they're dealing with.

244

: 00:13:22

So it does need to be explained

accessibly.

245

: 00:13:24

The other aspect

was that they really appreciate

246

: 00:13:28

kind of policy recommendations as well.

247

: 00:13:30

I suppose not just here's the information,

248

: 00:13:32

but here are some of the sort of actions

that could be taken off the back of that.

249

: 00:13:37

Is that, something that you,

that you work to do as well?

250

: 00:13:41

Yes. This I mean, this really chimes

with our strategy as well.

251

: 00:13:44

So I think it's I'm thinking back

to theories I learned in graduate school.

252

: 00:13:49

This really comes down to choice

architecture, right.

253

: 00:13:52

So if you are giving choices

to these policymakers,

254

: 00:13:55

if you are kind of doing

some of the legwork for them,

255

: 00:13:59

it makes it so much easier

for them to take action.

256

: 00:14:02

And so as much as possible, particularly

when we're writing a policy

257

: 00:14:05

facing briefing,

258

: 00:14:07

the recommendations

are the sort of the thrust of the report

259

: 00:14:11

that the most important part of the report

260

: 00:14:13

and we work really hard

to get those right.

261

: 00:14:15

And what sort of recommendations and and

communications are you having with them?

262

: 00:14:19

Do you have any sort of examples of any of

the research that you've been doing?

263

: 00:14:23

Ada. And the outputs that were directed

towards policymakers?

264

: 00:14:28

I did, so I, I'd like to give

265

: 00:14:30

an example of a summit

that happened last year.

266

: 00:14:35

So this was the global AI summit, that it

it was in the news.

267

: 00:14:40

You may have heard about it.

268

: 00:14:41

It took place at Bletchley.

269

: 00:14:43

It was a

it was organized by the UK government

270

: 00:14:46

and it was a gathering of world leaders,

271

: 00:14:49

luminaries, tech company executives,

not so much civil society,

272

: 00:14:54

which we will get to

to talk about the future of AI.

273

: 00:14:58

And it was really framed around

this existential risk of AI.

274

: 00:15:03

Right.

275

: 00:15:03

So the idea that in the distant future

276

: 00:15:07

I could be used, maliciously

277

: 00:15:10

in terms of nuclear warfare

278

: 00:15:13

and sort of apocalyptic scenarios,

that kind of thing.

279

: 00:15:17

And Ada

wanted to bring a new perspective to that.

280

: 00:15:22

Ada, as well as our colleagues

in civil society, wanted to bring

281

: 00:15:25

a new perspective to that, which is that

harms are actually happening right now.

282

: 00:15:30

When you apply for a loan, AI can decide

whether or not you get a loan.

283

: 00:15:36

When you apply for benefits,

it's the same thing.

284

: 00:15:39

Biometrics and facial recognition

technology

285

: 00:15:42

can have adverse impacts when it comes

to misidentifying someone as a criminal.

286

: 00:15:47

So these are real harms happening now.

287

: 00:15:50

So we're not really wanting to talk

about the distant future when there are

288

: 00:15:54

things that are happening right now.

289

: 00:15:55

And that was a framework that wasn't

really in place for the summit.

290

: 00:15:58

And so we, along with our civil society

291

: 00:16:02

colleagues, work

together to sort of shift this narrative.

292

: 00:16:05

And there are a few ways

that we did this right.

293

: 00:16:07

So first of all, we wanted to provide

a frame

294

: 00:16:11

that connected to our mission of AI

and data working for people in society.

295

: 00:16:16

Right.

296

: 00:16:16

And so when our director had

the opportunity to speak at the summit,

297

: 00:16:20

her first words were,

298

: 00:16:23

this summit

299

: 00:16:23

isn't fundamentally about technology

or regulation.

300

: 00:16:26

It is about people,

the people who make technologies,

301

: 00:16:29

the people who regulate them,

302

: 00:16:31

and most importantly, the people

whose lives are affected by them.

303

: 00:16:35

At this lot,

these lines were incredibly powerful,

304

: 00:16:38

and our director

had multiple senior people in government

305

: 00:16:42

coming up to her after her speech

and acknowledging the importance

306

: 00:16:46

of this framing and the fact that it

it wasn't loud enough, this narrative.

307

: 00:16:51

So that was really powerful.

308

: 00:16:53

First of all, so the first thing we did

was provide a frame.

309

: 00:16:56

The second tactic we used was giving

examples and analogies that really stick.

310

: 00:17:01

So often talking about something

like regulation

311

: 00:17:04

or governance is really intangible,

and you need these analogies

312

: 00:17:08

so that they just stick in people's

brains. Right.

313

: 00:17:11

So we had a few things that we want to see

314

: 00:17:14

in the regulation of AI,

315

: 00:17:17

and we connected them to things

that are already happening.

316

: 00:17:19

So for example, for AI manufacturers,

in the same way that cars are crash

317

: 00:17:24

tested, liable

318

: 00:17:26

for when things go wrong, like when

a restaurant gives you food poisoning,

319

: 00:17:30

and then a voice for people

who were going to be impacted,

320

: 00:17:33

like what happens

with planning permission.

321

: 00:17:35

So we use these three everyday examples

that most people are going to know about,

322

: 00:17:39

and we connected them

to how we think I should be governed

323

: 00:17:43

and regulated and

and those are really effective as well.

324

: 00:17:46

A third tactic,

which is one of my favorites,

325

: 00:17:48

one of my personal favorites

is if you want to influence policymakers,

326

: 00:17:54

it's really helpful to use their messaging

to get your own messaging across.

327

: 00:17:58

So one of the key messages

ahead of this summit

328

: 00:18:03

was that the UK wants to be an

AI superpower.

329

: 00:18:06

They want to be at the forefront

of innovation.

330

: 00:18:09

Right.

331

: 00:18:10

And so we made sure to use this language

to say that if the UK

332

: 00:18:15

is going to be an AI superpower, then

we need protections for everyday people.

333

: 00:18:20

We need a safety net, we need regulation.

334

: 00:18:23

So that was a tactic for us to be able to

335

: 00:18:29

turn the government's

messaging on its head

336

: 00:18:32

and use it to make sure

that we really emphasize the importance

337

: 00:18:36

of people in society

and this ecosystem really.

338

: 00:18:40

You mentioned that

339

: 00:18:42

you'll come back to it, which was perhaps

a lack of representation of,

340

: 00:18:46

I think you said civil society,

the Bletchley Park summit,

341

: 00:18:51

even though your target audience

then is policymakers.

342

: 00:18:54

How are you actually involving people?

343

: 00:18:57

I guess not that policymakers

aren't people, but other people,

344

: 00:19:01

other members of society

in the research that you do.

345

: 00:19:04

And how do they

the sort of ideas and values and concerns

346

: 00:19:08

perhaps feed into the research

and the reports that you're creating?

347

: 00:19:12

Public participation research

is a really important workstream at Ada,

348

: 00:19:17

and it actually cuts across

a lot of the research that we do.

349

: 00:19:21

So one of our main objectives

as an organization is to elevate

350

: 00:19:26

the voice of the public

to understand what they want and need.

351

: 00:19:31

When it comes to AI, one way

we did this in the context of the summit

352

: 00:19:35

is that we supported a brilliant

organization called Connected by Data

353

: 00:19:39

to hold a public deliberation on AI.

354

: 00:19:43

So it was essentially a five day exercise

where a group of members of the public

355

: 00:19:49

heard from experts, heard sessions

that were happening both at the Bletchley

356

: 00:19:54

AI summit and also the AI fringe event

that was happening in London as well,

357

: 00:19:59

and then came together

to make some recommendations

358

: 00:20:01

for what they want to see in terms

of governance and regulation

359

: 00:20:04

and how people can be protected,

as these technologies involved.

360

: 00:20:09

So that was

that was a really interesting and concrete

361

: 00:20:14

way that the public can be involved

in these decisions.

362

: 00:20:17

And I think that's probably

another podcast actually.

363

: 00:20:19

Like, how can you use public voice

to shape policy?

364

: 00:20:24

But it's definitely something

that's really important to us at Ada.

365

: 00:20:27

And I would say from a comms

perspective to this summit

366

: 00:20:31

was a really unique situation for me

as a head of communications.

367

: 00:20:38

So in past organizations,

I've been used to the media coverage

368

: 00:20:43

being sort of the pinnacle of comms

victories, I guess.

369

: 00:20:47

Right.

370

: 00:20:48

So if it's what you want, you want to get

as much coverage as possible.

371

: 00:20:51

And what was really unique about this

summit

372

: 00:20:54

is that we had a convening role at Ada.

373

: 00:20:57

So we helped to amplify

civil society voice.

374

: 00:21:01

That was our objective,

even if that wasn't necessarily our voice.

375

: 00:21:05

So there was a real sense of community.

376

: 00:21:07

We were sharing media opportunities

around other civil society organizations,

377

: 00:21:12

which with my PR hat on,

I had never really done that before.

378

: 00:21:16

Usually

you're very protective of that, right?

379

: 00:21:18

But I think what happened

is the diversity of those voices

380

: 00:21:22

really did shift

the narrative in the media as well.

381

: 00:21:25

So the media had really been covering

this existential risk

382

: 00:21:28

that I talked about before,

sort of the apocalyptic scenarios.

383

: 00:21:32

And I really think and we

we monitored the media coverage

384

: 00:21:36

and even to this day,

the coverage of existential risk

385

: 00:21:40

is almost nil,

because I really think in that moment,

386

: 00:21:45

we, the, the media and policymakers alike

387

: 00:21:49

realized that that wasn't

really the issue to be talking about.

388

: 00:21:51

It is the current harms

that are happening now.

389

: 00:21:53

And a big a big indicator of impact for us

was that we saw

390

: 00:21:58

our messaging echoed in Kamala

Harris's remarks around the summit.

391

: 00:22:02

So she she made a very pointed remark

392

: 00:22:05

that it's not about these future

existential scenarios.

393

: 00:22:11

It's about the harms

that are happening here now.

394

: 00:22:14

And so that was a big win for us

and for the rest of civil society

395

: 00:22:17

that felt quite underrepresented

at the summit.

396

: 00:22:20

And where was that focus on the sort of

present day dangers and risks

397

: 00:22:26

over the existential?

398

: 00:22:27

Was that reflected in the concerns of the

through

399

: 00:22:30

the sort of public deliberations

that you did as well?

400

: 00:22:34

Yeah, absolutely.

401

: 00:22:36

And I think the, the,

the number one consensus

402

: 00:22:40

that came out of those deliberations and,

and that has come out of our research

403

: 00:22:44

as well, is that people,

the public wants regulation with teeth,

404

: 00:22:51

as in robust regulation and safeguards

as this sort of unpredictable

405

: 00:22:58

and new technology is deployed, more

and more in our everyday lives.

406

: 00:23:04

So this isn't a particularly

comms question, I guess is more

407

: 00:23:07

just a question about

because I'm interested to know

408

: 00:23:10

it's who stands to gain

from not having a new regulation.

409

: 00:23:14

It's industry.

410

: 00:23:14

It's this sort of technology companies.

411

: 00:23:16

Right? Yeah. That's it essentially.

412

: 00:23:18

That's it. Okay, fine.

413

: 00:23:19

That that's it. Exactly.

414

: 00:23:21

But but I think that

how how that becomes a comms question

415

: 00:23:26

is, is the whole regulate

to innovate theory.

416

: 00:23:30

Right.

417

: 00:23:30

So it's how can we

how can we convince tech companies

418

: 00:23:36

that regulation and innovation

are not mutually exclusive?

419

: 00:23:39

And obviously,

420

: 00:23:40

that's not just a comms question,

but comms plays a really big role in.

421

: 00:23:43

So that's in that storytelling. Yeah.

422

: 00:23:45

We spoken about it.

423

: 00:23:47

Well, obviously, you know, the work

that you and your team Ada have done from

424

: 00:23:51

as a comms team to,

to to get these messages out and so on,

425

: 00:23:55

how about the researchers themselves?

426

: 00:23:59

A I guess question is

what kind of research are they doing?

427

: 00:24:03

I'm sure it's very broad, but just yeah,

you know, it would be fascinating

428

: 00:24:07

to hear about some of the things

that they're exploring at the moment.

429

: 00:24:09

And secondly.

430

: 00:24:13

What's your

sort of working relationship with them?

431

: 00:24:15

Like are you trying to empower,

432

: 00:24:17

enable them to sort of

tell their own stories as well?

433

: 00:24:20

Do they just kind of getting on

434

: 00:24:21

with the research, feeding that to you,

435

: 00:24:23

and then you go

and communicate it on their behalf?

436

: 00:24:25

How does that, relationship work?

437

: 00:24:27

So first of all, we have some really

interesting projects coming out soon.

438

: 00:24:31

Our scope is very wide

in terms of the actual topics we cover,

439

: 00:24:36

because if you imagine

the intersection of technology

440

: 00:24:40

and people and society is very broad,

there are lots of ways

441

: 00:24:44

that people and society can be impacted

by the deployment of technology.

442

: 00:24:48

Some of the reports we're working on

right now are around,

443

: 00:24:52

I genomic powered health prediction,

which is really fascinating.

444

: 00:24:57

So it it's around

how I can be used to predict disease risk,

445

: 00:25:03

to predict the way that someone, responds

to a certain medication.

446

: 00:25:09

And the report is about the implications

of this positive and negative

447

: 00:25:12

that this could have on the health care

system and people in society.

448

: 00:25:16

We're doing a report on gender and AI,

which explores

449

: 00:25:20

the lived experience of transgender

and non-binary people

450

: 00:25:26

and how data driven systems in healthcare

have affected them.

451

: 00:25:31

And then we're also

we're doing a big workstream

452

: 00:25:33

on the use of AI and data driven systems

in public services as well.

453

: 00:25:38

I think often there is a lot of optimism

454

: 00:25:42

about how I can transform public services,

455

: 00:25:46

and we know at Ada

that there is a lot of nuance to this.

456

: 00:25:50

Technology doesn't exist in a vacuum

and has knock on effects for people

457

: 00:25:54

in society when it is deployed in a system

458

: 00:25:57

that already exists and functions.

459

: 00:26:00

So we are we're looking at those knock

460

: 00:26:03

on effects

and offering recommendations on how

461

: 00:26:08

this technology can be deployed

in the public sector in a responsible way.

462

: 00:26:11

So those are

some of the things we're working on.

463

: 00:26:14

And in terms of our actual research

process, this is also something unique

464

: 00:26:20

about Ada that I've never experienced

in another research institute,

465

: 00:26:25

that I've worked out before.

466

: 00:26:26

So typically the workflow, at other places

I've worked

467

: 00:26:31

would be that researchers

468

: 00:26:34

draft a report, fully pass

469

: 00:26:37

the draft to Collins, Collins polishes it,

and then it goes out into the world.

470

: 00:26:41

So there's very little dialog back

and forth at those early stage is

471

: 00:26:46

what's really unique about Ada

is that Collins is involved

472

: 00:26:50

from the inception of an idea.

473

: 00:26:53

And what that means is that at those very

early stages, we can define audiences.

474

: 00:26:59

We can ensure that the framing reflects

475

: 00:27:01

our mission, vision and values

as an organization.

476

: 00:27:05

Right.

477

: 00:27:05

So I think when you're writing

about technology,

478

: 00:27:08

it's very easy to fall into a slightly

479

: 00:27:13

jargony and overly technical way

of describing things.

480

: 00:27:16

So our comms team is there to ensure

that people in society are centered

481

: 00:27:21

and also that these reports are written

in a way that are accessible,

482

: 00:27:26

for the audiences

that we want to reach as well.

483

: 00:27:29

And so that what it means is

that the process is probably a bit longer,

484

: 00:27:34

than what it would be

if it were just sort of this, you know,

485

: 00:27:37

research has it and then comms has it

because there is lots of back and forth.

486

: 00:27:41

There's lots of dialog.

487

: 00:27:43

And we've been working

to really clearly define editorial roles

488

: 00:27:47

and responsibilities between research

and comms, so that everyone is comfortable

489

: 00:27:51

with the process

and that it runs as smoothly as possible.

490

: 00:27:55

But what what we get out of it

are evidence based, well-written,

491

: 00:28:00

and really robust reports, that

that we can be proud of and that

492

: 00:28:05

and that we know reflect really,

really excellent work.

493

: 00:28:09

And presumably,

494

: 00:28:11

you know, I don't know

if you've sort of had feedback

495

: 00:28:13

from the researchers about this,

but the hope would be that by asking

496

: 00:28:17

the questions

always about who this is for whilst

497

: 00:28:20

the research is happening,

you know, having that constant focus on,

498

: 00:28:24

I don't know what the term is, sort of end

user or, you know, the sort of people

499

: 00:28:28

who they're doing the research

on throughout the process, one would hope

500

: 00:28:32

would have a beneficial effect

on the research itself as well.

501

: 00:28:35

Yes, absolutely. Defining that.

502

: 00:28:37

Yeah.

503

: 00:28:37

Defining that even before

the research really begins.

504

: 00:28:40

And but also keeping in mind

that these things can shift and evolve.

505

: 00:28:44

Right.

506

: 00:28:44

So there's nothing about this process

that we want to be too

507

: 00:28:47

set in stone and too rigid

because we know that research can evolve.

508

: 00:28:52

We know that writing can evolve.

509

: 00:28:53

And and also the external environment

can evolve, right?

510

: 00:28:56

So can evolve.

511

: 00:28:58

So, you know,

there could be a surprise election.

512

: 00:29:02

There could be something that means

that your audience has changed and

513

: 00:29:06

and that's built into the process

processes.

514

: 00:29:08

This slight flexibility.

515

: 00:29:10

But if you have an idea

of who you're talking to

516

: 00:29:12

before

you even start conducting the research,

517

: 00:29:15

then that really helps

to focus the research team.

518

: 00:29:18

Yeah.

519

: 00:29:18

So that's really interesting

to hear about,

520

: 00:29:20

sort of the process of that report

writing embedded in with the researchers

521

: 00:29:24

and the reports themselves, presumably,

you know,

522

: 00:29:27

they're quite sort of text heavy,

I suppose, sort of dominated

523

: 00:29:31

by copy and the writing

and the statistics and findings and so on.

524

: 00:29:35

One of the areas

I sort of interested in as well

525

: 00:29:38

was AI communications

about artificial intelligence, visuals.

526

: 00:29:42

I spoke a few years ago with Research

cats.

527

: 00:29:45

Do you how who was talking

about global narratives in AI?

528

: 00:29:48

And one of the side projects she was doing

529

: 00:29:49

was about representing regions of,

of artificial intelligence.

530

: 00:29:53

And it was all those sort of

531

: 00:29:55

glossy

532

: 00:29:56

white robots,

basically a sort of one of the things,

533

: 00:30:00

how are you coping

534

: 00:30:03

with that side of things

with the visuals? Yes.

535

: 00:30:06

I think this is a perennial challenge

for anyone in comms, who works in the tech

536

: 00:30:11

or AI space is that if you go on a stock

537

: 00:30:15

photo library, you're exactly right.

538

: 00:30:18

What you will see is fluffy white robots.

539

: 00:30:21

You will see Tech Blue, as we call it,

which is a very specific

540

: 00:30:25

shade of blue

that's used to illustrate technology.

541

: 00:30:29

You will see lots of people holding

smartphones right.

542

: 00:30:32

You will see very futuristic images.

543

: 00:30:36

And for us as a brand, this does not

reflect what we want to talk about.

544

: 00:30:41

As I said at the beginning of this chat,

545

: 00:30:44

we want to focus on people in society,

not on the actual technologies themselves.

546

: 00:30:49

And the solution that we found is that we

are actually we're working on a project

547

: 00:30:53

now where we are creating our own library

548

: 00:30:57

of illustrations alongside a designer.

549

: 00:31:01

We're focusing on our areas of work,

which are society,

550

: 00:31:04

justice and public services,

emerging technologies,

551

: 00:31:08

policy, governance and regulation,

and then public participation.

552

: 00:31:14

It is a real

553

: 00:31:16

challenge to illustrate

something like governance or regulation.

554

: 00:31:20

And we've had some really interesting

chats with our designers

555

: 00:31:24

about how to how to convey these things

556

: 00:31:28

in images in a way

that also centers people in society.

557

: 00:31:32

So this is definitely a work

in progress for us, but it's something

558

: 00:31:36

that will hopefully be

a really useful tool for us and our brand.

559

: 00:31:40

Right.

560

: 00:31:41

And when we when will we be able

to see those illustrations?

561

: 00:31:44

Do you have a.

562

: 00:31:45

We're hoping

by the end of the year. Great.

563

: 00:31:47

I look forward to it.

564

: 00:31:48

Before I'm going to ask

a couple of sort of final questions

565

: 00:31:51

that are a bit more general.

566

: 00:31:52

But before we move away from I, I do have

one question I'm very curious to know

567

: 00:31:56

because there's a lot of conversations,

568

: 00:31:58

obviously, around

how is AI impacting society?

569

: 00:32:03

There's a lot of conversations

570

: 00:32:05

within the communications, research,

communications sector around

571

: 00:32:08

how AI is going to impact

in that sector in particular.

572

: 00:32:11

So very curious to know

your thoughts on on that, on how you feel.

573

: 00:32:16

It's going to sort of influence

and transform

574

: 00:32:19

the work of professional communicators

and also how you currently, if at all.

575

: 00:32:24

I imagine you are in some capacity

using AI in

576

: 00:32:28

your own kind of daily or weekly

workflows.

577

: 00:32:32

This is my personal view,

and I'm very much

578

: 00:32:35

in the creative writing

editorial space. I.

579

: 00:32:40

I realize that other comms professionals

might might think of using

580

: 00:32:45

AI in different ways.

581

: 00:32:46

So I think that generative

AI is the thing that I think about

582

: 00:32:50

when it comes to potential threats

for comms professionals.

583

: 00:32:54

And at the moment,

I can say with confidence that ChatGPT

584

: 00:32:57

is not going to replace any good writers.

585

: 00:33:01

I can give you an example of that.

586

: 00:33:03

I, I occasionally use ChatGPT

587

: 00:33:06

to brainstorm title ideas or reports, so,

588

: 00:33:11

you know, giving a prompt

that says, I'm writing a report about X,

589

: 00:33:14

please could you give me ten titles

that have puns?

590

: 00:33:18

I think one example recently is that

we were writing a report about education,

591

: 00:33:22

and one of the titles it gave back to me

592

: 00:33:24

was eye of the Tiger

Taking a Bite out of Education.

593

: 00:33:30

I, I,

594

: 00:33:31

I just I'm very confident by the,

by the outputs

595

: 00:33:35

they get from ChatGPT that they are not

replacing actual writers any time soon.

596

: 00:33:39

I'm very confident with that. Yeah.

597

: 00:33:41

Are there any sort of useful?

598

: 00:33:43

I mean, sounds like

you're experimenting with it thus far.

599

: 00:33:45

It's not doing the job that,

600

: 00:33:47

you know, doing that job,

but are there any useful applications

601

: 00:33:51

you're finding or you're still just

in the kind of experimental phase?

602

: 00:33:54

So our data, we are really cautious about

603

: 00:33:58

using AI in our research

and in our evidence building.

604

: 00:34:02

And that's simply

because we want to make sure that we're

605

: 00:34:05

examining all of the knock on effects

of the technologies that we use.

606

: 00:34:09

And so because of that, like, ironically,

because of that,

607

: 00:34:12

I think we're probably slower

608

: 00:34:13

adopters of these technologies because

we really want to just examine them.

609

: 00:34:17

I mean, I would say that in a

perfect world, if there were some kind of

610

: 00:34:23

AI system that generated perfect footnotes

that I didn't have to edit,

611

: 00:34:27

I would really appreciate,

I would pay a lot of money for that.

612

: 00:34:31

I have not found that yet,

613

: 00:34:32

but that would be my

that would be my dream application of AI.

614

: 00:34:36

Yeah, I think that's it, isn't it?

615

: 00:34:37

Most people's hope is that

it will be able to replace the sort

616

: 00:34:40

of the drudgery of some of the work,

the sort of more mundane tasks

617

: 00:34:44

such as the fun, creative.

618

: 00:34:45

Yes, that's really fascinating.

619

: 00:34:47

Thank you.

620

: 00:34:47

So finally, a couple of questions.

621

: 00:34:50

One, that AI questions

that I ask all my guests.

622

: 00:34:52

The first is whether you have a particular

piece of advice that's been given to you

623

: 00:34:58

throughout your career and comms

that you come back to most often,

624

: 00:35:02

or that sort of help to help steer

your the way you work.

625

: 00:35:06

For me, that's a really easy question.

626

: 00:35:08

It is always go back to the strategy.

627

: 00:35:10

When in doubt,

always go back to the strategy.

628

: 00:35:14

Go back to the mission.

629

: 00:35:16

When it comes to making tough

630

: 00:35:18

decisions

when it comes to crisis communications,

631

: 00:35:23

just go back to the strategy.

632

: 00:35:24

What does your organization stand for?

633

: 00:35:27

Who are the people that you're talking to?

634

: 00:35:30

Yeah, but it's just a fundamental thing.

635

: 00:35:32

And it's helped me through

lots of tough situations.

636

: 00:35:36

That's the sort of the overarching

strategy of the

637

: 00:35:39

of the organization as a whole.

638

: 00:35:40

The comms strategy. Exactly, exactly.

639

: 00:35:44

Yeah.

640

: 00:35:44

And is that

641

: 00:35:44

I mean, that's something presumably

periodically you're looking to kind of

642

: 00:35:48

renew or revisit those strategies

643

: 00:35:51

anyway in other sort of evolving

processes, right?

644

: 00:35:54

Yes, absolutely.

645

: 00:35:55

And Ada, we have a new director

and we're actually working on a new Strat,

646

: 00:35:59

a new three year strategy right now,

which should launch next year.

647

: 00:36:03

And so, yes, I think this might tie back

to what I was saying

648

: 00:36:05

about the definitional piece earlier

in that it

649

: 00:36:09

it feels comfortable to have something

that's evergreen,

650

: 00:36:12

but in reality, a definition will evolve,

an organization will evolve.

651

: 00:36:17

And I'm lucky to work in an organization

where comms

652

: 00:36:20

is always a really close collaborator.

653

: 00:36:23

But those processes.

654

: 00:36:25

Final question is a recommendation.

655

: 00:36:28

So often people recommend

a particular book that has

656

: 00:36:32

sort of inspired the work that they do,

but it could be anything,

657

: 00:36:35

could be documentary,

radio show, podcasts, whatever it is.

658

: 00:36:38

What? What's yours?

659

: 00:36:39

So mine is very topical.

660

: 00:36:41

It's not exactly about communication,

but it is the perfect example

661

: 00:36:46

of using storytelling

to illustrate a complex issue.

662

: 00:36:50

So it is a new documentary

called daughters.

663

: 00:36:54

It's on Netflix in the UK,

and it zooms in on a father

664

: 00:36:59

daughter dance that is being held

in a prison in Washington, DC.

665

: 00:37:04

So that's the story that's being told.

666

: 00:37:06

You see interviews with the fathers,

667

: 00:37:10

you see the daughters outside of prison

living their everyday lives and the

668

: 00:37:14

struggles that they're encountering,

you know, with their fathers in prison.

669

: 00:37:18

And then everyone

is preparing for this dad.

670

: 00:37:20

So that is the story that's being told.

671

: 00:37:23

But it illustrates the systemic problems

of racism,

672

: 00:37:27

of poverty, of poverty,

of the prison industrial complex,

673

: 00:37:31

and without actually,

674

: 00:37:34

you know, being heavy handed about it

and talking about these things

675

: 00:37:37

in an academic way, it shows it through

the stories of these actual human beings.

676

: 00:37:42

And it is so powerful and moving.

677

: 00:37:44

And I think it's a really good lesson,

actually.

678

: 00:37:47

I will I can't wait to check

that out. Thank you so much.

679

: 00:37:51

What a fascinating and fun chat that was.

680

: 00:37:54

I'm enormously grateful

to Catherine for her time,

681

: 00:37:57

and also very grateful to everybody

working at the Ada Lovelace Institute

682

: 00:38:02

for all they're doing to keep society's

interests front and center

683

: 00:38:06

when it comes to the development

of new technologies involving data and AI.

684

: 00:38:12

Next episode is going to be

another research unraveled extra.

685

: 00:38:16

That's our bonus episode where Bianca,

my co-host, and I will reflect

686

: 00:38:20

on this interview and more importantly,

we'll hear what you thought about it.

687

: 00:38:24

So please do let us know what resonated.

688

: 00:38:28

Do you have any of your own tips

for how to engage with policy makers?

689

: 00:38:31

I know that this is an area that lots of

you want to explore in greater depth,

690

: 00:38:35

so it would be wonderful to hear

what you think.

691

: 00:38:38

As ever, you can either write to me

directly at Peter at Orinoco comms.com

692

: 00:38:43

spark conversation on social media

using the hashtag Research Unraveled.

693

: 00:38:48

Or if you'd like your actual voice

to be heard and featured in that bonus

694

: 00:38:52

episode, you can drop us a voice memo

by at the link in the episode notes.

695

: 00:38:58

If you do one thing today to

696

: 00:38:59

help us, it would be to share this episode

with someone in your network.

697

: 00:39:03

If you could do two things,

698

: 00:39:05

it would be to also leave a review and

rating wherever you listen to podcasts.

699

: 00:39:09

All would be massively appreciated

and help us to grow this podcast

700

: 00:39:13

for the benefit of the wider

research communications community.

701

: 00:39:17

So thank you and see you next time.