Latest / 6 Ps in AI Pods (AI6P) / 🗣️ AISW #133: Radina Shipkovenska, UK-based strategy and growth advisor

Transcript

🗣️ AISW #133: Radina Shipkovenska, UK-based strategy and growth advisor · 6 Ps in AI Pods (AI6P)

0:00
SPEAKER_00: Hi everyone.
0:00
SPEAKER_00: This is Karen Smiley.
0:02
SPEAKER_00: Welcome to 6P's in AI Pods.
0:07
SPEAKER_00: This post is part of our AI6P interview series on AI software and wetware.
0:13
SPEAKER_00: Our guests share their experiences with using AI and how they feel
0:17
SPEAKER_00: about AI using their data and content.
0:23
SPEAKER_01: I'm delighted to welcome Rodina Shipkowlenska from the UK and originally from Bulgaria.
0:29
SPEAKER_01: As my guest today on AI, Software and Wetware.
0:32
SPEAKER_01: Roddy, thank you so much for joining me on this interview.
0:34
SPEAKER_01: Please tell us about yourself, who you are and what you do.
0:37
SPEAKER_02: Yeah, well, thank you very much for having me, Karen.
0:40
SPEAKER_02: I'm delighted to be here today with you.
0:41
SPEAKER_02: I'm Braddy.
0:43
SPEAKER_02: My background is economics and finance.
0:46
SPEAKER_02: I've always kind of been working within that three minutes,
0:48
SPEAKER_02: then I've moved to data,
0:50
SPEAKER_02: then I did a bit of a lateral move to strategies, non-linear path, I guess.
0:54
SPEAKER_02: I've been always following what I was curious about and what I wanted to explore,
0:58
SPEAKER_02: but I'm a mix of strategist and data.
1:01
SPEAKER_02: What that actually means is that I'm working with companies
1:03
SPEAKER_02: and helping them sell their new propositions for products to a new market or think
1:08
SPEAKER_02: of a strategy of how they can launch a new product.
1:10
SPEAKER_02: So I'm a strategy director at a company called WPP,
1:14
SPEAKER_02: which is an advertising and marketing conglomerate.
1:17
SPEAKER_02: If people know anything about marketing, probably they know Ogilvy.
1:20
SPEAKER_02: So it's one of the OG marketing companies.
1:23
SPEAKER_02: If you think about Mad Men, they actually took a lot of inspiration from Ogilvy.
1:27
SPEAKER_02: So they're part of the group.
1:28
SPEAKER_02: But also over the years,
1:29
SPEAKER_02: they started acquiring management consultancies and I'm part of
1:32
SPEAKER_02: the management consultancy arm of the marketing company.
1:35
SPEAKER_01: It's a bit of a weird way of
1:36
SPEAKER_02: explaining what I do.
1:37
SPEAKER_02: Lots of people don't exactly get what I do, but let's put it this way.
1:41
SPEAKER_02: I'm using numbers and strategy to help companies launch their products to new markets.
1:48
SPEAKER_01: That sounds good.
1:49
SPEAKER_01: Tell us a little bit about your level of experience with AI,
1:52
SPEAKER_01: machine learning or analytics,
1:54
SPEAKER_01: whether you use it professionally or personally, or if you've ever studied it.
1:58
SPEAKER_01: Formally or informally, neither is mine.
2:01
SPEAKER_02: I haven't studied AI or machine learning, but I've always been really fascinated by this world.
2:07
SPEAKER_02: As I was mentioning my backgrounds in data, I've done advanced maths as well as my high school.
2:12
SPEAKER_02: So I was really always fascinated by how we can use math
2:16
SPEAKER_02: and how we can use engineering to solve challenges and problems.
2:20
SPEAKER_02: I also have a brother that I'm really close with and he's in computer science
2:24
SPEAKER_02: and he's developing specifically using AI.
2:27
SPEAKER_02: So we've always talked a lot about the uses of AI and how it can solve challenges and problems.
2:32
SPEAKER_02: I'm using it more in my personal life and professional settings as well in personal life.
2:37
SPEAKER_02: To be honest, I've been using it for very small things.
2:40
SPEAKER_02: I'm a bit ashamed to admit it,
2:41
SPEAKER_02: but sometimes when I need to do a bit of quick research,
2:44
SPEAKER_02: what I need to buy from the grocery shop or when I need to do something,
2:48
SPEAKER_02: I'm doing a quick search or I need to do quick research about something.
2:52
SPEAKER_02: In the professional context,
2:54
SPEAKER_02: we've been using AI for a lot of the work that we're doing for our clients.
2:59
SPEAKER_02: So if you think about optimizing workflows or processes or even anything related
3:05
SPEAKER_02: to marketing,
3:06
SPEAKER_02: whether that's marketing copy,
3:07
SPEAKER_02: images, production, a lot of these things we actually implemented for our clients.
3:11
SPEAKER_02: So I've been using it quite a lot in both personal and professional setting.
3:16
SPEAKER_01: That sounds good.
3:17
SPEAKER_01: Okay.
3:18
SPEAKER_00: Can you give us a specific example,
3:20
SPEAKER_00: a story about a time
3:22
SPEAKER_01: that you've used a tool that has AI
3:24
SPEAKER_01: or machine learning features?
3:25
SPEAKER_01: I'm wondering what went well and what didn't go so well with using it.
3:33
SPEAKER_02: Claude quite a lot.
3:34
SPEAKER_02: And I know everyone is a big fan of Claude and specifically co-work.
3:38
SPEAKER_02: In my job, we do a lot of slides and we do a lot of research to start with.
3:41
SPEAKER_02: So I feel like it's been really great to do deep research,
3:45
SPEAKER_02: especially if you have the right prompts.
3:46
SPEAKER_02: And if you look at the specific industry or a client or a company that you want
3:50
SPEAKER_02: to look at as a stock price,
3:52
SPEAKER_02: or you want to look at their product portfolio and how they're presenting themselves
3:55
SPEAKER_02: to clients, you can do this very much in-depth work, which I find really valuable.
4:00
SPEAKER_02: But then a lot of the job also entails presentations and slides creation.
4:04
SPEAKER_02: And a lot of people are using it actually creating PDFs.
4:07
SPEAKER_02: And I know that's a very basic use case that we're discussing,
4:10
SPEAKER_02: but for me,
4:12
SPEAKER_02: I haven't been able to master slides or haven't been able to master it to the level
4:17
SPEAKER_02: of something that's presentable or something that's super useful.
4:21
SPEAKER_02: I've also created agents.
4:23
SPEAKER_02: So part of my job, I could create agents, I can create a persona.
4:27
SPEAKER_02: Like a strategist and then you can give it critical thinking.
4:29
SPEAKER_02: So do you want it to be pushing you to think in a specific way?
4:32
SPEAKER_02: Do you want it to use frameworks or do you want it to be actually more agreeable,
4:36
SPEAKER_02: more disagreeable?
4:37
SPEAKER_02: So I'll create different agents and depending on the context and the challenge
4:40
SPEAKER_02: that I have,
4:41
SPEAKER_02: battle with them or think of my ideas through and basically look at loopholes
4:45
SPEAKER_02: in my thinking and ask them to challenge me back.
4:51
SPEAKER_01: That sounds good.
4:51
SPEAKER_01: So what part of it works well and what part of that doesn't work so well?
4:56
SPEAKER_02: I think the reasoning is quite good.
4:58
SPEAKER_02: You can see very clearly the thought process,
5:00
SPEAKER_02: but I think what doesn't really work well is sometimes it's lacking context.
5:04
SPEAKER_02: So a lot of what I've been seeing is you really need to become very good
5:08
SPEAKER_02: at prompt engineering and giving a very specific context.
5:11
SPEAKER_02: Otherwise you go a hundred miles in a very different direction
5:14
SPEAKER_02: and it's very difficult to backtrack and actually see
5:17
SPEAKER_02: at what point things went wrong.
5:20
SPEAKER_02: I think this specifically what's something that I'm seeing with Claude
5:23
SPEAKER_02: and there's not criticism with them,
5:24
SPEAKER_02: but it's just something I'd see with them specifically is that the prompt needs
5:29
SPEAKER_02: to be perfect.
5:29
SPEAKER_02: Otherwise it goes like a hundred miles an hour into very, very wrong direction.
5:34
SPEAKER_02: Yeah.
5:36
SPEAKER_01: Can you think of an example of something that you were trying to get it to do and
5:40
SPEAKER_01: then you prompted it and it went off in the wrong direction
5:44
SPEAKER_01: on not giving you what you wanted?
5:46
SPEAKER_02: Yeah, well,
5:47
SPEAKER_02: for example,
5:48
SPEAKER_02: I was doing research about a pharma company and what was the future of pharma
5:54
SPEAKER_02: that you have AI and actually you can accelerate the development of drugs
5:59
SPEAKER_02: and what this could entail innovation and drug development.
6:02
SPEAKER_02: But I didn't give it a very specific context of the fact where the company is based in the U.S.
6:08
SPEAKER_02: I kind of missed this part and obviously you have very different regulations
6:11
SPEAKER_02: between EU and U.S. and missing a very, very small piece of the puzzle.
6:16
SPEAKER_02: Basically I was starting to say that the industry is going to grow
6:19
SPEAKER_02: into exponential X amount of numbers.
6:21
SPEAKER_02: It was making up numbers and it was hallucinating and it was saying something
6:26
SPEAKER_02: that I was just reading, it's like, doesn't make sense.
6:28
SPEAKER_02: I know that the farm industry is going to grow, but not as much.
6:31
SPEAKER_02: I really couldn't find a way to backtrack in correct thinking.
6:36
SPEAKER_01: Okay.
6:36
SPEAKER_01: So that's an example where it didn't do a very good job.
6:39
SPEAKER_01: Is there an example where you could think of that it did do a good job?
6:43
SPEAKER_02: Oh, lots of examples.
6:44
SPEAKER_02: As a non-native speaker,
6:47
SPEAKER_02: I think a lot of people would agree with me that it really tremendously helps
6:50
SPEAKER_02: with grammar check, corrections, copy, any of that.
6:55
SPEAKER_02: Image generation is actually quite good comparatively to a few years ago,
6:58
SPEAKER_02: especially because we do a lot of image generation in marketing.
7:02
SPEAKER_02: So I see that some of the challenges that we had a few years ago with people
7:05
SPEAKER_02: with six fingers, seven fingers, now it's no longer a challenge.
7:08
SPEAKER_02: So very quickly, I can get very good images that I can use.
7:13
SPEAKER_01: Okay.
7:14
SPEAKER_01: So it sounds like you're making good use of AI tools.
7:16
SPEAKER_01: Are there any things that you would avoid using AI for?
7:21
SPEAKER_01: Something that you say,
7:22
SPEAKER_01: I will never use AI to do X when you would not use it and why would you choose not
7:27
SPEAKER_01: to use it?
7:28
SPEAKER_02: Yeah.
7:29
SPEAKER_02: An example of where I wouldn't use it is when I need deep thinking on the problems.
7:33
SPEAKER_02: I use AI to do quick research.
7:36
SPEAKER_02: To get the admin work done.
7:38
SPEAKER_02: So for example, if I need to do research on the industry or a client to get some stats going on.
7:43
SPEAKER_02: But if I need to think about the strategies like deep cognitive loaded work,
7:48
SPEAKER_02: I wouldn't necessarily use AI because what I've been noticing a lot in part
7:54
SPEAKER_02: of my research with Google that I mentioned to you earlier was looking
7:57
SPEAKER_02: at why people are adopting AI and why not,
7:59
SPEAKER_02: is that there is this concept of cognitive surrender.
8:02
SPEAKER_02: So I see a lot of people trusting AI to the level that they think that everything
8:07
SPEAKER_02: that AI is giving them is true and they're kind of leaving their critical thinking
8:10
SPEAKER_02: behind.
8:11
SPEAKER_02: So I'm very cognizant of that, especially if it's deep work and it's something very creative.
8:16
SPEAKER_02: I'm trying to not outsource it to AI because I do see,
8:21
SPEAKER_02: especially with younger people that I'm working with,
8:24
SPEAKER_02: this tendency of cognitive surrendering using to outsource your thinking to AI.
8:31
SPEAKER_01: I'm chuckling here at you saying younger people because to me, you look very young.
8:36
SPEAKER_01: How young does she think that
8:37
SPEAKER_01: she's saying younger people?
8:40
SPEAKER_02: Thank you very much.
8:41
SPEAKER_02: I'm not that young, but we've done a few experiments with students.
8:45
SPEAKER_02: Those people between 23 and 27.
8:47
SPEAKER_02: We did different experiments and we were observing how people were using AI and
8:53
SPEAKER_02: for what kind of tasks.
8:55
SPEAKER_02: And we saw that people are more and more reliant for this kind of deep thinking.
8:59
SPEAKER_02: They wouldn't even spend any time.
9:01
SPEAKER_02: Their default would be to go and use AI.
9:04
SPEAKER_01: So they trusted it to be accurate and they believe it even if their own
9:08
SPEAKER_01: judgment is, yeah, that definitely is happening.
9:12
SPEAKER_01: I've heard some interesting comments about that.
9:14
SPEAKER_01: That's a good example.
9:15
SPEAKER_01: So thank you for sharing that.
9:16
SPEAKER_00: Yeah.
9:17
SPEAKER_01: You mentioned the study that you did with Google.
9:19
SPEAKER_01: I'd love to hear more about that.
9:22
SPEAKER_02: Yeah, of course.
9:23
SPEAKER_02: It's a two year study that we started with Google.
9:26
SPEAKER_02: Basically, two years ago, they did a big study with BCG.
9:30
SPEAKER_02: Looking at why people are not adopting AI.
9:33
SPEAKER_02: So they did a quantitative study with BCG and they looked at companies adopting AI
9:38
SPEAKER_02: and they saw less than 1%
9:39
SPEAKER_02: of companies actually achieving any transformational impact using AI.
9:43
SPEAKER_02: And they started digging into, okay,
9:44
SPEAKER_02: so we have the technology,
9:45
SPEAKER_02: it's been developing quite fast, why are companies not adopting it?
9:48
SPEAKER_02: What is stopping them to do that?
9:50
SPEAKER_02: So they wanted to run a bit more
9:51
SPEAKER_02: of something that's not very typical Google I guess it wants to look at
9:55
SPEAKER_02: the psychology aspect and people side of the issue and what is stopping them.
9:59
SPEAKER_02: So we did a qualitative study where we interviewed more than 50 people asking them
10:04
SPEAKER_02: about the reasons of using and not using AI and we go to the point of Mindset,
10:11
SPEAKER_02: and I know sometimes these two concepts are kind of conflating,
10:15
SPEAKER_02: talking about mindsets and talking about AI,
10:18
SPEAKER_02: but I think AI as technology,
10:20
SPEAKER_02: it's bringing existential threats to who you are,
10:23
SPEAKER_02: and especially if you are as a person describing yourself through your profession,
10:28
SPEAKER_02: through what you do day in and out.
10:30
SPEAKER_02: When AI is automated, a lot of you start thinking, what am I doing?
10:34
SPEAKER_02: What am I going to do going forward?
10:36
SPEAKER_02: And for a lot of people, this leap is very difficult to do, especially if you are a specialist.
10:41
SPEAKER_02: We've seen this difference between generalists and specialists,
10:44
SPEAKER_02: where generalists Like myself,
10:46
SPEAKER_02: for example,
10:47
SPEAKER_02: like other people,
10:48
SPEAKER_02: which have a bit more broad specialism,
10:50
SPEAKER_02: they're a bit more comfortable,
10:51
SPEAKER_02: but people that have been honing their craft for 30 years and they have,
10:55
SPEAKER_02: they're carrying the pride of doing research and they want to go into depth
11:00
SPEAKER_02: of some topics, they find it very difficult to do the switch.
11:04
SPEAKER_02: We came up with three different mindsets of how people are implementing AI.
11:08
SPEAKER_02: We've abbreviated them so they're kind of sexy and easy to remember.
11:12
SPEAKER_02: So we have a FOMO, which is Fear of Missing Out.
11:15
SPEAKER_02: It's quite famous.
11:17
SPEAKER_02: But in this context,
11:18
SPEAKER_02: it's all about looking at what your competition is doing and trying to catch up
11:23
SPEAKER_02: and trying to show to investors or to your consumers or your customers
11:27
SPEAKER_02: that you're using AI because using AI is something sexy.
11:31
SPEAKER_02: And then you have FOMU, which is Fear of Messing Up.
11:34
SPEAKER_02: And this is usually legacy companies like Pharma,
11:37
SPEAKER_02: law companies,
11:37
SPEAKER_02: which are holding a lot of consumer data and they're not allowed
11:40
SPEAKER_02: to make any mistakes in the public domain.
11:43
SPEAKER_02: They feel like they would be persecuted.
11:45
SPEAKER_02: We also see this with a lot of creative companies that have been traditionally seen
11:50
SPEAKER_02: as the lighthouse of creativity and suddenly they're starting using AI
11:54
SPEAKER_02: in their marketing production and they're like, but why are you using AI?
11:58
SPEAKER_02: We saw a lot of backlash with Coca-Cola,
11:59
SPEAKER_02: for example,
12:00
SPEAKER_02: in their Christmas campaign where they started using AI and it's been like this,
12:04
SPEAKER_02: you know,
12:04
SPEAKER_02: everybody thinks of Coca-Cola and Christmas together and they got such
12:07
SPEAKER_02: a big backlash.
12:09
SPEAKER_02: So a lot of companies are really struggling with that.
12:11
SPEAKER_02: And then we see a very small fraction of companies which are focusing on maximizing advantages.
12:16
SPEAKER_02: So that's FOMA.
12:18
SPEAKER_02: And these are usually companies which are very structured thinking.
12:22
SPEAKER_02: They have head of engineering, they have head of data, they have head of AI.
12:26
SPEAKER_02: People who are really thinking not only about implementing the technology as it is,
12:31
SPEAKER_02: they're not trading as a tech platform,
12:33
SPEAKER_02: it's more trading it as something that people need to adopt and they need
12:36
SPEAKER_02: the proper training to do so.
12:38
SPEAKER_02: So that was last year when we did this study and then we thought,
12:41
SPEAKER_02: okay, so we know what makes people successful.
12:44
SPEAKER_02: How can we learn from the successful companies and people?
12:47
SPEAKER_02: How can we help others?
12:50
SPEAKER_02: So this year was more around the behavioral science of the challenge.
12:54
SPEAKER_02: This is where we run behavioral labs and behavioral science experiments
12:58
SPEAKER_02: where we're observing in real time how people behave
13:00
SPEAKER_02: when you were giving them different tools.
13:03
SPEAKER_02: So this is where I started talking about this cognitive surrender concept
13:06
SPEAKER_02: where we had to give people two tasks.
13:08
SPEAKER_02: The first one was you need to create the campaign using AI,
13:12
SPEAKER_02: but you're not allowed to question the output.
13:15
SPEAKER_02: You can spend as much as possible on the prompt.
13:19
SPEAKER_02: But on the actual output, you can just take it as it is.
13:22
SPEAKER_02: And the second one was the ownership one,
13:24
SPEAKER_02: which is basically we were giving them different frameworks.
13:28
SPEAKER_02: They can use any resources online.
13:29
SPEAKER_02: They could talk to other people in the room and they can share experiences.
13:32
SPEAKER_02: So they can spend as much as possible perfecting their craft and perfecting
13:36
SPEAKER_02: the campaign they were working on.
13:38
SPEAKER_02: And we're basically,
13:39
SPEAKER_02: throughout the experiment,
13:40
SPEAKER_02: we were comparing how they feel and their sense of pride and ownership.
13:44
SPEAKER_02: And we saw something quite fundamental.
13:46
SPEAKER_02: It's called the IKEA effect in behavioral science.
13:49
SPEAKER_02: So people actually take pride when they're built and they don't take as much pride
13:54
SPEAKER_02: in something that's been given to them quite easily,
13:56
SPEAKER_02: which is not something that I personally expected because I thought
13:59
SPEAKER_02: that people want things to be easy,
14:01
SPEAKER_02: but over time,
14:03
SPEAKER_02: It becomes too easy and we see people feeling disengaged,
14:06
SPEAKER_02: they don't feel any pride,
14:07
SPEAKER_02: they're kind of twimbling their thumbs and thinking about what they're going
14:12
SPEAKER_02: to be doing next.
14:13
SPEAKER_02: So yeah,
14:13
SPEAKER_02: we did the experiment and we came up with a playbook which is going to be published
14:19
SPEAKER_02: in the next month with some behavioural science nudges,
14:23
SPEAKER_02: tips and tricks of how you can nudge people in the right behaviours.
14:26
SPEAKER_02: Of course we're not How do I say?
14:29
SPEAKER_02: We're not overly confident.
14:30
SPEAKER_02: We don't want to say that this playbook is going to fix all of your issues and all
14:33
SPEAKER_02: of your challenges within your organization.
14:35
SPEAKER_02: There are so many things that you need to think about.
14:38
SPEAKER_02: Obviously, retraining is a big challenge.
14:41
SPEAKER_02: How do you create career paths now that everything has been changing so fast?
14:44
SPEAKER_02: And especially with these specialist routes that I've been discussing.
14:48
SPEAKER_02: So there are a lot of challenges,
14:49
SPEAKER_02: but there's like some small things that you can start thinking about
14:52
SPEAKER_02: an organization level.
14:54
SPEAKER_01: One of our writers,
14:56
SPEAKER_01: one of the chapters in our next She Writes AI book was writing about this
15:00
SPEAKER_01: the sense of identity and how people's sense of identity in
15:03
SPEAKER_01: an organization is being changed by this.
15:08
SPEAKER_01: If I am not this, what am I and who am I?
15:10
SPEAKER_01: And it's not just something that is a personal level.
15:13
SPEAKER_01: It actually affects the way an organization works.
15:16
SPEAKER_01: That's what her chapter is about.
15:17
SPEAKER_01: And I thought it was really cool that you mentioned that.
15:19
SPEAKER_01: We also have a chapter on drug discovery.
15:24
SPEAKER_01: So you've touched on two of them already, which is really cool.
15:27
SPEAKER_01: But it sounds like you get to deal with quite a lot of different domains,
15:30
SPEAKER_01: different areas in your work.
15:32
SPEAKER_01: So this playbook that's coming out, how is that being published?
15:36
SPEAKER_02: It's going to be published by Google.
15:38
SPEAKER_02: So it's going to be a downloadable artifact on Think with Google,
15:42
SPEAKER_02: but I cannot give you timelines.
15:44
SPEAKER_02: It's in the review.
15:47
SPEAKER_01: By the time this interview comes out,
15:49
SPEAKER_01: it'll probably be out and we can put the link
15:51
SPEAKER_01: into your interview is what I'm thinking.
15:53
SPEAKER_01: So that would be very cool to have that.
15:54
SPEAKER_01: What is it that you're hoping will happen as a result of that playbook being published?
15:58
SPEAKER_01: Who you were hoping will read it and what do you think they'll do differently?
16:01
SPEAKER_02: As a person, I'm bordering between optimistic and realistic.
16:09
SPEAKER_02: So I am optimistic that people will read it and they'll take whatever they need to take from it.
16:14
SPEAKER_02: But I'm realistic that a lot of these changes are very difficult,
16:17
SPEAKER_02: especially if you think about a hundred-year-old organization that has
16:20
SPEAKER_02: so many structural challenges and old systems that need to be changed.
16:24
SPEAKER_02: And I'm not talking about technology,
16:25
SPEAKER_02: I'm talking about ways of working, what is being promoted, what is the hierarchy.
16:30
SPEAKER_02: And even we started talking about M-shape individuals within our research,
16:35
SPEAKER_02: and it's a concept that we took from a person from academia that is now working
16:39
SPEAKER_02: within WPP.
16:40
SPEAKER_02: So she coined this term,
16:41
SPEAKER_02: which is like in the past, we used to have I-shaped, then it was T-shaped.
16:45
SPEAKER_02: Now it's M-shaped because AI is kind of allowing you to have multiple specialties.
16:50
SPEAKER_02: So you're no longer just researcher.
16:51
SPEAKER_02: You can also do copywriting, or you can also do strategy to some degree.
16:56
SPEAKER_02: You're not never going to be specialists in these domains, but you can develop these over time.
17:01
SPEAKER_02: So we were saying,
17:02
SPEAKER_02: what is the future of these M-shaped specialists and how can you enable them to grow
17:07
SPEAKER_02: and cultivate them through the organization.
17:09
SPEAKER_02: So back to your question,
17:11
SPEAKER_02: I hope that the playbook is going to make leaders think
17:14
SPEAKER_02: about how they're approaching this challenge from a human perspective,
17:18
SPEAKER_02: because I think that the technology has been proving that it's developing so fast
17:21
SPEAKER_02: that it's not a technological challenge as much as it's a human challenge.
17:26
SPEAKER_02: And I think they need to start thinking of bringing people along not just
17:30
SPEAKER_02: through training but also explaining the why and why we're doing this
17:34
SPEAKER_02: and what is this going to mean for you, for your job, for your future.
17:38
SPEAKER_02: What I've learned from the study with Google is that the human factor is so strong
17:42
SPEAKER_02: that we cannot overlook it.
17:44
SPEAKER_02: We need to include people.
17:47
SPEAKER_02: I just hope that people are going to do a bit of thinking
17:49
SPEAKER_02: and what we are doing internally as well is Rethink what the future
17:53
SPEAKER_02: of different departments is going to look like.
17:55
SPEAKER_02: Because for strategy,
17:56
SPEAKER_02: for example,
17:57
SPEAKER_02: and management consulting in general,
17:59
SPEAKER_02: historically information has been what companies are paying for,
18:03
SPEAKER_02: but now information economy has been democratized,
18:07
SPEAKER_02: so you no longer need to pay for that.
18:08
SPEAKER_02: So what does that mean for strategy and management consulting in general?
18:10
SPEAKER_02: I think rethinking needs to happen across different industries.
18:15
SPEAKER_01: Your description of FOMO and FOMU and FOMA was interesting.
18:20
SPEAKER_01: One thing that I was kind of listening for when you were talking about these are
18:23
SPEAKER_01: the reasons why people aren't adopting AI.
18:26
SPEAKER_01: Maybe it's because of the kind of work that I do,
18:28
SPEAKER_01: but I hear a lot from people that there are ethical concerns,
18:32
SPEAKER_01: that there are environmental concerns,
18:33
SPEAKER_01: and there are many reasons why people are hesitant to use it.
18:37
SPEAKER_01: And it's not so much that they're fear of messing up,
18:39
SPEAKER_01: I don't think, but that they have these reservations.
18:41
SPEAKER_01: But it doesn't sound like that really came out as a strong factor in your study.
18:45
SPEAKER_02: They do have these reservations, but we did study actually something that I forgot to mention.
18:49
SPEAKER_02: We did it across UK, France, and Germany, and we saw some differences across countries.
18:54
SPEAKER_02: We saw that French people,
18:55
SPEAKER_02: for example,
18:56
SPEAKER_02: they care a lot about ethics and they will kept on mentioning what would that mean
19:01
SPEAKER_02: for,
19:01
SPEAKER_02: as you said,
19:02
SPEAKER_02: sustainability,
19:03
SPEAKER_02: what would that mean for data centers which require a lot of water and obviously
19:06
SPEAKER_02: the ethics of using people's data without disclosing exactly how
19:10
SPEAKER_02: that data has been used.
19:11
SPEAKER_02: So I think that it did come out but in different levels and degrees
19:15
SPEAKER_02: across different nationalities and I feel like some countries care more about it
19:19
SPEAKER_01: than others.
19:21
SPEAKER_01: That's very interesting to hear about.
19:23
SPEAKER_01: So one concern that's kind of related to this is I hear a lot
19:28
SPEAKER_01: of people expressing concern about where these systems get the data and the content
19:33
SPEAKER_01: that they use for training the systems.
19:35
SPEAKER_01: There are writers,
19:37
SPEAKER_01: artists,
19:38
SPEAKER_01: all kinds of people,
19:39
SPEAKER_01: musicians who point out that their work was taken without their permission,
19:44
SPEAKER_01: without their consent, and then used.
19:46
SPEAKER_01: And now it's been trained into a model and the model is being used basically
19:49
SPEAKER_01: to compete against them and taking away their work and their right to earn a living
19:53
SPEAKER_01: from their skills and their body of work.
19:56
SPEAKER_01: And so I hear a lot of concerns about where this data came from and whether
19:59
SPEAKER_01: the companies got consent or are giving credit or are compensating the people
20:05
SPEAKER_01: whose works they're using to make these insane profits and valuations.
20:11
SPEAKER_01: So I'm wondering what your thoughts are about this,
20:13
SPEAKER_01: what some people call the three C's, which is consent, credit and compensation.
20:17
SPEAKER_01: That's from CIPRI.
20:19
SPEAKER_01: Do you feel like that is something that these companies should be required to do?
20:23
SPEAKER_01: Do you feel like the ship has sailed?
20:25
SPEAKER_01: Do you feel that we should be looking at this in a different way?
20:27
SPEAKER_02: Yeah.
20:30
SPEAKER_02: I think that we need to compensate people for their work and
20:32
SPEAKER_02: for what they're bringing to the table.
20:34
SPEAKER_02: As you're mentioning artists, they cannot make a living from the work that they are producing.
20:39
SPEAKER_02: As a non-technical person, as I was mentioning, I'm not sure how that is possible at the moment.
20:43
SPEAKER_02: So we all know that The models right now have been trained on data at large.
20:48
SPEAKER_02: So Reddit, anything that's been published online, blog posts.
20:51
SPEAKER_02: So I do think that we need to do it technically.
20:55
SPEAKER_02: I don't know if that's even possible to backtrack and see what data has been used.
20:59
SPEAKER_02: Can we actually separate it?
21:01
SPEAKER_02: And can we give credit or royalties to people's data?
21:05
SPEAKER_02: I do believe yes, but I don't know if that's possible.
21:08
SPEAKER_02: Or should we start from scratch?
21:09
SPEAKER_02: I think that's a bigger conversation.
21:11
SPEAKER_02: In the EU, there was a big discussion around it.
21:15
SPEAKER_02: And now we have an EU AI Act,
21:17
SPEAKER_02: which is protecting people and protecting people's data and specifically artists
21:22
SPEAKER_02: and songwriters as well.
21:24
SPEAKER_02: But obviously legislation is very slow to react to a lot
21:28
SPEAKER_02: of this technological development, as we know.
21:30
SPEAKER_02: So maybe that ship has sailed.
21:32
SPEAKER_02: You can tell me your professional opinion.
21:34
SPEAKER_02: That's even possible to backtrack.
21:37
SPEAKER_01: Yeah, I've heard a lot of different solutions from some people.
21:39
SPEAKER_01: Some say that there are actually technical ways where it can be done,
21:43
SPEAKER_01: where it can be backed up, where you can trace.
21:45
SPEAKER_01: Thinking of traceability,
21:47
SPEAKER_01: you can tell there's over a hundred lawsuits that are active now and a lot
21:50
SPEAKER_01: of them have to do with music.
21:52
SPEAKER_01: And there are cases where they could very clearly trace a whole song and
21:56
SPEAKER_01: the song gets reproduced and you could say,
21:59
SPEAKER_01: yep, this came from that artist, that recording, that.
22:02
SPEAKER_01: So in some cases there are ways,
22:03
SPEAKER_01: and I know there are people working on finding ways to do this kind of traceability
22:09
SPEAKER_01: after the fact.
22:10
SPEAKER_01: Other people say the real way to do it is to just look at the output and not try
22:13
SPEAKER_01: to trace it,
22:14
SPEAKER_01: but just look for the different pieces that figure out where they came from and give
22:18
SPEAKER_01: the credit.
22:18
SPEAKER_01: Or just say, look,
22:20
SPEAKER_01: we took from this whole artist community,
22:22
SPEAKER_01: we need to give something back to the artist community.
22:26
SPEAKER_01: And how to do that and do it fairly is another question.
22:29
SPEAKER_01: But I've heard a lot of proposals and I think the main problem is
22:34
SPEAKER_01: that there isn't any political or commercial force that's compelling it to happen.
22:40
SPEAKER_01: And in the absence of that, it's not going to.
22:43
SPEAKER_01: Companies do what they are rewarded for doing.
22:46
SPEAKER_01: And right now there's no strong enough incentive.
22:49
SPEAKER_01: I think some of the actions that we're seeing come out of the EU
22:54
SPEAKER_01: and others are leading the way on this in many regards.
22:57
SPEAKER_01: But whether it's enough,
22:59
SPEAKER_01: 3% of billion dollars of revenue sounds like a lot of money,
23:02
SPEAKER_01: but they still have many billions left over and they're still profiting greatly,
23:07
SPEAKER_01: whereas the artists are,
23:09
SPEAKER_01: even in cases where there have been settlements,
23:10
SPEAKER_01: like for books,
23:11
SPEAKER_01: some of the writers that I've heard from said they got maybe $500, $1,000.
23:18
SPEAKER_01: it's not enough to make up for what they've lost and so there's a lot
23:23
SPEAKER_01: of concern around that so yeah it's a tough problem but you know the fact
23:28
SPEAKER_01: that we're just kind of saying oh well it's a tough problem and not dealing
23:31
SPEAKER_01: with it is just letting it get worse.
23:34
SPEAKER_02: Yeah, exactly.
23:34
SPEAKER_02: I absolutely agree.
23:35
SPEAKER_02: And this is why I think it's so important to discuss it here,
23:39
SPEAKER_02: but also as part of what Unmute is doing.
23:42
SPEAKER_02: Something that I didn't mention at the beginning is that I'm also program lead
23:45
SPEAKER_02: for Unmute, which is a community of women in AI.
23:50
SPEAKER_02: So we start with women.
23:51
SPEAKER_02: We also like to amplify different diverse voices in AI to talk exactly about those challenges.
23:58
SPEAKER_02: We are at a time where all these systems are being developed.
24:03
SPEAKER_02: And if we don't interject now,
24:06
SPEAKER_02: we're looking at the next 50,
24:08
SPEAKER_02: 100 years of models that have been trained on specific data carrying biases from
24:13
SPEAKER_02: a very small percentage of the population.
24:15
SPEAKER_02: And if we don't have those conversations now,
24:17
SPEAKER_02: we're looking at the very automated and biased future, I think.
24:23
SPEAKER_02: I don't want to sound pessimistic,
24:25
SPEAKER_02: but I do think that these conversations are really important
24:28
SPEAKER_02: and this is what we're also trying to do, so publish as much as possible.
24:31
SPEAKER_02: And back to the creator economy,
24:33
SPEAKER_02: we recently published an article about influencer economy
24:37
SPEAKER_02: and how AI is impacting them.
24:40
SPEAKER_02: Influencers,
24:40
SPEAKER_02: for a lot of women,
24:42
SPEAKER_02: actually has been a way to get extra income,
24:45
SPEAKER_02: whether that's their mothers or they're in a different situation with their partner.
24:49
SPEAKER_02: Through influencing, they've been able to make some additional income.
24:53
SPEAKER_02: Now we have a lot of AI influencers which are taking over and they're almost non-recognizable.
24:59
SPEAKER_02: People can't tell the difference between an influencer and an AI chatbot.
25:03
SPEAKER_02: So it's kind of like shifting this dynamic and women are yet again in
25:06
SPEAKER_02: the receiving end of technology and being pushed away from something
25:11
SPEAKER_02: that they managed to carve out for themselves.
25:13
SPEAKER_02: So it's changing a lot of what's been built,
25:17
SPEAKER_02: I think, in the last 10 years, what we
25:19
SPEAKER_01: are seeing on social media as well.
25:21
SPEAKER_01: Yeah I want to come back to unmute and what you mentioned about bias
25:23
SPEAKER_01: because I think that's an important part that isn't always recognized
25:26
SPEAKER_01: or isn't given enough attention you know these systems reproduce
25:30
SPEAKER_01: and reinforce patterns and people women are noticing different areas where
25:35
SPEAKER_01: the bias comes up you ask for a photo of a CEO and it's a white males usually in
25:42
SPEAKER_01: a suit and tie in an office.
25:44
SPEAKER_01: A lot of CEOs are not that.
25:46
SPEAKER_01: And a lot of them aren't men.
25:48
SPEAKER_01: A lot of them aren't white, but that's the default it gravitates towards and will reinforce.
25:54
SPEAKER_01: And we say biases in hiring.
25:56
SPEAKER_01: And so there's so many ways that it's coming out
26:00
SPEAKER_01: and that's not being adequately discussed or addressed either.
26:04
SPEAKER_01: So I want to hear more about unmute though.
26:06
SPEAKER_01: I'm glad you mentioned it there because I didn't mean to come to it and talk about it.
26:11
SPEAKER_02: So it started with four co-founders on the 8th of March,
26:16
SPEAKER_02: actually, in the International Women's Day this year.
26:18
SPEAKER_02: It started with the mission to bring diverse voices and to amplify them.
26:22
SPEAKER_02: So we started with the Substack.
26:24
SPEAKER_02: We've been growing.
26:25
SPEAKER_02: We have around 400 subscribers.
26:27
SPEAKER_02: We've also launched our LinkedIn.
26:29
SPEAKER_02: We're reposting a lot of the articles,
26:31
SPEAKER_02: but also what we're trying to do is think about the pain points
26:35
SPEAKER_02: that our community have and how we can develop using AI different products
26:38
SPEAKER_02: that we can use to solve these challenges.
26:41
SPEAKER_02: So one of the products that we're working on is called Nuria.
26:45
SPEAKER_02: So this is an AI agent who is basically preparing parents to come back to work
26:51
SPEAKER_02: after their parental leave or they've been away for a while.
26:54
SPEAKER_02: So it's an agent that sits on top of Copilot and it's connected to your Outlook,
26:59
SPEAKER_02: it's connected to Teams and it's basically looking at data.
27:02
SPEAKER_02: From those channels and consolidating this data when you're back
27:04
SPEAKER_02: from your maternity leave you can ask questions about specific projects
27:08
SPEAKER_02: or you can ask about abbreviation or you can ask about updates
27:10
SPEAKER_02: so you feel like you're equipped to come back to work.
27:14
SPEAKER_02: What we see Some of the statistics are actually quite dire.
27:18
SPEAKER_02: We see a lot of women leaving their work,
27:20
SPEAKER_02: especially in the U.S.
27:21
SPEAKER_02: after having their first child because of lack of maternity leave policies.
27:25
SPEAKER_02: We saw some statistics recently about a month ago that about 800,000 jobs in
27:30
SPEAKER_02: the U.S.
27:31
SPEAKER_02: have been lost and a lot of these are women,
27:34
SPEAKER_02: 90% or if not 100% since the beginning of the year.
27:38
SPEAKER_02: It's pure assumption on our side,
27:39
SPEAKER_02: but probably a lot of these are mothers that cannot simply come back after a month
27:45
SPEAKER_02: or two months to work with a newborn.
27:47
SPEAKER_02: And even if they do and they have a bit more relaxed policies like we have
27:51
SPEAKER_02: in Europe,
27:51
SPEAKER_02: I think it's still really hard to manage your new identity,
27:55
SPEAKER_02: being a new mom with a quite demanding and challenging job.
27:58
SPEAKER_02: So we're trying to provide as many tools as possible.
28:01
SPEAKER_02: We've also launched something which is called AI Patent Tracker,
28:04
SPEAKER_02: which you can find on our website,
28:06
SPEAKER_02: which is basically tracking women's patents in AI versus men.
28:10
SPEAKER_02: And we are trying to encourage women to apply for patents,
28:14
SPEAKER_02: to do more research and to feel courageous enough to go and apply and think a lot
28:21
SPEAKER_02: of women are really struggling with imposter syndrome.
28:24
SPEAKER_02: And we keep on talking about this quite a lot,
28:26
SPEAKER_02: but we really want to support women in STEM and we really want to support just women
28:31
SPEAKER_02: in general, talk about AI.
28:35
SPEAKER_01: It's cool that you formed your group on March 8th, the International Women's Day.
28:39
SPEAKER_01: That's the day that we chose for launching our She Writes AI Community's first book,
28:44
SPEAKER_01: AI Everywhere Volume 1.
28:45
SPEAKER_01: We launched it on March 8th as well.
28:47
SPEAKER_00: Amazing.
28:49
SPEAKER_01: Yeah, we're aiming for Ada Lovelace today for the next set of books in October.
28:53
SPEAKER_01: Amazing.
28:55
SPEAKER_01: So how is Unmute organized?
28:56
SPEAKER_01: You said you started with,
28:57
SPEAKER_01: I think,
28:58
SPEAKER_01: four founders,
28:58
SPEAKER_01: but is this something that is open to anyone within your subset group?
29:03
SPEAKER_01: Who are you looking for, I guess, is a good question.
29:06
SPEAKER_02: Well, at the moment we're looking at women which are interested in AI that can talk
29:11
SPEAKER_02: about their experiences.
29:12
SPEAKER_02: We're always looking for contributors.
29:14
SPEAKER_02: You can write for us, you can reach out to us and offer your articles.
29:17
SPEAKER_02: So we're quite open to the types of topics that we are welcoming to publish.
29:22
SPEAKER_02: Quite similar to what you're doing,
29:23
SPEAKER_02: we just want to see different voices and experiences and people talking about AI,
29:27
SPEAKER_02: not just from an engineering perspective, but also creative perspective.
29:30
SPEAKER_02: And a lot of the people within our network are creative technologists or using AI
29:35
SPEAKER_02: for creativity, let's say.
29:37
SPEAKER_02: So we start with that.
29:38
SPEAKER_02: So you can reach out to us.
29:40
SPEAKER_02: You become part of the community.
29:41
SPEAKER_02: So we have a WhatsApp group that you'll be part of.
29:43
SPEAKER_02: We're also looking into having more community events,
29:47
SPEAKER_02: so just gathering together and being able to mingle with like-minded individuals.
29:51
SPEAKER_02: The majority of us are based in London,
29:53
SPEAKER_02: but also we're looking at events in New York as well because one of
29:55
SPEAKER_02: the co-founders is based there and looking into launching new products as well.
29:59
SPEAKER_02: So if any of your listeners has an amazing point of view on AI, please reach out to us.
30:06
SPEAKER_02: We'll be more than happy to collaborate and see how we can publish it.
30:10
SPEAKER_01: Yeah, absolutely.
30:11
SPEAKER_01: I don't know if I had mentioned,
30:12
SPEAKER_01: but on this interview series,
30:13
SPEAKER_01: I'm doing this interview series just as me personally,
30:16
SPEAKER_01: this is my personal mission,
30:19
SPEAKER_01: but about two thirds of my guests have been women and that's not really an accident
30:22
SPEAKER_01: because I feel like our voices need this extra amplification.
30:27
SPEAKER_01: So that's been a focus for them,
30:28
SPEAKER_01: but SheRace AI has spun off into its own identity and we've got,
30:32
SPEAKER_01: I think, 790 women listed in our directory now.
30:35
SPEAKER_01: They're not all active.
30:37
SPEAKER_01: But I would love to get you and some of your leaders and members able
30:42
SPEAKER_01: to participate there and be included in our weekly digest that we send out.
30:46
SPEAKER_01: And there are probably ways that our groups could help each other.
30:49
SPEAKER_02: And this is why I reached out to you because I felt
30:52
SPEAKER_02: that our goals and our vision is very similar.
30:54
SPEAKER_02: So we can collaborate and not create more silos.
30:59
SPEAKER_01: Yeah, it's quite interesting.
31:01
SPEAKER_01: One of the things that I've noticed of our 790 some newsletters,
31:05
SPEAKER_01: we have 16 different areas of AI that we group them into.
31:08
SPEAKER_01: Ethics is one of the biggest and these business applications are the biggest,
31:11
SPEAKER_01: but the one for parenting and family life, you might think that would be very big.
31:15
SPEAKER_01: It's actually one of our smallest, it seems like, who are writing aren't writing about that.
31:20
SPEAKER_01: They're writing about other things about AI.
31:23
SPEAKER_01: I really feel like that's an area that is underrepresented right now
31:27
SPEAKER_01: in our community is,
31:28
SPEAKER_01: hey, this is what's happening when I'm dealing with this, or this is helping me.
31:32
SPEAKER_01: One of the other chapters in our next book is a woman who used AI
31:36
SPEAKER_01: for her fertility journey and then her resulting pregnancy.
31:40
SPEAKER_01: And I feel like those are the kinds of stories that we need to share more of.
31:45
SPEAKER_01: So if anyone in your community is interested in sharing that,
31:48
SPEAKER_01: I think they would find that people are kind of looking for it,
31:51
SPEAKER_01: but we don't have that many people that are actually writing about it,
31:54
SPEAKER_01: which is counterintuitive.
31:57
SPEAKER_02: Yeah.
31:57
SPEAKER_02: Why do you think that is?
31:59
SPEAKER_02: Is it because people don't think that they can share?
32:01
SPEAKER_02: It's too personal?
32:04
SPEAKER_01: I don't know if it's that.
32:05
SPEAKER_01: It's a really good question.
32:07
SPEAKER_01: I'd be very curious to know the answer.
32:09
SPEAKER_01: I suspect it's more that when the women who are making the time to come write,
32:13
SPEAKER_01: they're doing it because they want to advance themselves professionally.
32:18
SPEAKER_01: They want to write about their work.
32:19
SPEAKER_01: They want to be taken seriously in the world of AI.
32:22
SPEAKER_01: And so they're writing about the applications and women,
32:26
SPEAKER_01: the representation and ethics, I think it makes sense.
32:28
SPEAKER_01: That's one of the areas where women actually aren't vastly outnumbered by men is in
32:35
SPEAKER_01: the areas of ethics and the impacts on society.
32:37
SPEAKER_01: And you and I both see that, I think.
32:39
SPEAKER_01: So I can understand why that one's big,
32:41
SPEAKER_01: but why the parenting and family life is one of the smallest.
32:44
SPEAKER_01: I would really love to know more about that.
32:48
SPEAKER_01: So I'm curious if maybe your community has some thoughts on it.
32:51
SPEAKER_02: Yeah, yeah, I'll definitely ask around.
32:53
SPEAKER_02: It's quite interesting.
32:54
SPEAKER_02: So I went to this talk.
32:55
SPEAKER_02: There's another publication which is called Godmothers of AI,
32:59
SPEAKER_02: which is also very similar to what our communities are doing.
33:03
SPEAKER_02: So they had a launch a month or two months ago and I went to the launch and they had
33:08
SPEAKER_02: a panel and one of the speakers on the panel was the founder
33:11
SPEAKER_02: of this company called IV.
33:14
SPEAKER_02: So it's I-V-E-E.
33:16
SPEAKER_02: And what they're doing is they're working with the UK government
33:19
SPEAKER_02: to retrain people starting with the UK government professionals,
33:23
SPEAKER_02: but also people within large organizations teaching them how to use AI tools.
33:28
SPEAKER_02: And one of the things that the co-founder of Ivy was saying is that she thinks
33:34
SPEAKER_02: that absolutely we need to discuss all these topics about motherhood
33:38
SPEAKER_02: and sustainability and ethics and these are so important topics but she also wants
33:42
SPEAKER_02: to see the creators and the people which are building and people
33:46
SPEAKER_02: which are training models and she feels as a founder,
33:50
SPEAKER_02: I don't want to misquote her,
33:51
SPEAKER_02: but she felt as a founder that women are always boxed in the sustainability corner
33:57
SPEAKER_02: of AI or ethics corner.
33:58
SPEAKER_02: And we see like you,
34:01
SPEAKER_02: for example,
34:01
SPEAKER_02: we see so many other brilliant women developing those tools
34:04
SPEAKER_02: and developing those systems.
34:06
SPEAKER_02: So how can we bring this even more to the forward of the conversation?
34:11
SPEAKER_02: So my personal goal for unmute is to have more and more people building with AI.
34:17
SPEAKER_02: Women training models coming out and talking about their challenges
34:20
SPEAKER_02: or what they're fascinating about or what they want to share with us.
34:24
SPEAKER_02: I'd like to see this as well.
34:26
SPEAKER_01: Yeah, that's very cool.
34:27
SPEAKER_01: Well, we're going to have to continue the conversation offline.
34:29
SPEAKER_01: I'm sure we could talk a lot longer
34:31
SPEAKER_00: about some of the things that we could
34:32
SPEAKER_01: do,
34:32
SPEAKER_01: but we
34:33
SPEAKER_00: certainly agree.
34:34
SPEAKER_01: And there's more than ample evidence that the biases are there in the tools,
34:39
SPEAKER_01: partly because women aren't on the teams,
34:42
SPEAKER_01: aren't in the rooms or people of color are in the rooms and the systems are built
34:46
SPEAKER_01: by people who just don't even think about some of the concerns or some of
34:50
SPEAKER_01: the considerations and that's something that we all could maybe help to remedy.
34:56
SPEAKER_01: I'm not sure if you're calling them returnships or what getting women back in there
35:00
SPEAKER_01: and helping them transition back is a really great initiative so I'm really happy
35:04
SPEAKER_01: that you're doing that.
35:06
SPEAKER_02: Thank you.
35:06
SPEAKER_02: It's very early on, thanks.
35:07
SPEAKER_02: So we're testing it internally, but I'm really hopeful that we can get more products.
35:12
SPEAKER_02: And now we do have developers within the company,
35:15
SPEAKER_02: but also AI is allowing us to do a lot of the groundwork ourselves to solve some
35:19
SPEAKER_02: of these challenges.
35:20
SPEAKER_02: So I'm super excited to hear more from the community and from your community as well.
35:25
SPEAKER_01: Yeah.
35:25
SPEAKER_01: So when you think about the AI tools that you use day to day,
35:28
SPEAKER_01: do you feel like the providers are transparent about where the data came from,
35:33
SPEAKER_01: whether it was equally trained on men or women.
35:36
SPEAKER_01: I think we know the answer to that.
35:39
SPEAKER_01: Do you get any sense that they are aware of it or that they're being upfront about
35:44
SPEAKER_01: where the data comes from and whether it's balanced,
35:47
SPEAKER_01: whether it's biased, anything along those lines?
35:51
SPEAKER_02: I think they're transparent that it's public source.
35:54
SPEAKER_02: I think they're transparent as much as they can be.
35:56
SPEAKER_02: I've listened to an interview with Miro Murati.
36:00
SPEAKER_02: Who actually I really respect and I think it's great as we have
36:03
SPEAKER_02: a strong female leader within the space but she herself didn't know the answer or
36:08
SPEAKER_02: at least she didn't want to share the answer of what data has been scraped and how
36:13
SPEAKER_02: the model has been trained which can tell you that was
36:15
SPEAKER_02: about two years ago I
36:16
SPEAKER_02: think yes I remember that interview yeah
36:19
SPEAKER_02: so I think they're providing you some sort of transparency
36:22
SPEAKER_02: but actually I don't think they even know or some
36:26
SPEAKER_02: of these sources are just too mixed up with everything they've been scraping
36:30
SPEAKER_01: So it's people who are consumers or existing in public.
36:34
SPEAKER_01: We use public transportation or just any kind of systems around the street.
36:39
SPEAKER_01: Our personal data is getting used by these different AI tools or systems.
36:44
SPEAKER_01: Do you know of any cases you could share where your own data has been used that you're aware of?
36:50
SPEAKER_02: Personally,
36:51
SPEAKER_02: I'm not aware of a case where my data has been used,
36:54
SPEAKER_02: but I know there was this big case with Clearview.
36:57
SPEAKER_02: At least the UK was a big scandal when there was a US company,
37:00
SPEAKER_02: which was basically scraping information from Facebook and using it for profiling.
37:05
SPEAKER_02: And that was used for defense purposes and law enforcement,
37:08
SPEAKER_02: but also banks have been using it as well.
37:11
SPEAKER_02: And so now I know this was a big scandal.
37:13
SPEAKER_02: I don't know if they've used my photos specifically.
37:15
SPEAKER_02: I feel like you'd get an email if there has been a data breach
37:18
SPEAKER_02: and somebody stole your email from a database,
37:21
SPEAKER_02: but you'll never get an email if somebody stole your photo or scraped your photo
37:25
SPEAKER_02: of Facebook.
37:27
SPEAKER_02: Maybe it has been, I just am not aware.
37:32
SPEAKER_01: Does awareness change the way that you use different social media systems
37:37
SPEAKER_01: or do you kind of feel like it's already out there, it's too late?
37:41
SPEAKER_02: I'm definitely more conscious.
37:43
SPEAKER_02: I am using only Instagram out of all the social medias and then they have a toggle
37:47
SPEAKER_02: where you can say that you're not allowing any of your posts to be used
37:51
SPEAKER_02: to train AI models.
37:52
SPEAKER_02: But at least the companies are reassuring you that they're not using your data.
37:58
SPEAKER_01: I have a friend who lives on the West Coast here.
38:01
SPEAKER_01: She just went through Heathrow Airport and she was with her partner who's from the UK.
38:07
SPEAKER_01: And she commented that she didn't even have to take out her passport
38:09
SPEAKER_01: because they scanned her face and automatically recognized her.
38:13
SPEAKER_01: And she said, you know, I think I would have rather waited in line.
38:18
SPEAKER_01: And had to go through it than to know that I could just walk through
38:20
SPEAKER_01: and they've got my face and they tagged who I was.
38:23
SPEAKER_01: I think there are a lot of cases like that where she became aware
38:27
SPEAKER_01: that they had her face already when she went through security at Heathrow.
38:33
SPEAKER_02: That's so interesting because we do have those electronic gates
38:36
SPEAKER_02: when you scan your passports and basically you have biometrics within your passport
38:40
SPEAKER_02: and they're scanning your face to match them.
38:43
SPEAKER_02: But yeah, I didn't know that even walking around they have this data about you.
38:48
SPEAKER_01: She's an American citizen with a U.S.
38:50
SPEAKER_01: passport,
38:50
SPEAKER_01: so I don't know if the biometrics would have been in there,
38:52
SPEAKER_01: so they got her face
38:53
SPEAKER_00: some other way.
38:54
SPEAKER_01: But there are still so many situations.
38:56
SPEAKER_01: And of course,
38:57
SPEAKER_01: as you mentioned with the data brokers,
38:58
SPEAKER_01: them stealing our data, this is not new with AI.
39:01
SPEAKER_01: It's just that AI makes it worse in some ways.
39:04
SPEAKER_01: Once that data goes somewhere,
39:05
SPEAKER_01: the EU has this right to be forgotten where they should be taking your data back out
39:10
SPEAKER_01: if you object.
39:12
SPEAKER_01: Don't know if any of that is happening because once it goes into those systems,
39:16
SPEAKER_01: it's very difficult to get it back out.
39:19
SPEAKER_02: It's very difficult to enforce and
39:23
SPEAKER_01: persecute, I think.
39:25
SPEAKER_01: So with LinkedIn, there are opt-outs for telling them not to use your data.
39:28
SPEAKER_01: Have you opted out?
39:29
SPEAKER_01: Yeah.
39:30
SPEAKER_01: Okay.
39:31
SPEAKER_01: And how about on Substack?
39:32
SPEAKER_01: Are you aware that there's a setting in your newsletter that allows you to opt out?
39:36
SPEAKER_01: Substack will try to block third-party AI bots from copying your newsletter.
39:41
SPEAKER_01: Do you know if you've opted out of that one?
39:43
SPEAKER_01: Yeah.
39:43
SPEAKER_01: Some people actually haven't been aware of that option until I've asked them,
39:46
SPEAKER_01: like, oh, I didn't know we had that.
39:50
SPEAKER_02: Yeah.
39:50
SPEAKER_02: But it's also always really hidden.
39:52
SPEAKER_02: So it is available, but you need to look for it.
39:55
SPEAKER_01: Yes.
39:55
SPEAKER_01: Yeah.
39:56
SPEAKER_01: LinkedIn was especially not very forthcoming about that
39:59
SPEAKER_00: when they first introduced this change and they took more data,
40:02
SPEAKER_01: but that's a whole other thing.
40:04
SPEAKER_01: In the US, certainly we have data breaches regularly.
40:08
SPEAKER_01: And I think you mentioned that earlier.
40:10
SPEAKER_01: Have you ever had a data breach that you've been notified about
40:13
SPEAKER_01: where someone got your information?
40:15
SPEAKER_02: Yeah.
40:15
SPEAKER_02: I think it was an email and something that I don't know what I do,
40:21
SPEAKER_02: but I'd never give my actual date of birth.
40:24
SPEAKER_02: That's smart.
40:25
SPEAKER_02: I've always put my random date of birth,
40:27
SPEAKER_02: so they stole my date of birth, which was fake, and email, I think, only.
40:33
SPEAKER_02: So nothing else, thankfully.
40:35
SPEAKER_02: But yeah, I've had a few emails from retailers telling me that.
40:42
SPEAKER_01: Yeah, yeah.
40:43
SPEAKER_01: Overall, it seems like public distrust.
40:47
SPEAKER_01: Of these companies has been growing.
40:48
SPEAKER_01: We hear more about breaches.
40:49
SPEAKER_01: We hear more about them doing things with our data that we never consented to.
40:54
SPEAKER_01: I'm wondering if you think that there is some way that a company could persuade you
40:59
SPEAKER_01: to trust them, or do you think that's not even possible?
41:03
SPEAKER_01: Or if it could be done, what would it take?
41:05
SPEAKER_01: What would be the main thing they'd have to do to earn and then keep your trust?
41:10
SPEAKER_02: I think that's a very difficult question
41:12
SPEAKER_02: because historically we've been quite distrustful and things happening in
41:19
SPEAKER_02: the last months and years have,
41:21
SPEAKER_02: yeah, kind of time and time again reassured that big tech cannot be trusted.
41:26
SPEAKER_02: But I think just transparency on how models have been used and transparency,
41:31
SPEAKER_02: especially on big contracts.
41:33
SPEAKER_02: Probably know that a lot of companies got a lot of backlash into being involved
41:38
SPEAKER_02: with law enforcement, defense contracts.
41:41
SPEAKER_02: So I think personally for me,
41:43
SPEAKER_02: I'll feel a lot more comfortable if I know that
41:46
SPEAKER_01: My data
41:47
SPEAKER_02: or companies
41:48
SPEAKER_02: that I'm using are not being involved in any sort of war
41:54
SPEAKER_02: or any law enforcement
41:57
SPEAKER_02: or defending or anything like that.
41:59
SPEAKER_02: I do know it's a very probably lucrative industry to get involved,
42:02
SPEAKER_02: but I think people would value this transparency.
42:06
SPEAKER_02: And I think a big question that you've mentioned or a big topic
42:10
SPEAKER_02: that you mentioned earlier was sustainability and how we're planning to tackle this
42:15
SPEAKER_02: and I know a lot of data centers especially in the US have been built in areas
42:19
SPEAKER_02: where you don't have water we have water shortage so basically you're depriving
42:23
SPEAKER_02: the local population of their resources so how are we planning to tackle that
42:27
SPEAKER_02: and how can we make data centers more energy efficient I think more efforts need
42:33
SPEAKER_02: to be done in that regard and personally for me As much as I enjoy generating images
42:40
SPEAKER_02: and loved being creative and unleashing different parts of me
42:44
SPEAKER_02: that I don't necessarily see myself as a creative person,
42:47
SPEAKER_02: but now I started seeing myself a bit more.
42:49
SPEAKER_02: I would love to see more advancement in research and innovation and how we can use
42:55
SPEAKER_02: the models to bring goods to humanity and how we can advance science.
43:00
SPEAKER_02: So I would love to see more companies pledging innovation funds and part
43:06
SPEAKER_02: of their resources into actual science and innovation.
43:09
SPEAKER_02: I don't know if that's possible.
43:15
SPEAKER_01: So going back to the whole idea that AI was supposed to be good for everyone, right?
43:19
SPEAKER_01: For the good of the world, the good of humanity.
43:21
SPEAKER_01: Yeah.
43:22
SPEAKER_01: You mentioned data centers and we're starting to see a lot of pushback here
43:26
SPEAKER_01: where communities are protesting proposed data centers and getting some
43:29
SPEAKER_01: of them blocked in some cases,
43:30
SPEAKER_01: especially as You mentioned in areas where water resources are tight
43:34
SPEAKER_01: and it's competing with homes,
43:36
SPEAKER_01: it's competing with agriculture and other considerations, including power.
43:40
SPEAKER_01: Are you seeing any of that happening in England right now?
43:43
SPEAKER_01: Any resistance to data centers, or is it not really a factor in the London metro area?
43:48
SPEAKER_02: Yeah, not really, or at least it's not spoken about as much.
43:54
SPEAKER_02: I think people mention it here or there, but it's not as much of a challenge.
43:57
SPEAKER_02: Maybe because we do have water resources.
44:00
SPEAKER_02: Spain, on the other hand,
44:01
SPEAKER_02: is predicted to become a desert in the next hundred years with climate change and
44:06
SPEAKER_02: so many challenges with water shortages.
44:09
SPEAKER_02: So I think there will definitely be protests if they're happening at the moment.
44:14
SPEAKER_02: I'm not aware of them, but I'm not conscious of it.
44:18
SPEAKER_01: Yeah.
44:19
SPEAKER_01: It's interesting to see that there's some momentum gathering around the idea
44:22
SPEAKER_01: that they can't just put these into these communities that are underprivileged
44:28
SPEAKER_01: and impose on them and damage their quality of life for these massive data centers.
44:35
SPEAKER_01: And I personally would rather see them put their money and resources
44:38
SPEAKER_01: into finding more efficient algorithms so you don't need quite as big of
44:43
SPEAKER_01: a data center to do some of the things that need to be done.
44:46
SPEAKER_01: Obviously,
44:47
SPEAKER_01: if we want to look for drug discovery or doing things that make
44:50
SPEAKER_01: the world better using AI for good,
44:52
SPEAKER_01: we want to have some computing horsepower, but where it goes matters.
44:57
SPEAKER_02: Yeah, absolutely.
44:58
SPEAKER_02: And you previously asked me about the challenges that I was seeing with some of the models.
45:01
SPEAKER_02: I saw that some of the models are using a lot more tokens for smaller tasks than others.
45:06
SPEAKER_02: So that's also another challenge.
45:07
SPEAKER_02: How do you balance the usage, not only electricity, but tokenization?
45:11
SPEAKER_02: How do you charge for that?
45:12
SPEAKER_02: It just gets very complicated very easily, I think.
45:15
SPEAKER_01: Yes.
45:17
SPEAKER_01: Yeah, it does.
45:17
SPEAKER_01: Well, those are all my standard questions.
45:19
SPEAKER_01: Is there anything else that you would like to add for our audience?
45:22
SPEAKER_01: Any thoughts about AI?
45:23
SPEAKER_01: Anything that you were wishing I would ask you that I didn't ask you?
45:27
SPEAKER_02: No, I think it has been a great conversation and I find it so refreshing to see
45:33
SPEAKER_02: that I can find people that have similar passions to me around the world
45:39
SPEAKER_02: and I've spoken to you,
45:40
SPEAKER_02: I've spoken to another great founder a few weeks ago and we have very similar chat
45:45
SPEAKER_02: and it's just reassuring that we're all trying to do our bit and trying to change
45:52
SPEAKER_02: the world, we're trying to change our small ecosystems at least a bit.
45:57
SPEAKER_02: What I would love to see is more and more girls and more and more women going
46:03
SPEAKER_02: into STEM or at least being vocal about their challenges with AI
46:08
SPEAKER_02: and not being afraid of the tech bros who are going to try to shut them down.
46:14
SPEAKER_01: Yes.
46:16
SPEAKER_01: Awesome.
46:17
SPEAKER_01: I think we agree on that.
46:17
SPEAKER_01: Yeah.
46:18
SPEAKER_01: Great.
46:18
SPEAKER_01: And I'm so happy that we've connected.
46:20
SPEAKER_01: We can all do some things and they all add up,
46:23
SPEAKER_01: but maybe we can do some things together and have even more of an impact.
46:27
SPEAKER_01: And I'm really looking forward to talking more with you about that.
46:30
SPEAKER_02: Thank you.
46:31
SPEAKER_02: Thank you so much.
46:32
SPEAKER_01: Well, thank you, Ravi, for making time for the interview.
46:35
SPEAKER_01: Appreciate it.
46:36
SPEAKER_01: And we'll be talking.
46:38
SPEAKER_01: Thank you.
46:44
SPEAKER_00: This post is part of our AI6P Interview Series on AI Software and Wetware.
46:50
SPEAKER_00: It showcases how real people around the world are using their wetware,
46:55
SPEAKER_00: brains and human intelligence,
46:57
SPEAKER_00: with AI-based software tools who are being affected by AI.
47:02
SPEAKER_00: And we're all being affected by AI nowadays in our daily lives, perhaps more than we realize.
47:08
SPEAKER_00: For some examples, see post, but I don't use AI.
47:13
SPEAKER_00: We want to hear from a diverse pool of people worldwide in a variety of roles.
47:18
SPEAKER_00: If you're interested in being a featured interview guest,
47:21
SPEAKER_00: anonymous or with credit, please get in touch.
47:27
SPEAKER_00: 6P's in AI Pods AI6P is a 100% reader-supported publication.
47:34
SPEAKER_00: All new posts are free to read and listen to.
47:37
SPEAKER_00: To automatically receive new AI6P posts and support our work, Consider becoming a subscriber.
47:44
SPEAKER_00: It's free.