Latest / 6 Ps in AI Pods (AI6P) / 🗣️ AISW #126: Suzanne Mueller, Esq., Egypt-based executive leadership advisor

Transcript

🗣️ AISW #126: Suzanne Mueller, Esq., Egypt-based executive leadership advisor · 6 Ps in AI Pods (AI6P)

0:00
SPEAKER_01: Hi everyone, this is Karen Smiley.
0:02
SPEAKER_01: Welcome to 6P's in AI Pods.
0:07
SPEAKER_01: This post is part of our AI6P interview series on AI, software, and wetware.
0:13
SPEAKER_01: Our guests share their experiences with using AI and how they feel
0:17
SPEAKER_01: about AI using their data and content.
0:20
SPEAKER_02: I'm delighted to welcome Suzanne Muller Esquire from Egypt as my guest today on AI,
0:25
SPEAKER_02: software, and wetware.
0:27
SPEAKER_02: Suze, thank you so much for joining me on this interview.
0:29
SPEAKER_02: Please tell us about yourself, who you are, and what you do.
0:33
SPEAKER_00: Certainly, Karen.
0:33
SPEAKER_00: Thank you very much for inviting me this evening.
0:38
SPEAKER_00: A former chief legal officer for several airlines in the United States.
0:44
SPEAKER_00: In that capacity, I bought and sold airlines somewhere around four.
0:48
SPEAKER_00: I probably put down somewhere around 120 tails in my time,
0:54
SPEAKER_00: and I built the world's largest 767-200 fleet to turn it from passenger
0:59
SPEAKER_00: to cargo configuration from both ANA and Delta combined.
1:04
SPEAKER_00: And over the years,
1:06
SPEAKER_00: I've done a lot of basically heavy regulated work because aviation is of
1:10
SPEAKER_00: that nature.
1:12
SPEAKER_00: The problems are a mile wide, an inch deep, and they run frequent.
1:17
SPEAKER_00: Again, I also did a lot of deal-making at the table, a lot of mergers and acquisitions work.
1:23
SPEAKER_00: I was working with Gerson Lehman as one of their member advisors
1:27
SPEAKER_00: for approximately 11 years, and I worked with Fortune 500 companies.
1:34
SPEAKER_00: Large consulting firms and told them about whether I thought the deal was
1:38
SPEAKER_00: a good deal and where I thought the possible holes were and from
1:41
SPEAKER_00: a regulatory standpoint.
1:43
SPEAKER_00: So building on that background approximately 11 years ago,
1:47
SPEAKER_00: my husband and I made the decision to return to his home country in Germany.
1:52
SPEAKER_00: And in doing so, I could no longer practice law.
1:55
SPEAKER_00: So we famously as lawyers have learned to open consulting firms and through
1:59
SPEAKER_00: that venue,
2:01
SPEAKER_00: I've again worked with Fortune 500 companies,
2:05
SPEAKER_00: boards of directors,
2:06
SPEAKER_00: leadership,
2:07
SPEAKER_00: and given them strategies,
2:08
SPEAKER_00: particularly in heavily regulated areas regarding not just operations and business,
2:14
SPEAKER_00: but acquisitions, divestitures, and mergers.
2:18
SPEAKER_00: Today, I help companies around the world.
2:21
SPEAKER_00: Approximately two and a half years ago, I started to see what a big deal AI was beginning to be.
2:27
SPEAKER_00: And it struck me it was very much like aviation in that the decisions that had
2:32
SPEAKER_00: to be made were mission critical.
2:35
SPEAKER_00: There were many people that were affected by it that weren't part of the conversation.
2:41
SPEAKER_00: And so what I decided to do was start at the top because I believe that's
2:45
SPEAKER_00: where corporate responsibility lies in the fiduciary decision.
2:49
SPEAKER_00: I put together an AI governance framework I called BORA.
2:54
SPEAKER_00: Which is the AI Utility Risk Alignment System.
2:58
SPEAKER_00: And what that does is it takes OECD principles which embeds ideas of diversity
3:03
SPEAKER_00: and ideas of building economies,
3:05
SPEAKER_00: not just favoring big economies but making sure all economies participate.
3:12
SPEAKER_00: And I build a governance system around vision, around testing, around auditing, around training.
3:19
SPEAKER_00: Around making sure you take that auditing information and feed it back in
3:23
SPEAKER_00: so we can improve the deployment.
3:25
SPEAKER_00: And lastly,
3:26
SPEAKER_00: around decommissioning,
3:27
SPEAKER_00: because I believe it's important we understand when we start a deployment how to end
3:31
SPEAKER_00: a deployment.
3:33
SPEAKER_02: Very interesting transition from legal world in aviation to AI governance.
3:38
SPEAKER_02: I appreciate you explaining all of that.
3:40
SPEAKER_02: So tell us a little bit about your level of experience with AI and machine learning
3:46
SPEAKER_02: and analytics,
3:47
SPEAKER_02: whether you use it professionally or personally,
3:50
SPEAKER_02: or if you study the technologies, either formally or informally.
3:54
SPEAKER_00: A little of all of the above.
3:55
SPEAKER_00: Back in 1998, I was one of the first interns with GE Aircraft Engines, which is now GE Aviation.
4:03
SPEAKER_00: One of the first agreements I worked on
4:06
SPEAKER_00: in their M&A area was actually their engine health agreement,
4:09
SPEAKER_00: which is a telemetric agreement,
4:11
SPEAKER_00: the predecessor And my job was to make sure I could write an agreement about how
4:17
SPEAKER_00: the plane would beam down data with respect to the engine health to
4:22
SPEAKER_00: the EMP mechanics on the ground and they would act upon that information.
4:28
SPEAKER_00: And so I had to write an agreement the board would get comfortable with
4:32
SPEAKER_00: a machine saying,
4:33
SPEAKER_00: diagnostically, this is the issue and the EMP mechanics responding to it.
4:38
SPEAKER_00: So that was one of the first ways I got interested in machine learning.
4:42
SPEAKER_00: And the idea of telemetry,
4:44
SPEAKER_00: of beaming down information,
4:46
SPEAKER_00: the meeting of the mind between what the diagnostic is from the machine to what
4:51
SPEAKER_00: the human does in response.
4:53
SPEAKER_00: Over the years, I've continued IP work.
4:57
SPEAKER_00: I've continued telemetry work with respect to my agreements in aviation.
5:01
SPEAKER_00: It's a big part of it.
5:04
SPEAKER_00: And then about a year ago, I took a course Mindvalley, actually, which was a mastery course.
5:11
SPEAKER_00: And I spent six months doing nothing,
5:14
SPEAKER_00: but learning AI,
5:15
SPEAKER_00: building deployments,
5:17
SPEAKER_00: figuring out LMs,
5:18
SPEAKER_00: figuring out how to make them rag-friendly with the information that they wanted.
5:24
SPEAKER_00: Did a couple of hackathons because they were very fun.
5:27
SPEAKER_00: And so I got my hands into the deployment learning side of it,
5:31
SPEAKER_00: not so I could be a technician because I'm not.
5:35
SPEAKER_00: I'm someone who is on the governance side of things and on the M&A side of things,
5:40
SPEAKER_00: but I can see around corners when I understand what the developer is thinking.
5:45
SPEAKER_00: I'll never code for real, but I can VibeCode a meme program.
5:49
SPEAKER_00: So that was one of the ways that I took my actual learning to the test, and it was six months.
5:54
SPEAKER_00: I remember spending hours on a couch teaching myself Airtable after they went
5:59
SPEAKER_00: through it, and I thought I was going to lose my mind.
6:01
SPEAKER_00: It was fantastic.
6:03
SPEAKER_00: In my personal life, I began to design tools to help me with my work.
6:09
SPEAKER_00: And that is something I believe greatly in is how the decision is made by the human,
6:14
SPEAKER_00: because the human's in charge of this process.
6:17
SPEAKER_00: So one of the first things I built after listening to Kazi Koziakoff,
6:21
SPEAKER_00: who who was Google's first AI chief decision maker,
6:25
SPEAKER_00: she was one of my teachers in this program, was I built a decision tool.
6:31
SPEAKER_00: And it wasn't one that made the decision,
6:32
SPEAKER_00: but it was the one that allowed me to see it through various lens of fairness,
6:37
SPEAKER_00: or perhaps it's, are we saying speed over accuracy?
6:42
SPEAKER_00: Or are we favoring consistency over bias?
6:45
SPEAKER_00: And it would give me these various lenses so you could really peel back the nature
6:50
SPEAKER_00: of what you were favoring, what lens you were privileging in decision-making.
6:54
SPEAKER_00: So I do use tools like that.
6:56
SPEAKER_00: Buttercream is one.
6:57
SPEAKER_00: Sage is another.
6:58
SPEAKER_00: I build tools that help me idea things.
7:02
SPEAKER_00: But I, of course, am the person who's responsible and I take that position very seriously.
7:10
SPEAKER_02: Yeah, you mentioned setting up your applications and setting up systems to be rag-friendly.
7:16
SPEAKER_02: Obviously,
7:16
SPEAKER_02: in the legal business,
7:19
SPEAKER_02: having accurate sources,
7:20
SPEAKER_02: completely accurate,
7:21
SPEAKER_02: and only drawing from credible sources is exceptionally important.
7:25
SPEAKER_02: So it's interesting that you spent some time digging into that.
7:28
SPEAKER_02: Absolutely.
7:29
SPEAKER_00: I'm a big fan.
7:31
SPEAKER_00: Of checking and rechecking sources.
7:33
SPEAKER_00: Look, there are some decisions that are what I call low-hanging fruit.
7:36
SPEAKER_00: They're just not consequential.
7:38
SPEAKER_00: They're revocable.
7:39
SPEAKER_00: They're not a big deal.
7:40
SPEAKER_00: But there are many decisions that are consequential.
7:43
SPEAKER_00: And so knowing the difference in those decisions is the key to being successful with AI.
7:47
SPEAKER_02: Yeah.
7:49
SPEAKER_02: You mentioned that you've added a couple of tools for yourself.
7:52
SPEAKER_00: Yes, just several that have helped workflows.
7:56
SPEAKER_00: I don't use them all the time, day to day, but for instance, I love to argue and negotiate.
8:02
SPEAKER_00: I shouldn't say argue, I like to negotiate.
8:04
SPEAKER_00: It was one of my favorite things as a dealmaker that I've done over the years.
8:08
SPEAKER_00: So a sparring partner was something I put together so I could four
8:12
SPEAKER_00: or five levels down into a deal and say, well, I wouldn't come back with that.
8:16
SPEAKER_00: I want to get to push against what it does is it offers me clarity in The point
8:22
SPEAKER_00: with negotiations,
8:23
SPEAKER_00: that's the most important thing is things can go left, things can go right.
8:26
SPEAKER_00: Don't go in with a hard and fast position.
8:29
SPEAKER_00: Go in with puzzle pieces, drop them on the floor and be ready to move quickly.
8:33
SPEAKER_00: And so having an ideation partner that I could build keeps that in mind for me
8:37
SPEAKER_00: and makes me a better negotiator.
8:40
SPEAKER_02: It's very interesting that you enjoy negotiation.
8:43
SPEAKER_02: One thing that I hear a lot when I'm talking to women especially is
8:46
SPEAKER_02: that they didn't get comfortable with negotiation as they were growing up and
8:50
SPEAKER_02: that they find it more of a challenge.
8:52
SPEAKER_02: So I think it's really interesting that you've excelled in that area.
8:56
SPEAKER_00: I think part of that was I grew up in New York and we would go down to Chinatown
8:59
SPEAKER_00: and spend literally weekends haggling.
9:02
SPEAKER_00: It just became a thing.
9:03
SPEAKER_00: And when I got into business, honestly, I've been doing this 30 years.
9:07
SPEAKER_00: I walked into some spaces There were people sitting at the table.
9:11
SPEAKER_00: Many people have sat at the table and said, could you get me a coffee with cream?
9:14
SPEAKER_00: I would say, sure.
9:15
SPEAKER_00: And I'd turn around, go back, and I'd get them a coffee and make it up.
9:19
SPEAKER_00: I'd put it down and sit across from them and say,
9:21
SPEAKER_00: I don't know how good it is, but let's go ahead and let's go with that.
9:25
SPEAKER_00: And then I would be at the table across from them.
9:27
SPEAKER_00: But I understood the rules.
9:29
SPEAKER_00: I understood that if I was more prepared than they were,
9:32
SPEAKER_00: and I wasn't going to allow my emotions to get in the way,
9:35
SPEAKER_00: and I had strategy,
9:36
SPEAKER_00: but in a hard and fixed position,
9:39
SPEAKER_00: I was always going to come out with a fair deal,
9:42
SPEAKER_00: especially if I approached it with the problem is not between us,
9:46
SPEAKER_00: it's in the center.
9:47
SPEAKER_00: And I became very comfortable in the arena.
9:50
SPEAKER_02: That's a very interesting background and story.
9:53
SPEAKER_02: So thank you for sharing that.
9:55
SPEAKER_02: And by the way,
9:55
SPEAKER_02: I think your story about the GE engines and working with the health
9:59
SPEAKER_02: of them is super important.
10:00
SPEAKER_02: I spent a lot of time working in what ADD called asset health at the time,
10:05
SPEAKER_02: which is looking at the health of the big power transformers and other equipment,
10:09
SPEAKER_02: shore grids and factories.
10:10
SPEAKER_02: So that whole area.
10:12
SPEAKER_02: Legally, of course, it's very interesting.
10:13
SPEAKER_02: And then what people do with telemetry is very interesting.
10:17
SPEAKER_02: All the tools you're coming for our cars nowadays and what permissions people have around that.
10:22
SPEAKER_02: So it's a full area.
10:23
SPEAKER_02: I don't know if I'm trying to get into that.
10:25
SPEAKER_02: I think that's so interesting that you spent some time working in that.
10:28
SPEAKER_02: To get back to the AI specific features,
10:31
SPEAKER_02: I'm wondering if you can share a specific story on a time when you've used
10:35
SPEAKER_02: an AI tool.
10:35
SPEAKER_02: It could be a chatbot or it could be something else.
10:38
SPEAKER_02: AI features of different tools that you have used and how you use it.
10:43
SPEAKER_02: Maybe something that went well when you used it and something that didn't go so well.
10:48
SPEAKER_00: Well, I think there's lots of times it doesn't go well.
10:50
SPEAKER_00: I just start off on the negative side and we'll end on the happier note.
10:54
SPEAKER_00: There are certain problems that we recognize and limitations that AI has.
10:59
SPEAKER_00: It's a pattern seeker.
11:01
SPEAKER_00: So one of the biggest problems that AI has,
11:03
SPEAKER_00: and I think we're seeing this in business now,
11:05
SPEAKER_00: is that as a matter of strategy,
11:08
SPEAKER_00: when you say, hey, what's the best way to approach it?
11:11
SPEAKER_00: Because it's a pattern seeker,
11:14
SPEAKER_00: it's going to drop off what I call the soft edges,
11:18
SPEAKER_00: the ways that are normally But if you think about business,
11:22
SPEAKER_00: it's doing it differently.
11:24
SPEAKER_00: That is the key to strategy.
11:26
SPEAKER_00: So we've got to be careful, and I've seen it happen with AI.
11:30
SPEAKER_00: It'll give me back an answer where optionality is really removed
11:34
SPEAKER_00: because it's not strong enough in its pattern.
11:37
SPEAKER_00: And by doing so, you're really planning on strategy and becoming the same as everyone else.
11:43
SPEAKER_00: And when you have homogeny and have strategy, you just have nothing unique to offer.
11:50
SPEAKER_00: So I think that's a pattern that I've seen very often.
11:55
SPEAKER_00: I think we can all speak to the written word in AI and how flat and lifeless it can sound.
12:01
SPEAKER_00: You can recognize it.
12:03
SPEAKER_00: There's a tone about it.
12:04
SPEAKER_00: It's like looking at someone with eyes that can't see.
12:07
SPEAKER_00: It's just rote information that's losing its depth.
12:11
SPEAKER_00: It's losing its insight.
12:13
SPEAKER_00: What it does well on the other side of that is when you add the human factor,
12:18
SPEAKER_00: because it's just a tool,
12:20
SPEAKER_00: but when you add the human creativity, the human mind, It's the uplift AI.
12:26
SPEAKER_00: That's where the magic is.
12:28
SPEAKER_00: And so I can start with an idea and I can ideate something and I can say, how does that work?
12:34
SPEAKER_00: What corner am I forgetting here?
12:36
SPEAKER_00: What would this, ask me 10 questions that would make this argument 30% stronger.
12:43
SPEAKER_00: And now I'm doing two things.
12:44
SPEAKER_00: I'm increasing the strength of my argument, but I'm inviting AI into my thinking.
12:49
SPEAKER_00: So whatever it outputs automatically becomes more tailored.
12:54
SPEAKER_00: So it does that very well.
12:56
SPEAKER_00: I think AI by itself is meaningless.
13:00
SPEAKER_00: It truly is to me.
13:01
SPEAKER_00: I recently did a podcast for a law school and they said, what about lawyers in the future?
13:07
SPEAKER_00: Are we doing anything anymore?
13:08
SPEAKER_00: Are we training anyone to do anything significant?
13:12
SPEAKER_00: And I said, you absolutely are.
13:14
SPEAKER_00: We're each talented in our unique ways.
13:17
SPEAKER_00: And it's the unique way that makes AI magic.
13:20
SPEAKER_00: It's not that AI is magic.
13:22
SPEAKER_00: It's that our control of it and our ability to use our creativity to lift it as magic.
13:27
SPEAKER_00: So AI to me is neither good nor bad.
13:30
SPEAKER_00: It's a tool to be utilized and controlled with human kindness.
13:34
SPEAKER_02: Do you use any other AI-based tools to help support you in your work?
13:40
SPEAKER_00: I use a lot of them.
13:41
SPEAKER_00: If I really want to get to the best sources and I'm researching, I'm going to go to perplexity.
13:46
SPEAKER_00: I just know that it's going to take me down deep.
13:48
SPEAKER_00: I'm going to say, give me the best of the best.
13:50
SPEAKER_00: Give me the academic.
13:52
SPEAKER_00: I tend to pull my resources from academic sources on my RSS feeds.
13:57
SPEAKER_00: I don't tend to do downstream articles because then I'm dealing with someone else's thing.
14:02
SPEAKER_00: So I like to go to the most honest source of things.
14:06
SPEAKER_00: That helps me create solutions to the problems that they're pointing out.
14:12
SPEAKER_00: So I use perplexity a lot of times to make sure my sources are clean, knowledge is clean.
14:17
SPEAKER_00: I use Claude in various ways.
14:20
SPEAKER_00: I think Claude is very good at writing.
14:22
SPEAKER_00: It can be great in helping me parse through all 200 articles because I got so excited.
14:28
SPEAKER_00: You know,
14:29
SPEAKER_00: and I still have to bring it down to something that I can utilize and saying,
14:33
SPEAKER_00: hey, statistically, what seems to match up and give a theme?
14:37
SPEAKER_00: So I'll use it in that context, and I find that very helpful.
14:42
SPEAKER_00: ChatGPT I use if I'm having a bad day and I need someone to tell me how wonderful I am.
14:47
SPEAKER_02: Oh, it's good at that, certainly.
14:50
SPEAKER_00: So I use lots of tools, N8N, if I want to build something cool and play around with some nodes.
14:56
SPEAKER_00: If I want to vibe code when I'm just having a good old day where something comes up,
15:00
SPEAKER_00: I once vibe coded for the EU AI.
15:04
SPEAKER_00: It was a developer tool which would tell them as they were ideating which level
15:10
SPEAKER_00: of restriction it fell into under the EU AI so they could change the deployment
15:16
SPEAKER_00: if it was too high for their comfort level.
15:18
SPEAKER_00: So again, it's a world of fun, but it's in my control.
15:23
SPEAKER_02: That sounds like an interesting application and certainly one that maybe a lot
15:26
SPEAKER_02: of people could use.
15:27
SPEAKER_02: Have you thought about releasing it as a tool or even as an open source product?
15:32
SPEAKER_02: I made it as an
15:33
SPEAKER_00: open source.
15:34
SPEAKER_00: It was actually a hackathon and everyone else was doing these great travel apps.
15:39
SPEAKER_00: This just seems to me that's something that they need to understand
15:43
SPEAKER_00: before they start getting too far down the coding,
15:47
SPEAKER_00: where they're falling with respect to their risk level.
15:50
SPEAKER_00: So it's an intentional deployment.
15:53
SPEAKER_00: So I don't know what will come in the future, but that was actually a hackathon that I did.
15:59
SPEAKER_02: And the tool that I think we all use,
16:00
SPEAKER_02: for instance,
16:01
SPEAKER_02: where doing this interview over Zoom and right now Zoom is generating
16:05
SPEAKER_02: a live transcript of our words as we're talking in addition to doing
16:08
SPEAKER_02: the audio recording.
16:10
SPEAKER_02: Some people use AI note takers
16:12
SPEAKER_02: and I've even seen some people using virtual employees as assistants that join
16:17
SPEAKER_02: into meetings.
16:17
SPEAKER_02: Do you have any experience with those?
16:20
SPEAKER_00: I do.
16:20
SPEAKER_00: I actually have a real story about that that proves that I am anything but infallible.
16:27
SPEAKER_00: One of my associates came up to me, actually a great emerging head of technology's lead.
16:32
SPEAKER_00: There was an idea that for operations we could create an AI employee
16:36
SPEAKER_00: and we could have a whole bio around and then website and everything and put them
16:42
SPEAKER_00: the website and do all this stuff.
16:44
SPEAKER_00: We were really developing this idea.
16:46
SPEAKER_00: It just sounded fabulous.
16:48
SPEAKER_00: Of course, we gave it a job description.
16:49
SPEAKER_00: We gave it guardrails, did all the important things you should do with agentic releases.
16:54
SPEAKER_00: And if we discuss a little later about the agentic constitution,
16:57
SPEAKER_00: we can go deeper into how to control specific agentic employees.
17:01
SPEAKER_00: And we get through the process and we're pre-release and She calls me back
17:05
SPEAKER_00: and she said, we have an issue.
17:07
SPEAKER_00: And I said, what?
17:08
SPEAKER_00: And she said, well, it's a misrepresentation.
17:09
SPEAKER_00: We can't make it look like a live employee.
17:12
SPEAKER_00: So we can the whole thing.
17:15
SPEAKER_00: You have to always be looking.
17:17
SPEAKER_00: That would be,
17:18
SPEAKER_00: I don't want to say analogous to false advertising,
17:20
SPEAKER_00: but it's putting liability on something that doesn't actually exist.
17:24
SPEAKER_00: So this is all emerging for all of us.
17:27
SPEAKER_00: And as we all try to find new ways to get creative,
17:31
SPEAKER_00: there are guardrails and you must continually check for them.
17:35
SPEAKER_00: So there's my experience on how it can go awry as someone who's in the field.
17:41
SPEAKER_02: Yeah, that's a really interesting story.
17:43
SPEAKER_02: So thank you for sharing that.
17:44
SPEAKER_02: As you were talking toward the end,
17:45
SPEAKER_02: it was reminding me of the saying that they quoted some computer person when
17:50
SPEAKER_02: the computers first started coming out,
17:52
SPEAKER_02: basically saying that a computer cannot be accountable for a decision.
17:57
SPEAKER_02: Therefore, it cannot make a decision.
17:59
SPEAKER_02: There always has to be a human who remains responsible for whatever decision,
18:03
SPEAKER_02: even if the computer is providing advice or guidance or recommendations.
18:08
SPEAKER_00: That's true and that's a large part of what I do with Mueller Consulting.
18:12
SPEAKER_00: It's really not about just a governance framework or an agenda constitution
18:18
SPEAKER_00: when we teach them meaningful ways to control their deployment.
18:21
SPEAKER_00: It's about educating fiduciaries their decision because it is a human decision and it is human.
18:29
SPEAKER_00: I think it was Harvard Law Forum recently released an article that said,
18:34
SPEAKER_00: outside of technology,
18:36
SPEAKER_00: 12% of our boards felt comfortable making decisions around AI deployments.
18:44
SPEAKER_00: Concerning and obviously they're feeling the stress as well.
18:46
SPEAKER_00: So one of the things I do to mentor boards and to work with boards is teach them
18:51
SPEAKER_00: about decision hygiene around AI because to me the decision is
18:55
SPEAKER_00: the most important part of AI,
18:57
SPEAKER_00: the human decision,
18:58
SPEAKER_00: not the decision AI will make because it'll make the decision you allow it to make
19:03
SPEAKER_00: with your own guardrails.
19:05
SPEAKER_00: So how we make decisions in the board and in leadership around AI,
19:11
SPEAKER_00: whether we've reduced them to writing and put the reasoning behind them
19:14
SPEAKER_00: so we can revisit it.
19:16
SPEAKER_00: Because in AI, decisions decay.
19:19
SPEAKER_00: They do.
19:19
SPEAKER_00: What was a good deployment yesterday doesn't necessarily make sense today.
19:24
SPEAKER_00: And we need to be able to revisit those decisions.
19:28
SPEAKER_00: So decision is a huge part of what I do to give the board understanding of what
19:34
SPEAKER_00: an awesome responsibility they sitting with this fiduciary obligation.
19:40
SPEAKER_02: You've mentioned the Agenda Constitution.
19:42
SPEAKER_02: Could you say a few words about that?
19:44
SPEAKER_00: Sure.
19:45
SPEAKER_00: The Agenda Constitution, again,
19:46
SPEAKER_00: came on this idea,
19:48
SPEAKER_00: I want to make sure that fiduciary leadership understands what it's doing.
19:53
SPEAKER_00: It's got a vision and we do that with governance.
19:55
SPEAKER_00: The first thing we do is a charter.
19:57
SPEAKER_00: Then it struck me that as we were agents.
20:03
SPEAKER_00: We weren't giving them any specific job description.
20:06
SPEAKER_00: We weren't telling them the behavior.
20:08
SPEAKER_00: But we would do that to any employee that came in the door.
20:11
SPEAKER_00: But here we were giving a piece of machinery,
20:14
SPEAKER_00: passwords and privileges and access,
20:18
SPEAKER_00: in some cases unfettered, and saying, now go figure out what your job is.
20:22
SPEAKER_00: Because anything we don't code in there is going to make a decision right,
20:25
SPEAKER_00: wrong, or indifferent.
20:27
SPEAKER_00: So what I decided to do was to put together what I call the Agenta Constitution.
20:32
SPEAKER_00: 80% is the universal layer.
20:35
SPEAKER_00: That is about 12 principles that the board adopts with respect to bias drift
20:42
SPEAKER_00: and it's working against diversity in knowledge,
20:45
SPEAKER_00: diversity in tools,
20:47
SPEAKER_00: always making sure they're within jurisdictional regulatory confines.
20:51
SPEAKER_00: The types of things that can be static and should follow every deployment.
20:56
SPEAKER_00: Now that doesn't mean you won't make an exception.
20:57
SPEAKER_00: There may be an exception made.
20:59
SPEAKER_00: That is a governance event that with good decision hygiene is written down.
21:04
SPEAKER_00: As is the reasoning so someone who goes to revisit understands why the exception was made.
21:10
SPEAKER_00: Then you have what I call the 20% deployment level, and that's specific to each deployment.
21:16
SPEAKER_00: And what that really is is the conversation that I needed to see between the board
21:20
SPEAKER_00: that had to make the fiduciary decision and the CTO and risk management and
21:25
SPEAKER_00: the leadership that had to build, deploy, live, audit, and sunset by.
21:33
SPEAKER_00: And so it is six members of your leadership team or six functional positions as
21:37
SPEAKER_00: a matter of they're actually different people are having legal involved,
21:41
SPEAKER_00: having your finance person involved,
21:44
SPEAKER_00: having your CTO involved, having your risk person involved.
21:47
SPEAKER_00: All these positions are going to go through any suggested deployment that needs
21:53
SPEAKER_00: to be considered by the board and write a brief that goes through each part of what
21:58
SPEAKER_00: the board really needs to understand in English.
22:02
SPEAKER_00: So it understands and it can make a good decision on whether this deployment meets
22:09
SPEAKER_00: the needs of what they're trying to accomplish and remains in the That same document
22:15
SPEAKER_00: then can be used by the CTO to help the developer build it to the board's intent.
22:22
SPEAKER_00: They can take that document and code with it.
22:25
SPEAKER_00: Everything that has been decided has gone through this.
22:29
SPEAKER_00: And now intent is being shared between the technology and the fiduciary leadership.
22:35
SPEAKER_00: And that's the conversation in the real world that needs to be.
22:39
SPEAKER_00: So fiduciary intent is actually coded into the process.
22:43
SPEAKER_02: Yeah, that sounds like a very effective technique,
22:45
SPEAKER_02: and you've done this with a number of companies?
22:48
SPEAKER_00: Yes, I have.
22:49
SPEAKER_00: Again,
22:50
SPEAKER_00: it's about giving them control and showing them where human primacy comes into
22:55
SPEAKER_00: the picture.
22:56
SPEAKER_00: AI is a tool that is subject to what the human rules are,
22:59
SPEAKER_00: and we must make those rules,
23:02
SPEAKER_00: or it will make the rules for us, and that is unmitigated risk.
23:06
SPEAKER_02: That sounds good.
23:07
SPEAKER_02: So it sounds like you made a good headway in improving the state of the world
23:12
SPEAKER_02: in terms of responsible execution of AI,
23:16
SPEAKER_02: and you've made some good use of the tools for yourself.
23:19
SPEAKER_02: I'm wondering if there are any things that you specifically avoid using AI for,
23:24
SPEAKER_02: things that you would just never use AI to do?
23:28
SPEAKER_00: Well, I think when we talk about decision hygiene and we talk about decisions
23:32
SPEAKER_00: that are important,
23:33
SPEAKER_00: we talk about irreversibility,
23:35
SPEAKER_00: we talk about harm that can be caused,
23:37
SPEAKER_00: whether it's reputational,
23:39
SPEAKER_00: physical, mental, people who are never in the room to begin with.
23:43
SPEAKER_00: These are decisions that have to remain human and only human.
23:48
SPEAKER_00: I have seen companies utilize to what I call the authority chain,
23:53
SPEAKER_00: you know, the lower, the next rung, the next rung.
23:56
SPEAKER_00: And by the time they They flatten the optionality of the issue
24:01
SPEAKER_00: so much you no longer see the risk.
24:04
SPEAKER_00: So I think when we talk in terms of what's a higher risk to us,
24:09
SPEAKER_00: what are the things that actually bring danger or hurt people that were never part
24:14
SPEAKER_00: of the conversation to begin with?
24:16
SPEAKER_00: Those are absolutely human decisions and should remain that way because they require
24:23
SPEAKER_00: that type of consideration.
24:27
SPEAKER_00: There is an augmented middle.
24:28
SPEAKER_00: It's looking about how permanent it is,
24:31
SPEAKER_00: whether it's reversible,
24:32
SPEAKER_00: whether it's revisitable, what the harm is, and you can use.
24:37
SPEAKER_00: But again, the human in the loop then is still the important part of that.
24:43
SPEAKER_00: And then there's what I call the low-hanging fruit decisions.
24:46
SPEAKER_00: Those are the ones that at the end really don't have big consequences.
24:50
SPEAKER_00: What am I having for lunch today?
24:52
SPEAKER_00: Turkey or salad?
24:53
SPEAKER_00: Go ahead, Tom.
24:54
SPEAKER_00: That doesn't matter unless I'm deathly allergic to turkey or salad.
24:59
SPEAKER_00: It's just not a big decision and it can be automated without my input if I'm willing
25:04
SPEAKER_00: to live with the consequences.
25:05
SPEAKER_00: And that's the key decision here.
25:07
SPEAKER_00: Whatever decision it makes,
25:09
SPEAKER_00: if you're not willing to live with the consequences, it can make that decision.
25:13
SPEAKER_02: Yeah, you mentioned concern about hurting people who were never part of the conversation.
25:19
SPEAKER_02: That actually comes up quite a lot in AI ethics because the way a lot of
25:23
SPEAKER_02: the AI tools were built,
25:25
SPEAKER_02: they were built,
25:27
SPEAKER_02: for instance,
25:27
SPEAKER_02: from a very biased subset of the world's content and people's works,
25:33
SPEAKER_02: and so they're not representative, they can be very biased.
25:37
SPEAKER_02: The way that the tools are actually developed in a lot of cases,
25:41
SPEAKER_02: they're using exploitative labor practices and hurting people who were not in
25:46
SPEAKER_02: the room in a lot of cases.
25:48
SPEAKER_02: Women have been underrepresented in the rooms and so we find that the biases of
25:52
SPEAKER_02: the tools could cause harms in that way.
25:54
SPEAKER_02: So the whole idea of looking out for people who weren't in the room is something
25:59
SPEAKER_02: that I think we could use more of at the higher levels of the AI companies.
26:05
SPEAKER_00: That's why I ground systems and OECD principles.
26:09
SPEAKER_00: It is about taking care of the diversity.
26:11
SPEAKER_00: Look, it's a real concern that our knowledge pool,
26:16
SPEAKER_00: so limited to two or three voices that we kill our own ability to be creative
26:21
SPEAKER_00: as humans.
26:22
SPEAKER_00: That's a real concern.
26:24
SPEAKER_00: So one of the reasons I'm in MENA right now and in the GCC area is
26:29
SPEAKER_00: that I'm dedicated to making sure that the pool of thought and the pool
26:32
SPEAKER_00: of representation,
26:34
SPEAKER_00: and specific to women,
26:35
SPEAKER_00: I'm working to develop women in AI here because I think it's a new way
26:39
SPEAKER_00: to give them economic freedom.
26:41
SPEAKER_00: I believe women are naturally creatively drawn to AI and are natural superstars in the area.
26:48
SPEAKER_00: But the point is that I'm out here partially just to make sure that voice
26:54
SPEAKER_00: the pool remains, that all these voices in the pool remains.
26:58
SPEAKER_00: I don't have to agree with them all.
26:59
SPEAKER_00: It doesn't but if I can't even consider it her intellectual capacity and my ability
27:06
SPEAKER_00: to be creatively excellent.
27:09
SPEAKER_00: And that is not something I can support for our human race.
27:13
SPEAKER_00: Knowledge scarcity is not something I can get in mind.
27:16
SPEAKER_02: Thanks for sharing all of that.
27:18
SPEAKER_02: So we've touched on this a little bit,
27:19
SPEAKER_02: but one concern that comes up is When we talk about AI systems and
27:25
SPEAKER_02: whether they are ethical and suitable for humans to use
27:28
SPEAKER_02: and what our responsibilities are,
27:29
SPEAKER_02: we look at where these tools got the data that they used to train on the sourcing.
27:36
SPEAKER_02: Sometimes they've scraped data.
27:39
SPEAKER_02: Different available online sources that were not public domain.
27:42
SPEAKER_02: They were actually copyrighted.
27:43
SPEAKER_02: And I don't know how much you get into intellectual property issues
27:46
SPEAKER_02: in your governance work,
27:47
SPEAKER_02: but it's one of the big concerns in lawsuits and copyright infringement
27:52
SPEAKER_02: and unfair competition.
27:54
SPEAKER_02: So there's a lot of activity around that.
27:57
SPEAKER_02: Sometimes companies will use information that we put in online.
28:00
SPEAKER_02: LinkedIn is one example.
28:02
SPEAKER_02: They've done that.
28:03
SPEAKER_02: So I'm wondering what your thoughts are about the way that companies get the data
28:09
SPEAKER_02: that they use for training their systems and how those practices could
28:16
SPEAKER_02: or should impact the decisions people make about using the tools that are built
28:19
SPEAKER_02: on them.
28:20
SPEAKER_00: Well, data hygiene is something that's an important part of the process.
28:24
SPEAKER_00: If your data isn't good and what you're relying upon isn't good,
28:27
SPEAKER_00: then what you output is going to be even worse.
28:30
SPEAKER_00: I think we have an inherent responsibility to the human race to make sure
28:35
SPEAKER_00: that we're not creating a bunch of stuff that dirties the water and
28:39
SPEAKER_00: that we are offering diverse perspectives.
28:42
SPEAKER_00: It's not just the inclusivity of voices.
28:45
SPEAKER_00: It's the idea that thoughts, whole thought streams would be excluded.
28:52
SPEAKER_00: That's just a terrible way to imagine the future.
28:56
SPEAKER_00: I think it's very, very important that data hygiene be part of the conversation in Egypt now.
29:04
SPEAKER_00: They're actually doing their first big release around data privacy rules, which is huge.
29:10
SPEAKER_00: They've never had it.
29:11
SPEAKER_00: They've had the GDPR in Europe, which is very important.
29:14
SPEAKER_00: We've got HIPAA, of course, in the United States around data privacy.
29:18
SPEAKER_00: Egypt now understands how important it is to get the data,
29:23
SPEAKER_00: to have the proper permissions,
29:25
SPEAKER_00: to retire it when the permission is revoked,
29:27
SPEAKER_00: to use it responsibly,
29:30
SPEAKER_00: which is part of why I say when you deploy,
29:33
SPEAKER_00: make sure you've already got your sun setting because your data obligations live
29:37
SPEAKER_00: beyond your deciding to sunset your tool.
29:41
SPEAKER_00: So I think, again,
29:41
SPEAKER_00: it's being responsible making sure what we're doing is allowable the right way
29:46
SPEAKER_00: within the guidelines and is ethical.
29:48
SPEAKER_02: So how do you advise these companies on their responsibilities when it comes
29:53
SPEAKER_02: to using the major large language models,
29:58
SPEAKER_02: ChatGPT and Plot and Perplexity and such,
30:01
SPEAKER_02: as far as where those systems got their data and what risks they're taking,
30:05
SPEAKER_02: if they use them to generate words,
30:07
SPEAKER_02: to generate images, to generate music or anything like that?
30:11
SPEAKER_00: Yeah, well, around copyright issues, they're still figuring that out.
30:14
SPEAKER_00: We're seeing courts go various ways.
30:18
SPEAKER_00: We did just see a major case for copyright win, and that's great for the copyright holder.
30:23
SPEAKER_00: I think these issues are going to continue to be important.
30:27
SPEAKER_00: Most corporations are aware of copyright issues through non-disclosure agreements
30:31
SPEAKER_00: and things of that nature.
30:33
SPEAKER_00: They speak about them, but they're building the LLM and that's very opaque to them.
30:39
SPEAKER_00: I think what you have to say to leadership or what I do say to leadership is
30:45
SPEAKER_00: because you don't know the source, you don't know what you're giving up in using it.
30:51
SPEAKER_00: So again,
30:52
SPEAKER_00: we don't allow it to run with the human in the loop with the testing of
30:58
SPEAKER_00: whether that's a fair representation with our work of going and doing
31:03
SPEAKER_00: the good research.
31:04
SPEAKER_00: That's what makes us strategically placed to do responsible deployments that value today.
31:13
SPEAKER_00: And then tomorrow when you go to market,
31:14
SPEAKER_00: because people want to see you're using it responsibly.
31:17
SPEAKER_00: And when you go to due diligence,
31:19
SPEAKER_00: one of the things they're going to look at is if you are aware of not only
31:22
SPEAKER_00: where your data comes from,
31:24
SPEAKER_00: but what jurisdictions it's crossing into and could be stepped on further and
31:28
SPEAKER_00: whether you'll have access to it should another geopolitical burst.
31:34
SPEAKER_02: Yeah, a lot of aspects there that require some careful thought.
31:37
SPEAKER_02: And it's good that you're providing guidance on that because it seems like a lot
31:41
SPEAKER_02: that is missing.
31:42
SPEAKER_02: In a lot of the world nowadays.
31:44
SPEAKER_02: So another question that comes up has to do with when companies are accessing data,
31:51
SPEAKER_02: whether or not they should be required to do what some people call the three C's
31:54
SPEAKER_02: of creative rights, which is consent, credit, and compensation.
31:59
SPEAKER_02: If a company is going to use data that they collect from consumers
32:04
SPEAKER_02: or collect publicly,
32:06
SPEAKER_02: what are your thoughts about whether they ought to be required to get consent and
32:11
SPEAKER_02: to give credit and to potentially give compensation as well?
32:17
SPEAKER_00: Well, consent is a requirement under a data privacy scheme,
32:22
SPEAKER_00: certainly the GDPR, knowing consent is actually part of the process.
32:26
SPEAKER_00: When we're talking about compensation,
32:28
SPEAKER_00: I presume we're really looking more at the artistic or the copyright value stuff.
32:33
SPEAKER_00: I think that people who creatively have come up with something unique
32:38
SPEAKER_00: and is co-opted needs to be compensated.
32:42
SPEAKER_00: I think we're going to be arguing for many,
32:44
SPEAKER_00: many moons over what a derivative work is and how far derivative it has to be so
32:51
SPEAKER_00: that compensation isn't necessary because it's a new piece of work.
32:54
SPEAKER_00: And we've done this between humans and it's a gray area with machines and we need
33:00
SPEAKER_00: to figure out where that line is.
33:04
SPEAKER_02: So for the tools that you've used,
33:07
SPEAKER_02: you mentioned Perplexity and ChatGPT,
33:10
SPEAKER_02: but as someone who uses those tools,
33:13
SPEAKER_02: do you feel like the providers of those tools have been transparent about sharing
33:18
SPEAKER_02: where they got the data that they use for those tools?
33:22
SPEAKER_00: Well, the way I use them is I'm source searching.
33:25
SPEAKER_00: So I can actually see the original article.
33:28
SPEAKER_00: I'm actually reading the original stuff, and this stuff has been released already.
33:33
SPEAKER_00: And that's what I'm citing.
33:34
SPEAKER_00: I'm not having AI make my decisions about what it says.
33:38
SPEAKER_00: I'm actually going to the source.
33:40
SPEAKER_00: And for what I do in particular, look,
33:42
SPEAKER_00: I can't tell people to be responsible about it if I'm not pull up the source
33:46
SPEAKER_00: and read the actual article.
33:48
SPEAKER_00: And then you get the mismatch,
33:50
SPEAKER_00: how things are vetted,
33:51
SPEAKER_00: where the optionality was flattened,
33:53
SPEAKER_00: and where the strategy around the edges and the soft sides is dropped off.
33:57
SPEAKER_00: And that's why you get a lot of voices out there that sound the same.
34:01
SPEAKER_00: But no, I use it to go to the sources.
34:04
SPEAKER_00: I never say, hey,
34:05
SPEAKER_00: give me what Freddie Mercury sang and let's make a song off of it
34:08
SPEAKER_00: so I can sound just like him.
34:10
SPEAKER_00: That's not how I use AI.
34:13
SPEAKER_00: But I have no problem ideating with AI because, again, I'm involved in the process.
34:20
SPEAKER_00: I have no problem building with AI because, again, I'm in the process.
34:25
SPEAKER_00: I don't have a system that I say for anything important or anything that's going to infringe.
34:30
SPEAKER_00: Go have a good day.
34:32
SPEAKER_00: Let me know when you're done, and I'll see you on the other side.
34:35
SPEAKER_00: That's not how I utilize it.
34:37
SPEAKER_02: So you're really using perplexity and such more as a more powerful search tool,
34:43
SPEAKER_02: not having it generate anything for you?
34:46
SPEAKER_00: I use it to say, look, this is a theory that I, it's one of the ways I'll use it.
34:52
SPEAKER_00: This is a theory in a framework that I've been working.
34:56
SPEAKER_00: And I use LLMs and I cross check with them.
34:59
SPEAKER_00: That's one of the ways that I make sure my tools are good.
35:02
SPEAKER_00: It's never just one system.
35:03
SPEAKER_00: I will have Claude,
35:04
SPEAKER_00: check Perplexity,
35:06
SPEAKER_00: check ChatGPT because again, that's part of the due diligence in this, right?
35:11
SPEAKER_00: We want to flag problems.
35:13
SPEAKER_00: And I'll go to Perplexity and I'll say,
35:14
SPEAKER_00: I want you to pull up anything that sources that looks similar or is on point,
35:21
SPEAKER_00: who has come down this road and ideated in this fashion.
35:25
SPEAKER_00: And if it's out there, let's uncover it.
35:28
SPEAKER_00: So I can say whether I've developed or built on something.
35:32
SPEAKER_00: I'm not pulling stuff out of thin air.
35:34
SPEAKER_00: I'm obviously building on anything.
35:36
SPEAKER_00: The Art of Decision Hygiene has existed.
35:38
SPEAKER_00: Cassie Kaziakov has been speaking about it for years.
35:42
SPEAKER_00: Annie Duke has been speaking about it for years.
35:45
SPEAKER_00: Decision decay is something I built off that because it wasn't enough to make
35:51
SPEAKER_00: the decision in AI.
35:52
SPEAKER_00: I realized that in the fiduciary construct,
35:55
SPEAKER_00: there was a decay problem that they were going to be held accountable for.
36:00
SPEAKER_00: And so that's a build on something.
36:02
SPEAKER_00: But I always credit Cassie.
36:04
SPEAKER_00: I've done it several times, I think, during this conversation.
36:07
SPEAKER_00: I think it's important to talk about the people you're building on
36:11
SPEAKER_00: because adds not only credibility but gives recognition where recognition is due.
36:18
SPEAKER_00: These are the people who have inspired me.
36:20
SPEAKER_02: Yeah, Kathy was one of the first people that I found when I got to Substack
36:23
SPEAKER_02: and started looking for Substacks written by women about AI.
36:28
SPEAKER_02: And yeah, he's a great resource.
36:30
SPEAKER_02: And I love that you're calling out Annie as well, seeing women credited as your sources.
36:34
SPEAKER_02: That makes me happy.
36:36
SPEAKER_00: Yes, yes, absolutely.
36:37
SPEAKER_00: And Cassie, I was fortunate enough, she was one of the teachers in my program.
36:43
SPEAKER_00: So she really inspired me on the decision side of things and she inspired me.
36:49
SPEAKER_00: She does it from a tech side,
36:52
SPEAKER_00: you know,
36:52
SPEAKER_00: and having that opportunity to say, gosh,
36:55
SPEAKER_00: if I were sitting in the seat boardroom,
36:58
SPEAKER_00: what would I need to know and how would I need to remain responsible for the process
37:05
SPEAKER_00: and be able to do so without having to point to an agent.
37:10
SPEAKER_00: And so she really inspired my work.
37:13
SPEAKER_00: Noelle Russell has inspired a lot of my work with the agentic constitution.
37:18
SPEAKER_00: So I've been inspired by many women in AI.
37:20
SPEAKER_02: That's great to hear that.
37:22
SPEAKER_02: I want to switch for a minute to thinking about use of our data as consumers,
37:28
SPEAKER_02: as members of the public,
37:30
SPEAKER_02: our personal data and content is
37:32
SPEAKER_02: probably getting used by AI-based tools
37:34
SPEAKER_02: or systems somewhere.
37:35
SPEAKER_02: Do you know of any specific cases that you could share where you know
37:39
SPEAKER_02: that your information has been used?
37:41
SPEAKER_00: Well, I don't know that I can share that it's been used because I'm privy to that.
37:45
SPEAKER_00: I don't know what they're training.
37:47
SPEAKER_00: But for everyone who thinks wiping their past conversations
37:51
SPEAKER_00: and deleting does any bud,
37:53
SPEAKER_00: everyone should understand that Sam Alt and ChatGPT are involved in
37:57
SPEAKER_00: a very large lawsuit.
37:59
SPEAKER_00: And as part of the lawsuit, they must hold up.
38:02
SPEAKER_00: To all conversations and they are available to the court.
38:07
SPEAKER_00: So is saying don't hold on to my information actually happening?
38:12
SPEAKER_00: No, it will be used in that court case if it can be used by either side.
38:16
SPEAKER_00: So I think we need to be aware that that is happening even when we think it's not.
38:20
SPEAKER_02: Yeah, one area is the LLMs and the way that people interact with them,
38:26
SPEAKER_02: even the titles that get assigned to the chats.
38:29
SPEAKER_02: That information,
38:30
SPEAKER_02: even if the content of the chat isn't disclosed,
38:32
SPEAKER_02: I understand that now a lot of the titles,
38:35
SPEAKER_02: the metadata about what the chats were and the discussions are more,
38:40
SPEAKER_02: I won't say public necessarily, but certainly more accessible to others.
38:45
SPEAKER_02: They're not private, but much more so than I think people would realize.
38:48
SPEAKER_02: If someone who travels,
38:50
SPEAKER_02: you've gone through TSA and they use biometrics now and they use facial recognition
38:56
SPEAKER_02: to match people up to their passports.
38:58
SPEAKER_02: Obviously social media sites are big for that.
39:01
SPEAKER_02: You know, we sign up for websites for streaming.
39:03
SPEAKER_02: They want to know our birthdays.
39:05
SPEAKER_02: And so our data is everywhere.
39:07
SPEAKER_02: And I think sometimes people don't always realize what's being done with that data.
39:13
SPEAKER_00: Interestingly enough,
39:14
SPEAKER_00: one of the first cases I worked on as a young editor was with a company that does
39:20
SPEAKER_00: a lot of research.
39:23
SPEAKER_00: And as part of their process, they were getting a lot of driver's licenses information.
39:30
SPEAKER_00: And it was one of the first cases that gave rubber stamp to say, well,
39:35
SPEAKER_00: actually that's something that has valued them that they've purchased with
39:39
SPEAKER_00: the database.
39:39
SPEAKER_00: They can sell it to providers.
39:42
SPEAKER_00: And so that was back in 1999.
39:45
SPEAKER_00: Wow.
39:46
SPEAKER_00: The U.S. has begun to step it back since then.
39:49
SPEAKER_00: But that was a real case where the courts said,
39:52
SPEAKER_00: no, actually that in itself is an asset that they bought along.
39:56
SPEAKER_00: The data and the right to resell it had value to them and they had a right to exercise that.
40:01
SPEAKER_00: So we have seen the farthest expansion and we hope it comes back.
40:07
SPEAKER_00: The GDP are like legislation that's coming in November of 2026 in Egypt is to stop
40:18
SPEAKER_00: a lot of biometric use.
40:20
SPEAKER_00: We have clubs here.
40:22
SPEAKER_00: You pull up to the club and it's a biometric entrance.
40:25
SPEAKER_00: People are selling to real estate companies and everything else.
40:29
SPEAKER_00: This is just part of it.
40:31
SPEAKER_00: People are getting it and they're putting it together slowly.
40:34
SPEAKER_00: But to see a place like Egypt that actually is releasing a more strict standard, Dr.
40:39
SPEAKER_00: Suzanne El-Akabawi is leading the efforts here and watching what she's done
40:45
SPEAKER_00: to say what we really need to say to people is we're going to be so strict
40:48
SPEAKER_00: that you're not going to want to violate this.
40:51
SPEAKER_00: And we need to start thinking about that because it is a violation of individual rights.
40:56
SPEAKER_02: Yeah, absolutely.
40:58
SPEAKER_02: So I'm wondering,
41:00
SPEAKER_02: have ever had a time when a company's use of your data has caused problems for you?
41:05
SPEAKER_02: Phishing attacks, privacy violations, identity theft, data breaches.
41:10
SPEAKER_02: I think it may be better overseas from what I hear from my guests,
41:12
SPEAKER_02: but in the U.S. here, it's been quite a problem.
41:15
SPEAKER_00: Yeah, we maintain a residence in Germany as well.
41:18
SPEAKER_00: They had a major hacking that occurred in the late 80s,
41:22
SPEAKER_00: I think it was 80 years,
41:23
SPEAKER_01: around their banking system.
41:25
SPEAKER_00: It is illegal almost to do anything with data.
41:28
SPEAKER_00: You've got the GDPR, but Germany itself is very much against this use and proliferation.
41:36
SPEAKER_00: And it's incredibly difficult even to do online banking there.
41:40
SPEAKER_00: They send you a 20 number code and each one's unique every time.
41:43
SPEAKER_00: I never moved money over in Germany.
41:46
SPEAKER_00: I have an account and my husband does it.
41:48
SPEAKER_00: I have no idea how to do it.
41:50
SPEAKER_00: I would say other areas like the US are much freer about it and again
41:54
SPEAKER_00: the courts there have felt at the federal level that in itself people's information
42:00
SPEAKER_00: when it's bought for other purposes can be used as an additional sale is problematic
42:07
SPEAKER_00: because you didn't sign over ownership And that's where we saw HIPAA come
42:12
SPEAKER_00: into effect.
42:13
SPEAKER_00: You just couldn't wait and get someone's health data.
42:16
SPEAKER_00: But we need stronger guardrails around this because of what is occurring.
42:22
SPEAKER_00: It's not just the phishing.
42:23
SPEAKER_00: It's not just the data leak.
42:25
SPEAKER_00: I was part of the target.
42:27
SPEAKER_00: With the credit cards and stuff.
42:29
SPEAKER_00: I've gotten free monitoring of my credit for the rest of my life.
42:33
SPEAKER_00: But these are real concerns and data privacy is a huge part of AI and it's part
42:39
SPEAKER_00: of responsible AI.
42:40
SPEAKER_00: So we need to continue to monitor how data is being used,
42:44
SPEAKER_00: get more strict in it because it can be dangerous.
42:47
SPEAKER_02: Yeah, absolutely.
42:49
SPEAKER_02: So last question and then we can talk about anything else that you want.
42:52
SPEAKER_02: Public distrust of AI and tech companies has been growing lately.
42:57
SPEAKER_02: And it's a bit of a backlash to finding out more about what they're doing with our data.
43:02
SPEAKER_02: And I'm wondering what you think is the most important thing that companies need
43:07
SPEAKER_02: to do to earn and keep the trust of people, or if that's even possible.
43:12
SPEAKER_00: Well, I think we have been,
43:13
SPEAKER_00: and I've written a little bit about this deeply embedded in a time of opacity,
43:18
SPEAKER_00: where things were passed in the boardroom like we all agree passed.
43:23
SPEAKER_00: That is no longer going to work.
43:25
SPEAKER_00: Transparency in the boardroom, and that's part of my decision hygiene framework.
43:30
SPEAKER_00: Transparency in leadership is very important.
43:33
SPEAKER_00: What we did, why we did it, what the limitations are, what's going to cause us to revisit it.
43:39
SPEAKER_00: Are all part of the things that should be easily accessible and trans being recorded.
43:46
SPEAKER_00: And I think that if corporations want trust in AI,
43:50
SPEAKER_00: transparency with their people,
43:52
SPEAKER_00: what they're doing,
43:53
SPEAKER_00: what their intent is,
43:54
SPEAKER_00: what their employment plans are,
43:56
SPEAKER_00: what's their human talent plan, all these things will build trust.
43:59
SPEAKER_00: We can't stop every negative thing from happening.
44:03
SPEAKER_00: We can be honest and transparent about what's going on.
44:07
SPEAKER_00: And let people make up their mind if they want to engage with it.
44:11
SPEAKER_02: Yeah, the whole talent plan is an aspect that I think doesn't get talked about very much.
44:16
SPEAKER_02: There's a lot of announcements of, oh,
44:19
SPEAKER_02: we're starting to use AI and it's being blamed sometimes fairly,
44:23
SPEAKER_02: sometimes not as the reason for layoffs.
44:26
SPEAKER_02: But then there are companies like IKEA,
44:28
SPEAKER_02: which famously,
44:29
SPEAKER_02: when they adopted AI,
44:31
SPEAKER_02: they didn't lay off all of these customer service representatives.
44:34
SPEAKER_02: They re-skilled them and they added another billion dollars to the business for the year.
44:39
SPEAKER_02: And so they're going a completely different direction by the way they take advantage
44:43
SPEAKER_02: of it,
44:43
SPEAKER_02: not for cutting back and cutting costs, but to actually amplify their business.
44:48
SPEAKER_02: I think there's not much transparency around companies that are actively hiring now,
44:53
SPEAKER_02: even to, well, okay,
44:54
SPEAKER_02: we're hiring you now for this,
44:55
SPEAKER_02: but then we're going to have you use the AI tools and they'll take what you know and
45:00
SPEAKER_02: then you'll be gone in a couple of years.
45:03
SPEAKER_02: I think that's a scenario that some people are fearing
45:06
SPEAKER_02: and there's not much reassurance coming out on that perspective
45:10
SPEAKER_02: because companies are so focused on AI washing their business so that it looks like,
45:16
SPEAKER_02: oh, we're using AI too.
45:17
SPEAKER_02: Don't worry, we're not falling behind.
45:19
SPEAKER_00: Right.
45:20
SPEAKER_00: It's the FOMO issue.
45:21
SPEAKER_00: And that's actually a piece I'm writing right now for Substack.
45:24
SPEAKER_00: It'll be out in a couple of weeks.
45:25
SPEAKER_00: We just saw Ford go back to what it called its gray beard hiring.
45:29
SPEAKER_00: It's pulled back its engineers.
45:32
SPEAKER_00: One of the things I make sure I get across to leadership is that institutional knowledge.
45:38
SPEAKER_00: And it's walking out the door is cutting the legs off your sofa and expecting it not
45:43
SPEAKER_00: to smoosh your dust bunnies.
45:45
SPEAKER_00: They're diametrically opposed.
45:48
SPEAKER_00: Where Your AI system fails, and it will.
45:52
SPEAKER_00: If that institutional knowledge has walked out the door, you are beholden to a bad system.
45:59
SPEAKER_00: If tomorrow a geopolitical bomb falls in the place where your data sits,
46:04
SPEAKER_00: and you have to close your pipeline down and your institutional knowledge has left.
46:09
SPEAKER_00: You no longer have a strategy you are begging for survival.
46:14
SPEAKER_00: So what I always advise is to level their people up.
46:17
SPEAKER_00: Again, AI is great for some tasks.
46:21
SPEAKER_00: Take that and then allow your creativity of your people to be increased with
46:26
SPEAKER_00: the tool and they will find better productivity.
46:29
SPEAKER_00: They will find happier employees.
46:32
SPEAKER_00: They will find that they are meeting their goals economically and ethically
46:38
SPEAKER_00: when they use their human talent to level up, not their tools to replace them.
46:43
SPEAKER_00: So I'm writing a piece about that right now in Ford's Recall.
46:47
SPEAKER_02: Oh, awesome.
46:49
SPEAKER_02: Yeah, so by the time this interview publishes,
46:51
SPEAKER_02: we'll make sure we enter the link to your article here at this point in
46:54
SPEAKER_02: the interview.
46:55
SPEAKER_02: That's all the questions I had.
46:56
SPEAKER_02: I would actually like to hear more about what you're doing to support women in AI
46:59
SPEAKER_02: in Egypt, or if there's anything else that you'd like to talk about.
47:02
SPEAKER_00: Well, one of the things that I'm doing,
47:04
SPEAKER_00: and I had mentioned Dr.
47:06
SPEAKER_00: Susana Lakabawi,
47:07
SPEAKER_00: who is the advisor to the IT ministry here,
47:11
SPEAKER_00: and she also is the CEO of the Data Privacy Consortium in Egypt.
47:16
SPEAKER_00: She's a wonderfully talented attorney and we had the good privilege
47:20
SPEAKER_00: of working together previously in law.
47:23
SPEAKER_00: When I was here and I have approached her and spoken with her
47:27
SPEAKER_00: about developing specifically a women in AI program through one of
47:32
SPEAKER_00: the top universities here who I'm not going to mention because I'm meeting
47:35
SPEAKER_00: with them next week.
47:36
SPEAKER_00: So let's just keep our fingers crossed and getting some space.
47:40
SPEAKER_00: It's not going to be about lecturing them about governance.
47:43
SPEAKER_01: It's going to be about getting people in to teach them the tools.
47:46
SPEAKER_00: I'm not a person.
47:48
SPEAKER_00: I'm the person who can say, let's do it responsibly.
47:51
SPEAKER_00: Creativity into it.
47:52
SPEAKER_00: But I believe that women are in a unique position,
47:56
SPEAKER_00: particularly in places like MENA and GCC,
47:58
SPEAKER_00: where they haven't necessarily had the same economic opportunities as women,
48:04
SPEAKER_00: that perhaps we enjoy in the United States for good, bad,
48:07
SPEAKER_00: and we can argue about what those really look like, but it's very different here.
48:12
SPEAKER_00: And so I believe that exposing women to the tool
48:16
SPEAKER_00: and giving them their natural release
48:18
SPEAKER_00: to their creativity will create economic parity for them.
48:23
SPEAKER_00: So it's something that I strongly believe in developing Egypt and hope
48:28
SPEAKER_00: to see develop across the GCC, ASA, and MENA regions.
48:33
SPEAKER_02: That's really exciting.
48:34
SPEAKER_02: I love that you're doing that.
48:36
SPEAKER_02: If there's anything I can do to help you advocate for it.
48:38
SPEAKER_02: We haven't connected with a lot of women from Egypt yet on Substack writing about AI
48:43
SPEAKER_02: and I'd love to hear more voices coming from the area.
48:47
SPEAKER_00: I can certainly help with that.
48:48
SPEAKER_00: There's a lot of great women who are getting stirred in the environment
48:53
SPEAKER_00: and there are different considerations here.
48:55
SPEAKER_00: I mean, we could go around the world and they look at the tools very differently or
49:00
SPEAKER_00: in very different places with them.
49:03
SPEAKER_00: But as I tell them here,
49:04
SPEAKER_00: they're developing their vision and I'm very proud to support them and help them
49:10
SPEAKER_00: with that for what they're going to be doing with AI.
49:13
SPEAKER_00: It's not about, am I China or the US?
49:15
SPEAKER_00: It's where your unique talents fall and how to leverage them.
49:19
SPEAKER_00: For Germany,
49:20
SPEAKER_00: I still think they should be looking at robotics,
49:23
SPEAKER_00: but they're still trying to be one and two, get there eventually.
49:27
SPEAKER_00: But I think it's important that,
49:29
SPEAKER_00: especially in jurisdictions that are a little slower to adopt,
49:33
SPEAKER_00: that we take the shortest path between A and Z.
49:37
SPEAKER_00: So I want to help them avoid those unnecessary fishtail turns
49:43
SPEAKER_00: that other jurisdictions are making by going FOMO.
49:46
SPEAKER_00: We learn from them and we can take the shortest path, however long that takes.
49:51
SPEAKER_00: It'll be the most ethical, best path for their people.
49:55
SPEAKER_02: Well, that sounds great.
49:56
SPEAKER_02: Could you just say the name of your Substack for our listeners
49:59
SPEAKER_02: to I
50:01
SPEAKER_00: think it's Susan Mueller'sSubstack and underneath it, Ethics with Abundance.
50:05
SPEAKER_00: I write about business and how to do things in a way that I hate to just use ethical
50:12
SPEAKER_00: because I think ethical scares people.
50:14
SPEAKER_00: It sounds like a fugazi.
50:15
SPEAKER_00: It sounds like something they can't get in the arms.
50:18
SPEAKER_00: What I like to say is it's calling the lens you choose to privilege
50:22
SPEAKER_00: when two things are intentioned.
50:25
SPEAKER_00: There's no judgment behind that.
50:26
SPEAKER_00: Just know what you're calling.
50:28
SPEAKER_00: And that in itself is a vision.
50:31
SPEAKER_02: So I help with that.
50:33
SPEAKER_02: Yeah.
50:33
SPEAKER_02: When I think about ethics,
50:35
SPEAKER_02: it comes down to first people meaning to think about what their values are,
50:39
SPEAKER_02: what their priorities are,
50:40
SPEAKER_02: and then looking,
50:43
SPEAKER_02: here's what I use AI for,
50:44
SPEAKER_02: here's what I don't, and this is why, and this is how this fits with my values.
50:48
SPEAKER_02: And sometimes just doing this assessment or audit
50:50
SPEAKER_02: or defining your own personal AI policy Policy sounds scary, but we all have one.
50:56
SPEAKER_02: We have something that guides how we make our decisions about what we do and don't do.
51:00
SPEAKER_02: And I think that this going through that process of thinking
51:03
SPEAKER_02: about our values can bring some clarity and maybe open up to some options
51:06
SPEAKER_02: that we didn't realize we had or that we could look for.
51:09
SPEAKER_02: Like you said,
51:10
SPEAKER_02: there's no judgment in it,
51:11
SPEAKER_02: but everyone needs to,
51:12
SPEAKER_02: I think, make their own policy about what they use AI for and what they don't.
51:17
SPEAKER_00: Yeah, and that's really the first step in our system.
51:20
SPEAKER_00: I call it the AI Charter.
51:22
SPEAKER_00: That is the vision of the board.
51:24
SPEAKER_00: That is calling out what we favor and where we'll give privilege to.
51:28
SPEAKER_00: And again, it's not about judgment.
51:30
SPEAKER_00: It's about recognizing where you put your stake in the grant.
51:35
SPEAKER_00: And so every deployment thereafter will follow and be easier because you know where
51:41
SPEAKER_00: the stakes lie.
51:42
SPEAKER_00: If you need to make an exception, obviously business has exceptions, We have to record.
51:48
SPEAKER_00: We have to be transparent.
51:50
SPEAKER_00: And that's how visions come to life on paper by actually embedding them.
51:56
SPEAKER_02: Yeah, that makes total sense.
51:57
SPEAKER_02: So I made the time for this interview and appreciate you sharing that with us.
52:02
SPEAKER_00: I really enjoyed it.
52:04
SPEAKER_01: Yeah, I did as well.
52:05
SPEAKER_01: Thank you.
52:07
SPEAKER_01: About this interview series and newsletter.
52:11
SPEAKER_01: This post is part of our AI 6P interview series on software and wetware.
52:17
SPEAKER_01: It showcases how real people around the world are using their wetware,
52:21
SPEAKER_01: brains and human intelligence,
52:24
SPEAKER_01: with AI-based software tools who are being affected by AI.
52:29
SPEAKER_01: And we're all being affected by AI nowadays in our daily lives, perhaps more than we realize.
52:35
SPEAKER_01: For some examples, see post, but I don't use AI.
52:40
SPEAKER_01: We want to hear from a diverse pool of people worldwide in a variety of roles.
52:45
SPEAKER_01: If you're interested in being a featured interview guest,
52:48
SPEAKER_01: anonymous or with credit, please get in touch.
52:54
SPEAKER_01: 6P's in AI Pods AI6P is a 100% reader-supported publication.
53:01
SPEAKER_01: All new posts are free to read and listen to.
53:04
SPEAKER_01: To automatically receive new AI6P posts and support our work, consider becoming a subscriber.
53:11
SPEAKER_01: It's free.