The one skill AI can't replace | Alexander Schacht | Data Science Hangout

Hey there, welcome to the Paws at Data Science Hangout. I'm Libby Herron, and this is a recording of our weekly community call that happens every Thursday at 12 p.m. U.S. Eastern Time. If you are not joining us live, you miss out on the amazing chat that's going on. So find the link in the description where you can add our call to your calendar and come hang out with the most supportive, friendly, and funny data community you'll ever experience.

I would love to introduce you to our featured leader this week, Alexander Schacht, Senior Director at SciTel, but also the host of the Effective Statistician podcast. Alexander, welcome. I would love it if you could introduce yourself, say a little bit about who you are and something you like to do for fun.

Yeah. So Libby, thanks so much for having me here. I'm super excited to see so many people and even a couple of familiar faces. I'm a statistician by training and have worked since 2002 actually in the pharmaceutical industry, both at various bigger companies as well as CROs. So these are the service providers for the pharmaceutical industry. I'm pretty passionate about being a statistician and I also coded quite a lot in the past, but nowadays not so much anymore. So just so many young people that can do it so much better than me.

So what I really love doing outside of work is actually skiing. I'm just back from a skiing vacation together with my three kids. I have a podcast that is of course free for you. It's over 450 episodes and it's a lot about things that we can talk about today like career advice, like how can you have an influence as a data scientist beyond just your technical skills. I'm very passionate about us as data people to have more influence in our organizations, but also beyond. I think if as a general kind of thing in society, we make better decisions on data will be all better off.

I personally am very passionate about data around patients and health. And here we also need to convince people, stakeholders to make a better decision on data. So it can be kind of people that invest in these areas like sponsors, but it can also be regulators. So like in the US, it would be the FDA, or in Europe, the EMA, or it can be payers. So insurance companies or also national payers like in Canada and most of Europe, you would have national payers. And ultimately, all of us.

Most of us will sooner or later become a patient or have a patient maybe as a child, or as a spouse, or as an old mother or father. And then you want to look into these data and actually be able to make sense of it to decide whether a treatment is necessary or not. We have probably seen a lot about that during the COVID area, where everybody was talking about data and how to make sense of it. And that's where I'm really, really passionate, making sure that you have the right experiments, collect the data in the right way, and make meaningful conclusions out of it so that everybody can understand it and not just us data people.

Tensions between technical and executive work

So I think the biggest tension is kind of internally in us. We become data scientists because we're passionate about the technical aspects. Whenever you go to a conference and there's lots of technical sessions and there's maybe one session about communication, all the technical sessions are filled and so on. About communication, you see a couple of people.

I think we need to embrace all the communication aspects, become better in communication, even if that is not our comfort zone. We need to step outside of this comfort zone and embrace these other things so that we can actually work better with the technical stuff.

So you can have the best technical ideas. If you can't convince your non-technical management, they will not fly. Or if you are super technical, but nobody can effectively work with you, you'll not be effective. That's why I call it the effective statistician, or my book is also called How to Be an Effective Statistician. You can think about it like in a car. Your technical skills is your engine. That's all the power that you have. And your communication skills is everything around it, especially the wheels that bring it on the road. If you have a Porsche engine, but you have wheels like a bicycle, you'll not bring the power on the road. You can work on improving your technical skills so much, it will not help you move faster. And it would just be frustrating for you.

Your technical skills is your engine. That's all the power that you have. And your communication skills is everything around it, especially the wheels that bring it on the road. If you have a Porsche engine, but you have wheels like a bicycle, you'll not bring the power on the road.

Therefore, I think it's really, really important to work on these other skills. And I can tell you, as soon as I worked more on my communication skills, my influencing skills, my leadership skills, I was able to do much more fun things because I was also much more seen as a technical expert. My presentations, all the personal branding, all these things helped so much in that regard.

Increasing visibility and influence

So the first thing is you need for yourself to make a decision that you want to be more visible. And it takes courage to do so. Now, in order to have courage, you need to have a reason. Why do you want to be courageous? So for me, it's really important to understand, what do you want to personally achieve? So for me personally, what I talked earlier about is making the right decisions for patients based on data. It could be something completely different for you. But you need to know why you do these things so that when critique comes up, people kind of against you whatsoever, and you mentioned this kind of biased environment, when it's hard, you know why you're doing it.

The second thing is start with first listening. If you want to build trust with people, if you want to become a better leader, actually at all levels, especially the higher levels, start with working on your listening skills. Listen not like, you know, and listening is kind of, there's a big spectrum. There's this kind of listening like you're playing on your smartphone. And there's listening where you're really, really listening. You're 100% with the other person. You don't multitask. Maybe you take notes, but you see the other person. You have full screen or best is actually in person. And you really connect. You ask questions. You make follow-up questions. You have eye contact, all these different things that will help you to build trust.

And that will help you to make everything easier because the key thing is building trust. Now talking about visibility, there's of course visibility within your organization, but I think it's even easier to have visibility outside of your organization. And from a professional point of view, the number one place where you can build that is LinkedIn. So do two things. First, improve your LinkedIn profile. Make sure that it tells the story, that there's really something personal in it. Make sure that you know who you write that for. So that is a personal branding aspect. The second thing is regularly contribute on LinkedIn. It's so easy. Pick a topic that you're passionate about and then write about it at least once a week, at least once a week. And if you do that, you will increase your personal visibility. Now advanced, go to conferences. Present there. Join communities where you present as well. That will help you a lot.

By the way, Libby, one thing. Just post about something that you wish you would have known two years earlier.

Leadership, trust, and walking your talk

So I have actually trained people that own companies. I've trained all kinds of different levels within companies. My understanding is exactly the opposite. So in order to be effective, you need to be more humble, be more courageous, be more listening, yeah, be more trust-building. So the more higher up you are in the organization, the more exposed you are. And so everything you do will be critiqued by many more people. And then, despite that, doing the right things is even harder. So it takes more courage.

The higher you are up in the organization, the more people distrust you. People distrust you. Yeah. I don't know why, but that's across all the people that I talk to, the bigger your title, the less trust you have. I think this is what you need to do in that case, is you need to walk your talk. What a lot of people forget is that communication, the most powerful way of sending information and communicate is acting, is doing, is not talking or presenting. So how you behave and what you do when push comes to shove will show how much people will trust you. So whether people will go the extra mile for you depends on how you treat them when it's hard.

When you, as a leader, take the easy way and fire 20% of your people, well, good luck with managing the rest of the 80, the rest 80%.

Ethics and benefit-risk in data science

So as a data scientist or a statistician, I think it's really, really important that you understand the context in which you're working and have good business understanding. For me in the medical space, it means that you understand what are, for example, different side effects or what are consequences, what are outcomes? How important are certain outcomes versus other outcomes? So that you can actually make some kind of informed decision.

In medicine, you have always, in a sense, a sickle kind of problem. And that is a typical benefit-risk problem. No treatment is actually safe. There's always a risk with it. But we are willing to take the risk because we also get a benefit from it. And that's the so-called benefit-risk ratio. Now, how is that? And you need to have a, you know, a general understanding for how the benefit-risk ratio is and what is acceptable from the benefit-risk ratio so that you can make meaningful communication upwards.

Who will benefit and who will take the risk? And here as data scientists, we can kind of look into this and, for example, help and identify who are these high-risk people. Do they carry also most of the benefit or do it's actually the other way around? Do we have, you know, certain groups that carry most of the risk but get no benefit? Can we kind of exclude them from the intervention? Now, the interventions sometimes are not black and white. So, in medical treatments, for example, you could have different doses or different frequencies of application of a treatment. Can you do something similar? All of that is based in domain knowledge that you need to have. And then you can actually discuss that with your management so that they can also make informed decisions.

Managing feedback culture on a team

So, from a feedback perspective, everybody is encouraged to give feedback and to raise concerns. Now, there's basically two things you can raise concern about. One is concern about people and one is concern about things like processes, whatsoever, documents. And raising concerns about the latter is usually kind of the easy thing. Let's talk about raising concerns about people.

There's basically three forms of feedback, or three cultures. The first is, that's the worst one, no feedback whatsoever. Everything is kind of hidden somewhere and nobody really talks about it. Second is, there's feedback, but the feedback usually goes upwards. So, let's say Rachel is the team leader and then Isabel has a concern about Libby and goes with that to Rachel. That's kind of okay. But what I would do as Rachel, I would say, I would ask Isabel, have you already talked with Libby about it?

That is always the first thing. People should talk to those people that they have some kind of problem with whatsoever directly. Now, I give you the feedback formula that you can very, very easily use. It's three steps, super simple, and you can do it in two minutes. The feedback formula is, first, you ask for approval. So, Isabel goes to Libby and say, hey Libby, do you have time for feedback? Libby says, yes. Could also say, no. Can you come back in half an hour because then I'm done with my code? Okay. Libby says, yes. Isabel now states from her perspective what she has seen in terms of an action, a behavior. Libby, when you were five minutes late to the meeting, that's as objective as it can be. Or you can always say kind of from my viewpoint. And then you say, what was the impact? That delayed kind of our meeting start, and we anyway had only 30 minutes. And so, we needed to repeat a lot of things. And other people were, I saw that other people were annoyed. Okay. Stop. That's it. These three steps. Approval, observation, impact.

Now, the nice thing is you could do this feedback both for constructive and for supportive feedback. So, it can be something like what you just mentioned, constructive feedback, but you can also do it all the time for supportive feedback. Like, why that was a great presentation or why the LinkedIn post was good or why whatsoever. That is really, really good feedback culture. And so, if someone comes to you, you can train them on that feedback culture.

Now, there's one thing that I want to warn you about. If you look for feedback kind of help, there's very often people recommend a fourth step, and that's advice. Don't do that unless you are invited to. So, first, for positive feedback, you don't need to do it anyway. For constructive feedback, mostly as a person knows what to do anyway. And you, as a feedback provider, don't know what she or he can actually do best with it. And if you kind of impose your advice on the other person, yeah, that always ruins trust. So, don't, you know, jump on the advice monster unless you ask.

Navigating stakeholders who want predetermined conclusions

So, first is you need to establish trust. So, you need to make sure that they trust you, because if they don't trust you, nothing will change. In order to build that trust, you need to understand what they care about. What are their goals? What are their pain points? What are their fears? What are their hopes? Trust is basically three things that you need to have. The first is, and this is probably the most important, is you need to show that you care for the other person. The second is, so that's the care part, the second is you need to have, the other person needs to feel that you're competent. So, the other person, you can bring this across with your profile, with what you have done before, who you have helped before. And the last part is the character. And I think this is really important. You need to basically come across as a trustworthy person that wants to do the right things.

You need to help them understand what are the costs, what are the risks, and these kind of different things. And that you are not policing them, but helping them. So, in many pharma companies, the statisticians are, you know, behave like the P value police. Yeah. And that's not helpful. Yeah. That way, you exclude yourself, and you make yourself basically vulnerable for being outsourced. So, much better is understanding what is the overall risk for making a wrong decision, and clearly kind of understanding this risk and communicating this risk. But this is not something that you can do, you know, overnight. The whole kind of trust thing, as you know, trust takes time. So, I would not set my goals that, well, you will change the game within the next two weeks. Yeah. That takes a much kind of longer term effort.

The one skill AI can't replace

Interestingly, just this morning, I watched a short clip of the CEO of NVIDIA. He was asked about kind of the smartest person he ever worked with and, or he works with, or he knows. And he mentioned kind of, you know, all the kind of it was a little bit of an overstatement. I think that, you know, also coding skills, you know, become a commodity. I'm not sure that's completely the right thing because first you really need to understand what is needed.

What you can't automate are all our human to human interactions. It's really, really difficult. Yeah. Because our human to human interactions are built on trust. I mentioned that before. Yeah. We trust other humans. Machines here and there. But really, when it's about our usual workplace, it's about human interactions. So learning how to interact with others is one of the foundational thing. And asking good questions, yeah, is, I mentioned that earlier, is part of listening. Listening is not, you know, that you're just silent. Listening is also asking good questions. And you can only ask good questions, for example, if you have a good business understanding.

AI doesn't come with trust. And AI doesn't come with good business understanding. And these things, I think, is really, really difficult to integrate.

What you can't automate are all our human to human interactions. Because our human to human interactions are built on trust. AI doesn't come with trust. And AI doesn't come with good business understanding.

I want to throw in my own thought here, which is that something I've seen in community building over many years is that trust is built in repeated interactions over time. I don't know if anybody has heard of the marble jar concept, but like, you have this interaction with somebody, and they have like an imaginary marble jar, right? And they do something that garners your trust, and they kind of like get a marble, right? And over time, over interactions with these people, you kind of know who has a full marble jar and who doesn't, who has earned your trust. It's so hard to earn trust in just one interaction. It's like repeated measures over time.

Yeah, just one thing. The best tool for building trust are one-to-one meetings. Yeah. One-to-one where you can listen a lot, you know? So that's why every supervisor needs to have weekly one-to-ones with their direct reports. And if you don't have time for that, maybe you shouldn't be a supervisor.

Balancing employment and entrepreneurship

So I have set up my business in such a way that I can do a lot with minimal time investment. And that it aligns as much as possible with things that my employer actually stands for as well. So I started with my podcast. I was employed at Lilly and they were very positive in terms of contributing to the community. The same was later true with UCB. Usually, if you are working on things and publishing on things that are more helping the community rather than kind of promoting a specific product or service or things like that, I think most companies are usually kind of okay with that. So it's like, I don't know, if you would be super active in your local church, nobody would say something about this. So make sure that that works.

The second thing is I delegated a lot of things. So people will often ask me, how could I run a weekly podcast while having a full-time job? Well, most of the work I delegated. I have a team that sits in the Philippines that does also have a lifting in terms of podcast production. And then when I did all the leadership training, I set it up in such a way that it's also minimal effort or minimal time with maximum outcome. Having this constraint of not so much time actually makes you innovative and helps you to say, no, when I was full-time on myself, I actually did things that I shouldn't have done because I had too much time. Less time is not necessarily a problem. It actually is a very, very good filter for doing only the things that matter and applying the 20-80 rule.

When stakeholders trust AI more than experts

Yeah, I have seen someone that is more from the commercial side using AI to kind of think about products that we should offer. Not understanding what we are really offering and what is actually in the AI output. It's a little bit similar to the question from Frank before. You need to build trust with these different people so that they understand what are the consequences and that they by themselves understand that maybe that was not a good idea. So telling them that was BS is probably not the right thing. But more asking them about things and what do you mean by that? How will that kind of have an impact? Do you know what that means? It would mean kind of implementing these kind of different things. There's this risk. It takes that long. So more taking the person seriously.

Persons always act on their best interest and with their best tool and in what they think is the best thing for them. And they have very, very different tools and views and opportunities than you. So don't just dismiss it. That is their reality. And approach that with some grace and humbleness and see the person, not what they have said. I think this is really an important approach.

And approach that with some grace and humbleness and see the person, not what they have said.

Getting out of the weeds as a manager

Decline as many meetings as possible. Yeah, that's a short answer. People will figure things out themselves if you allow them to do things out. If you always come with your suggestions, with your advice, with these kind of things, they will not learn to do it for themselves and will always rely on you. Yeah. So delegate, delegate, delegate, delegate, delegate, delegate. Yeah. And you will, yeah, sometimes some things will go south. Yeah. And that is fine. Earlier in your career, you messed things up as well. Yeah. So let others make mistakes. Be there to help them, yeah, and coach them. But let them do these things by themselves.

You know, leaders speak last, said Nelson Mandela. Yeah. And I think the best is if leaders don't need to speak at all. So get out of the weeds. When people really need you, they will come to you. And even then, you know, ask them what they have done, what they have thought. So ask good questions, but don't give advice.

Fantastic. I love it. Well, we have reached the top of the hour, so we have to say goodbye. I don't want to. We could do another hour with you, Alexander. Thankfully, if you would like to hear more about Alexander, he's got a podcast, so you can go listen to The Effective Statistician and get more Alexander in your life. You can follow him on LinkedIn. This was such a wonderful time. Thank you so much for hanging out with us. I cannot wait to see you all next week as well.