Leadership is Changing – Are You Co-Thinking with AI?

Great Expectations Podcast – Episode 6: Elisa Farri

Paul Sephton:

The first thing that I'd love to dive into and start with is really looking at this evolution of leadership in an age of AI. I kind of want to go into the psychological piece of it. I'd also to go into the practical application of it, but if we start off with the big premise that you have in your new book around AI being described as a co-thinker, I'd love to get your take on what you mean by co-thinker and how that differs from how other people might perceive, the usage of AI to be.

Elisa Farri:

So, co-thinking implies that every leader has access to a 24-7 thought partner who is always on, always available, and this partner augments the thinking of leaders. How? By framing perspectives, challenging assumptions, spotting overlooked spots, structuring the thought process in a more effective way. So, AI can be the sparring partner that never gets tired and is always on. We are seeing in our research that co-thinking with AI leads to better quality results, if and only if three ingredients are present. Context, challenge, and conversation. To clarify on this, imagine you are trying to solve a complex problem, you can either ask AI to give you the best solution to your problem, and you will be in a prompt and weight mindset. Or you shift it to a different way of interacting with AI and you start having a dialogue with AI. This is what we call co-thinking with AI, and as you do the co-analysis together with AI, you share contextual knowledge. You challenge AI answers, but vice versa. You also ask AI to challenge your proposed solutions, and you have a dialogue. You build comparatively on the answers and the inputs of AI. And this is exactly what Valentin Marguet is doing, in his work. Valentin is powertrain project leader at Ferrari, and he has been a heavy user of generative AI in co-analysis of technical problem solving. And so that this example sheds light on the opportunities for leaders of co-thinking with AI beyond using AI for productivity assistance.

Paul Sephton:

The productivity assistance that you tap into at the end there is the key difference because most initial use cases were really focused on output and this idea of, having a, not necessarily a sparring partner, but something that could do something for you, which was measured on the quality of the output and maybe the volume of the output at the same time. The augmentation that would take a portion of your work, for example, writing your emails for you or drawing up basic calendar management or word docs, and it would take that and manage that portion of the workload thinking is maybe a less tangible sense of output. How do you quantify that shift between output-based, productivity versus this conversational nature of working with AI?

Elisa Farri:

So, when we look at the output and the benefits of embracing a collaboration between human and AI, it’s not about speed. It's not about going faster, which is, let's say the main KPI used when we're looking at AI as a productivity enhancement, as a productivity assistant. When it comes to co-thinking, deeper reflection is about seeing what you're not seeing or even more powerfully seeing something through lenses that you do not typically wear. Generative AI is powerful at taking perspectives, and that's why we need, from a leadership standpoint, to be aware that these are two different ways of interacting with AI. They are equally powerful. Leaders can equally benefit from productivity enhancement and deeper reflection. And it's a muscle they need to train because technology can do both. So, we need more and more leaders equipped with the muscle for human-AI collaboration, being knowledgeable and aware of when it makes sense to shift gears, to go from productive assistance to co-thinking and vice versa. Because remember, leadership tasks, the managerial tasks do not happen back home. And it's an interplay, intertwined, done in sequence. And once again, we are at the point of a new managerial capability. We need leaders to be equipped with this muscle and be aware of what technology can do in order to see this collaboration for, let's say not creating business results, but also for ensuring a responsible and ethical use of such a powerful technology.

Paul Sephton:

We saw this massive rise in collaboration and a lot of cognitive fatigue from one too many meetings, but traditionally leaders have relied on sparring with other people. How would you advise people to know when is the right time to perhaps spar with AI and co think with AI versus calling your peers or your team into a meeting and trying to have human to human collaboration?

Elisa Farri:

Thank you for this question because it sheds light on an aspect that is often overlooked, and this aspect is about the setting in which you're using generative AI. So, let's start from the traditional and most widely used setting, the one-on-one. This is a technology that runs on devices, and it automatically attracts you as a magnet in a one-on-one. But then as you said, the vast majority of tasks and activities and decisions that are typically made in large organizations are made in a very different setting. And this is the human to human and team setting. So, the answer to your question is that there could be various ways in which you inject AI into a team setting We are seeing, for instance, leaders who are leveraging AI before a team activity, which could be a workshop or ideation session. It could be a strategy session. And they use it in preparation for team, discussions. But we are also seeing other managers who are injecting AI during the team conversation, beyond note taking. You can ask the generative AI power chatbot to contribute to the conversation. I give you a practical example during a team meeting, which is happening online. You can ask the chatbot if there are any items that you haven't covered yet, and which are supposed to be part of the conversation as per the predefined agenda. Or you can ask the chatbot to build on the discussion to further articulate and elaborate what has been discussed so far, and even to take a specific perspective, for instance, of the client you're discussing, to give you additional insights, different lenses through which you can look at the topic that you're currently discussing. Yes, AI can be used beyond the one-on-one setting. And the main reason why we strongly recommend leaders to inject AI into a team setting is that based on our research last year, we found that leveraging the sense making of the team is also a powerful antidote to some of the most common traps that might be encountered when using AI. And leveraging team judgment, not for better use of this technology, but also for anticipating and mitigating potential risks related to this technology.

Paul Sephton:

When it comes to leaders adopting and leaning into this, is there anything that you've noticed from a trends perspective on which companies or organizations are particularly good at it or what is potentially holding other organizations back? Because the higher up you go in leadership, the more often you would find people who are there because of their competencies and who might be a little bit hesitant or cautious to either fully trust in AI systems or admit that they're relying on or engaging with, AI at such a high level of codependency, rather than it being seen as a positive trait of, having a secret weapon.

Elisa Farri:

So, what's the secret sauce for widespread adoption? Accelerated adoption of this technology in a large organization? The answer is simple, and people tend to be very surprised. What we found in our research is that one of the key accelerators is indeed leadership. When leaders do not talk about the importance of AI but show that they are using AI in first person. This activates a contagious wave inside the organization, and then it percolates across the various layers. If we want to truly accelerate adoption it starts with and from the leadership, we need custom training for leaders. And that's why also we wrote a book on this because the common use cases are unfit for the traditional leadership and executives’ role. And so, we need to tailor the use of this technology to the needs and the goals of a typical leader's job. Second, we need leaders who role model, who walk the talk when it comes to the use of this technology, and this is easy to do. This is not rocket science. With individual use cases, how you are leveraging it for productivity assistance, but also how you leverage it as a thought partner. It's about sending emails or also structured, let's say more structured than formal newsletters, with your learnings. What you're learning as a leader along your journey is about actively scouting for the mavericks, for the gen users inside the organization and leverage them for your personal reverse mentoring and for your personal learning journey. Yes, leadership can be the accelerator when it shifts from just saying and announcing that AI is imperative inside the organization to really embracing it as a force of transformation on a daily basis.

Paul Sephton:

And what do you think we should be expecting for the next 12 to 24 months in terms of productive output? Because all of this requires much learning and change management, the expectation has always been kind of underpinned by productive outputs. And how do we balance the expectation of the gain that we will get potentially long term versus the short term sacrifices we need to make to be able to learn these new technologies, socialize them, and adopt them across an organization.

Elisa Farri:

Personally, I feel that the key word is learning. We need to upscale and train the entire workforce in order to prepare the workforce for what will come next. And why I'm saying this is because the vast majority of leaders still look at this technology from an outside perspective, as spectators. We need to avoid this spectator view, but on the other hand, we need to reflect. Think how this technology will disrupt the way in which work is performed. The real challenge is that we need to do all of this at the same time because technology's evolving. Adoption is increasing yet slowly, not in all industries and organizations. And at the same time, we need to have a long-term view in order to anticipate what will be the impact on the workforce. Once again, these boil down to the role of leaders and managers across the entire organization. They need to activate an ongoing communication channel with their team, which is not about how they're using AI today, but also how they are envisioning this as a disruptive force in how work is performed. This is a challenge that we are seeing, widely debated compared for instance to one year ago, one year and a half ago, where the main focus was on the technology. Now, the big chunk of the debate at leadership level is indeed about the impact and about how to prepare the organization for unlocking the long-term value while balancing the short-term impact, which, as you said, can be positive, but also in some cases have some negative impact, especially on, certain jobs and tasks.

Paul Sephton:

So one of the interesting things that you touched on earlier was that set of skills or training that's required for leadership to really lean into co-thinking with AI is different to perhaps the broader set of advice given on how to use these tools. If you have someone listening to this, sitting in their office at home going “yes, I'm a hundred percent on board. I want to do this. How do I lean into it?” And they've got a prompt window up on their screen with whichever service they're using. How do you guide that initial jump into the ocean, in terms of the right way to go about using these tools as compared with what you might find elsewhere or as a directive more broadly for a workforce as a leader?

Elisa Farri:

First suggestion would be to go back to the most recent interactions that they had with their generative AI model and analyze the level of contribution of technology compared to their level of contribution. Why I'm saying this because in our master classes, when we train leaders, we, often realize that the level of contribution of the human, in this case, of the leader, is very, very limited. It's 90% contributed by AI and 10% contributed by the leader. And this sheds light on the fact that this is highly imbalanced. And when leaders are a bit upset, or they complain about the poor quality of the answers that they get from AI this is mainly because they're not interacting in the proper way. They're not sharing feedback, they are not building on AI answers. They are not having a real conversation. The second is that most managers have what we call the blank prompt syndrome. And this could be because they don't know what a good question could be, what a good prompt could be. And here our suggestion is very simple. Why don't you ask AI directly, to give you, for instance, 10 potential questions, good questions that you can ask, AI and then ask for another 10. As I said at the beginning, this is a thought partner that never gets tired, leverage it. And the third recommendation is. Challenge, challenge, challenge also in a human to human setting. We know that beauty unlocks when we challenge each other in a constructive way. This is more than important with AI. Why? Because AI tends to be very, very complacent. These bots are designed in a way that they tend to please you. They will say that, oh, Elisa, you had a very great idea. Oh, Elisa, you wrote a very great email. And our, tip for leaders is try out a simple prompt this. Please give me your brutally candid feedback, and you will see how the answer of AI will flip. These are three very practical recommendations that we're finding very useful to open the eyes of leaders and embark them on a journey of co-thinking with AI.

Paul Sephton:

It’s a fascinating journey of adoption and one that will rely heavily on trust. I'm curious to get your take on how do we balance trust with leaning into something where we maybe exercise some caution or fact check something or read through it versus leaning into a sparring partner and co-thinker who we are fully engaging in a high-trust relationship with off the bat?

Elisa Farri:

Judgment is imperative in every single interaction with AI. We need to ensure that our sensor of judgment is always switched on. That's our main, recommendation. And Paul, we are seeing judgment in action. Right now. We are having a conversation. You are asking question. I am thinking about the best answer to convince and pitch you my thesis. While you're listening to me, you're trying to validate and to challenge what I'm saying or to connect with other interviews that you might have or based on your experience and so on and so forth. The first antidote to the trust trap, in my opinion, is embracing critical thinking. Because if we have more and more people talking with AI. They will automatically tend to switch their judgment sensor on. Why? Because when you simply click and wait, prompt and wait for some magic to happen, the tendency is to switch it off. I give you a very practical example. When you use the calculator, you switch off your judgment because you blindly trust the calculator. In this case, we don't want leaders and people inside the organization to switch off their judgment sensor. The second recommendation is that, and we already touched briefly upon this, the fact that this is a technology that attracts you as a magnet in a one-on-one conversation. When we try to inject AI in a team setting, we see that the team sense making, the way in which judgment is exercised in a team setting is another powerful antidote to the trust trap. Why? Because you have different expertise, you have different point of views, different backgrounds, and plus, there's also this. Mandate at a team level that you want to reach a good quality output. In the age of AI, we need more judgment, not just as an individual, but also at a team level in order to deal with trust the traps.

Paul Sephton:

I think the way that you balance the team setting with the individual setting is one that I'm particularly keen to double click on. And the reason for this is if we look into some of the predictions that we've seen around gen AI and then agentic AI, it's looking like one scenario or outcome might be that we go from having individual agents for knowledge workers to ending up with people leading teams full of agents where you have one human leading a team of agents and basically becoming something of a fact checker and prompter. And that's for better output. Maybe that's for less work. If you ask the Americans, they might say it was for increased productivity. They probably think that the Europeans want longer holidays. Regardless of the outcome though, if we're talking about the role of leadership and how it's evolving, and we're talking about the role of teams and how they will evolve, how do you see, the role of leaders evolving and the role of teams evolving as AI becomes more embedded into our daily workplace?

Elisa Farri:

Another powerful question. We don't know yet how the distribution of tasks will happen inside the organization, but what we do know is that there will be three main categories. The first category is a set of tasks that you will automate, AI will perform such tasks. It could be in the form of agents, it could be in, the form of other technologies, applied to task execution. But this will be totally delegated to AI. Then there will be other tasks on the other extremes that will stay under the remit of humans. But the vast majority of tasks, we should expect them to fall into the third category; a blend of human-AI interaction. When it comes to the role of leaders, we need leaders who know how to control and oversee tasks that are delegated to AI. We need leaders who keep doing what they've been doing, overseeing tasks and decisions made by humans. But the real challenge will be how to equip leaders in overseeing a hybrid the way of working where humans interplay and interact with AI. This is the real challenge. You mentioned agentic AI. This is one of the driving forces for rethinking how work is performed inside the organizations and when it comes to this capability and skills. This is not about being aware and knowledgeable about what this technology can do for us, but we also need to counterbalance this with knowledge about how human should be interacting with AI. The fascinating aspect is that we are living in an era where the way in which work has been performed over the last centuries might be totally disrupted. And once again, we need leadership who is not a spectator. We need a leadership that is a driving force and a role model for embracing this human AI collaborative way of working.

Paul Sephton:

And Elisa to kind of try and future proof a team, what are some of the mistakes or common pitfalls that you see leaders making today as they try and into the technology or resist it, but where you think that they're not necessarily taking the right steps or what are the things to avoid?

Elisa Farri:

The first thing to avoid is to look at this technology only from the perspective of productivity assistance. There's much more beyond that. Here is the opportunity of embracing co-thinking. When the level of complexity increases, we always recommend shifting from productivity assistance to having a conversation with AI. This, I would say is the first thing to avoid. The second mistake is when leaders apply this technology as a layer for efficiency, for productivity. And this is pretty different from the first point because this entails a vision. You need to look at it as a pervasive and fully transformative force to rethink what have been doing. And the second recommendation is to ask themselves. How can AI help us do a certain thing differently? And this is typically the springboard prompt that executives and leaders give to trigger a first human-AI conversation. Ask AI to provide a novel view and also to inject an element of divergent thinking when you look at a specific task that you would like to do. Otherwise, the risk is that you will be trapped in the old way of doing things, adding the layer of AI on top. The real differentiator in the future will not be the layer that, let's be honest, it'll be soon a commodity because every organization will be applying the same layer, but it'll be in the transformative power, and here once again, becomes central the human. The real power is to leverage human creativity, the human thinking, the human expertise. Augment this with AI and transform how things are doing inside the organization.

Paul Sephton:

And I'm curious to get your take on how that applies in a global setting because we know the concept of a wealth gap growing in the past few generations, and there's quite a bit of speculation that there'll be a similar AI gap where future generations in the workforce will either have AI smarts or significantly lack them. Is there anything that you're seeing right now in global organizations in terms of different teams being more or less, resistant to adopting the technology? It could be down to different cultural settings. It could be down to different company sizes. It could be down to budgets and who has the budget to be able to deploy certain AI technologies or maybe sectors where there are some more regulated sectors. How do you see all of these factors playing out and then impacting the overall landscape of usage and familiarity or competence in AI?

Elisa Farri:

We need to differentiate between exogenous and endogenous factors because there are exogenous factors that apply. Let's say globally to all organizations and notwithstanding the size of the industry, and you mentioned some of them, regulation is indeed an exogenous factor that you do not control and you have to comply with. But then there are some, pretty, endogenous ones. The ones that are pretty peculiar of the specific organization at hand. And in this case, what we are seeing is that culture is the key differentiating factor when it comes to the adoption of AI and the ability to unlock benefits and potential of AI. We should never forget the human element. And the human element is about, more soft skills and more hard skills, a mix of mindset, culture, habits, behaviors, and never forget that culture is shaped on a daily basis. And leaders have an important role in shaping this aspect. Once again, we go back to the role of the leaders. We go back to the fact that they need to role model not the use of AI, but also some key behaviors that created a psychological safety inside the organization for embracing such a disruptive, change. When we see that leadership is actively using and not talking about AI. And when we see that there's the right, culture with the right ingredient that triggers an innovative experimental mindset, still very disciplined, and structured, we are laying the foundations for an organizations that can thrive, in the age of AI.

Paul Sephton:

Really, really interesting Elisa. Now, if I were to close up with one question, it would probably be the advice or closing words. We've covered much practical advice and tangible takeaways for leaders, but what do you think the one key ask or takeaway that you'd give to leadership is in terms of how to truly innovate and lean into this change to be able to yield positive results, in 1, 3, 5, 10 years time.

Elisa Farri:

First of all, be the accelerator. Remember that every day you can either accelerate or slow down this process. Feel this as your own responsibility. Be accountable. For the second, activate the communication channel within your team. Ask at least once a day to your team or to one of your team members, have you used generative AI today? What did you do? Was it helpful? What was the main benefit? What did you learn? And do this also from a selfish standpoint. It's your daily opportunity to learn from your peers, from your team members, how they are using it. And the third point is about judgment. Do not limit to, exercise your role as leader, as going through a checklist of do's and don'ts when it comes to the use of this technology. Create the right environment for your people, for your organization to share and learn on an ongoing basis what is working, what is not working, what are risks or traps that make them feel uncomfortable about using this technology? That is triggered from the top down, but you should never forget about the human dimension. And listen to your people and ask them how they are living, this journey also from an emotional standpoint.

Paul Sephton:

Thank you much for the time today, at least. I really appreciate you joining us on the show.

Elisa Farri:

Thank you, Paul, for having me...