Inteligencia Artificial

AI Leadership: why executives need social listening

Business leadership analyzing AI-powered social listening insights and digital customer conversations.

Discover how AI is transforming leadership and why social listening helps executives understand customer sentiment, protect brand reputation and make smarter decisions.

What a decade-old leadership study and Klarna’s AI reversal teach executives about listening in the age of generative AI.

In February 2024, Klarna announced that its AI assistant was doing the work of 700 customer service agents. It had handled 2.3 million conversations, two-thirds of all customer chats, in its first month (Entrepreneur). The company froze hiring, and its CEO, Sebastian Siemiatkowski, became one of the loudest voices claiming AI could do almost any job.

By May 2025, Klarna was hiring humans again. Siemiatkowski told Bloomberg that the cost-driven push had led to «lower quality» service (CX Dive). By September 2025, the company was reportedly moving engineers, marketers and analysts into customer support (Fluent Support, citing Business Insider).

It is tempting to read this as a story about technology. It is really a story about leadership. And a 2014 article in The Leadership Quarterly explains it better than most of the commentary written since.

Leadership has a source and a transmission

In «E-leadership: Re-examining transformations in leadership source and transmission,» Bruce Avolio and colleagues reviewed what a decade of digital technology had done to leadership. Their core idea is simple. Every act of leadership has a source, someone or something that originates influence, and a transmission, the channel through which that influence reaches people.

For most of management history, the source was the leader and the transmission was the leader’s own behavior: a speech, a memo, a meeting. Technology broke that tidy picture. The authors argued that leadership and technology shape each other over time, and that influence now also originates with followers, with groups, and with the wider context around the organization.

They also made an uncomfortable observation. Leadership research, they wrote, had taken an «anthropological» approach to technology: it studied the traces technology left behind after the fact, instead of anticipating its effects. Most companies still manage AI the same way. They find out what happened once the damage is visible.

Read through this lens, the Klarna case offers four lessons.

Lesson 1: Leaders decide what AI is for

The article draws on Adaptive Structuration Theory, which holds that technology is never simply «used.» People appropriate it according to how they interpret its purpose, or what the theory calls its «spirit.» Leaders shape that interpretation in three ways: through the vision they communicate, through what they legitimize by their own example, and through the mandates and policies they impose.

Klarna’s leadership did all three. The vision was AI as a replacement for people. The CEO legitimized it publicly and repeatedly. And the hiring freeze turned it into policy. The spirit of the technology inside Klarna became cost-cutting, not better service. The AI performed exactly the role leadership assigned it.

The lesson for executives is that an AI rollout is a leadership message before it is a technology project. Whatever you say AI is for, your organization will make it so.

Lesson 2: Your message travels without you

Avolio and colleagues described how a leader’s message, once sent through digital channels, gets received, reinterpreted and passed on by thousands of people. It can spread or be neutralized depending on whether audiences believe it is authentic. Their example from 2006 still lands: when an internal email from Cerner’s CEO criticizing employees’ work ethic was forwarded and leaked, the company’s stock fell 29%.

Klarna’s statements followed the same path at a far larger scale. A claim made to impress investors was also heard by customers, employees, job seekers and journalists. Each group interpreted it through its own experience. A customer struggling to reach a human was unlikely to hear «innovation.» More likely, they heard «we cut you off to save money.»

Today, every executive statement about AI is transmitted this way. The leader controls the source. The leader no longer controls the transmission.

Lesson 3: Customers now share the steering wheel

One of the article’s most important shifts is that the «locus» of leadership has moved. Technology empowered customers and followers to give transparent, often instant feedback, and to organize around it. The authors pointed to review sites that grade products, and to Glassdoor, where entire companies are judged by their employees.

At Klarna, the course correction did not originate in a strategy meeting. It followed reports of customer dissatisfaction and declining service quality (tech.co). In practice, customers exercised leadership: their collective reaction changed the direction of a multibillion-dollar company.

For marketing leaders, this reframes the role of customer voice. It is not only an input for campaigns. It is a force that can redirect strategy, whether the executive team is paying attention or not.

Lesson 4: Efficiency is not empathy

In 2014, the authors considered a provocative idea from robotics researchers: machines might take on leadership and service roles precisely because they are not influenced by emotion. The authors were skeptical. They cited critics who argued that machines could not yet exercise moral judgment, read unfolding situations or accumulate experience the way people do. They also warned that people might follow automated directives without weighing ethical and human considerations.

Klarna arrived at the same conclusion through experience. Siemiatkowski said the company learned that most of the time, people would rather speak to other people (tech.co). The company now describes a hybrid model in which AI provides speed and people provide empathy (CX Dive).

The point is not that AI failed. The point is that leadership chose to use it where human judgment mattered most, and only found out when customers said so.

Social listening: the missing feedback loop

If leadership is now transmitted through channels leaders don’t control, leaders need a way to see how it is being received. That is the strategic role of social listening: the structured monitoring and interpretation of public conversations about a company, its leaders, its products and its decisions.

We don’t know how Klarna monitored public reaction, and we won’t speculate. But the signals were public: customer complaints, social media debate, press coverage and, eventually, a widely shared label for the pattern, the «Klarna effect» (Fluent Support). The question for every executive is how early their organization would have heard those signals, and whether anyone would have connected them to a leadership decision.

Generative AI changes this picture in two opposite directions.

It makes listening possible at the speed of transmission. Until recently, reading thousands of comments meant keyword counts and crude sentiment scores. Large language models can read unstructured conversation at scale, summarize the narratives forming around a decision and flag shifts in tone within hours. This is how leaders can move from the «anthropological» approach Avolio and colleagues criticized, studying traces after the fact, toward something close to real time.

It makes what you hear harder to trust. The authors asked whether followers would always know they were interacting with an authentic leader, rather than someone falsely representing one. In 2014, that was speculative. Today, synthetic content, AI-written posts and deepfakes of executives are operational risks. Listening is no longer only about what people say about you. It is also about detecting what is being said as you, and separating organic reaction from manufactured noise.

Listening is not surveillance

The same article that makes the case for listening also draws a clear warning line. The authors cited companies monitoring employees’ social media to gauge whether leadership messages were landing, and a tool designed to flag «disgruntled» employees from their emails and posts. They raised ethical concerns about acting on such predictions, and noted that people who feel watched may simply mask their real reactions.

They also cited the «transparency paradox» identified by Harvard researcher Ethan Bernstein: constant observation can reduce performance, because people invest effort in hiding what they do rather than improving it.

For leaders, the distinction matters. Responsible social listening looks at public, aggregated narratives: how the market, customers and the public are interpreting a decision. It does not profile individuals, and it does not turn employee expression into a disciplinary signal. Under privacy laws such as Brazil’s LGPD and Europe’s GDPR, this is also a compliance question. Done badly, listening destroys the very trust it was meant to protect.

What leaders and marketers should do now

  1. Treat every AI announcement as a leadership message. Before going public, ask how customers, employees and candidates will each interpret it, not just investors.
  2. Listen to decisions, not only to brands. Most monitoring tracks brand mentions and campaigns. Tie listening to specific leadership decisions, such as an automation rollout, a pricing change or a restructuring, so you can see how each one is received.
  3. Define the signals that would make you change course. Klarna’s reversal came more than a year after its announcement. Agree in advance on which shifts in customer narrative would trigger a review, so evidence reaches decision-makers before the damage compounds.
  4. Put a human between the signal and the decision. Use AI to read at scale, but keep interpretation with people who understand context, culture and ethics.
  5. Verify before you react. In a world of synthetic content, confirm that a backlash is organic and that a quote attributed to your leaders is real.
  6. Set your ethical boundaries in writing. Listen to public conversation, not to private lives, and say so openly.

The sun is still rising

Avolio and colleagues closed their review by saying the field of e-leadership was still at dawn, with much left to learn. More than a decade later, generative AI has made their questions urgent rather than academic.

Klarna’s story is not a warning against AI. It is a warning against leading AI without listening. Leaders can still choose the source of their influence. What they can no longer do is assume they know how it is being received.

References

  • Avolio, B. J., Sosik, J. J., Kahai, S. S., & Baker, B. (2014). E-leadership: Re-examining transformations in leadership source and transmission. The Leadership Quarterly, 25(1), 105–131. doi:10.1016/j.leaqua.2013.11.003
  • Bernstein, E. (2012). The transparency paradox. Administrative Science Quarterly, 57(2), 181–216.
  • Klarna is hiring customer service agents after AI couldn’t cut it on calls, Entrepreneur, May 2025.
  • Klarna reverses AI customer service overhaul, tech.co, May 20, 2025.
  • Klarna reinvests in human talent for customer service, CX Dive, 2025.
  • Klarna AI demoted, Fluent Support (citing Business Insider).

Suscríbete a nuestro boletín

Recibe consejos y novedades de marketing directamente en tu correo.

Síganos

Mantente conectado y sigue nuestras últimas perspectivas, actualizaciones y tendencias del sector.

¿Listo para convertir la inteligencia en una ventaja estratégica?

Descubre cómo Loxias AI ayuda a organizaciones como la tuya.