The Future of AI Is Human-Led

January 25, 2026

Pamela Woitschach

Originally published on: https://www.linkedin.com/pulse/future-ai-human-led-pamela-woitschach-phd-mba-ckqsc/?trackingId=grSsBfsqR7GWtfQBlUF9RA%3D%3D

Before my master's and PhD, I studied my undergraduate degree in Clinical Psychology in the Faculty of Philosophy, and at that time, I never imagined that one day I would not only return to those foundational ideas, but also apply them to a technology capable of changing the course of the world.

As I have noted in other articles, I trust that developers will continue to create technology that wows humanity, so I will not attempt to train myself in the technical details or discuss them in depth.

What I can discuss, however, is the intersection of psychology, philosophy, data science, measurement and evaluation, and business. This series of articles I have been focusing on is about human-led AI transformation.

The Leadership Deficit at the Heart of Artificial Intelligence (AI)’s Global Moment, and Which 2050 Do We Want? Reflections on Leadership in the AI Era from Davos' opening session

As AI (Artificial Intelligence) becomes embedded across every aspect of daily life, from how we communicate and work to how services are delivered, society is struggling with how to remain in control of systems we did not evolve alongside. In response to this uncertainty, frameworks such as human-in-the-loop and human-oversight have emerged as attempts to manage risk in uncertain times.

Yet AI is now present everywhere, from household devices to vehicles and communication platforms. The scale of its reach demands a deeper leadership model.

This article reflects on insights I gathered from the following #WEF26 sessions:

• How Can We Cooperate in a More Contested World? Town Hall: Dilemmas around Ethics in AI. Speakers: Max Tegmark, Meredith Whittaker, Rachel Botsman, Mat Honan.

• How can we deploy innovation at scale and responsibly? Next Phase of Intelligence. Speakers: Nicholas Thompson, Eric Xing, Yoshua Bengio, Yuval Noah Harari, Yejin Choi

• How can we deploy innovation at scale and responsibly? Scaling AI: Now Comes the Hard Part. Speakers: Roy Jakobs, Ryan McInerney, Mat Honan, Amin Nasser, Julie Sweet

• Open Forum: Which 2050 Do We Want? Speakers: Arjun Prakash, Aruoture Oddiri, Zainab Azizi, Taylor Hawkins, Alois Zwinggi, Agnes Callamard, Adam Tooze

I will open this by defining words that came from Meredith Whittaker, President of the Signal Foundation, who articulated the stakes of AI integration with such clarity and impact at the How Can We Cooperate in a More Contested World session. Meredith said something along the lines of: 

“Signal is used by militaries, governments, boardrooms, and dissidents, including extensively in Ukraine, because private communication can be a matter of life and death. Signal is built on top of major operating systems, including Android, iOS, Windows, and macOS, and we take that responsibility seriously. All code and cryptographic protocols are open source and increasingly subject to formal verification, allowing mathematical proof that the software does exactly what it claims to do. This level of assurance is necessary because failures at this layer can have real-world consequences. At the same time, Signal ultimately depends on the security guarantees of the operating systems it runs on.

That foundation is now being challenged by the rapid integration of AI agents into operating systems. While marketed as convenient, these agents require broad access to calendars, files, screens, microphones, browsers, messages, and cloud processing to function. In practice, they operate with permissions similar to advanced malware, are highly susceptible to prompt injection, and cannot reliably distinguish authentic user intent from manipulation. This marks a fundamental shift in computing, where operating systems are no longer neutral platforms but are increasingly controlled by agent providers, reducing both developer autonomy and user agency and posing a serious threat to privacy-preserving applications like Signal.”

Okay, now let’s breathe, because that was a heavy one. : )

Meredith exemplifies why I wholeheartedly believe leadership must evolve into a model that includes aspects that allow leaders to care not only for productivity metrics but for present and future generations globally more than ever before.

Too often, discussions frame humans as passive participants within AI systems. This framing seems misleading to me. The future is not about humans just being in the loop or on the loop. It is about humans in the lead, as Julie Sweet pointed out. Our leaders and members of society must be active participants.

Human-led AI means humans set strategy, define values, establish guardrails, and retain final authority. AI remains a tool, not an autonomous decision maker.

We use the tools to amplify efficiency while remaining in control.

AI as a tool should be no different. The challenge is not whether AI should lead or follow, but whether humans are willing to remain accountable for what these systems include and can do.

This perspective was echoed by leaders from the session Scaling AI: Now Comes the Hard Part, and the Open Forum: Which 2050 Do We Want? who emphasized that systems are representations of human values. What AI reflects depends on what humans choose to encode, measure, reward, and restrict.

Human-led design means deciding intentionally what systems should represent and enforcing accountability over time, recognizing that governance is not a one-time exercise but an ongoing responsibility.

The next panel I would like to mention is, How can we deploy innovation at scale and responsibly? Next Phase of Intelligence, among topics on what is next on AI development, raised a critical question:

 Should AI agents behave like humans?

The consensus was no. And while there are aspects of human intelligence that could be required for AI future enhancements, human and AI intelligence are on different trajectories.

AI excels at specialized tasks, pattern recognition, and speed. AI does not understand human relationships, cultural nuances, or moral ambiguity. It does not possess philosophical curiosity or lived experience. These gaps are not flaws to eliminate but boundaries to value.

The path forward is not to replicate the human mind but to design systems that augment human judgment. I really liked the presenter’s point of view on the uncertainty we are currently living in. And the need for us to be humble.

There is no real precedent for what we are experiencing. That uncertainty makes leadership even more critical.

The AI-era leadership model

The future of AI will not be determined by technical capability alone. It will be determined by whether leaders are willing to stay accountable, exercise restraint, and insist that human values remain at the center of intelligent systems. Ethical leadership, moral courage, and purpose-driven decision-making are equally critical.

As the Signal example shows, delegating decisions to algorithms does not remove human responsibility. Leaders must be willing to name risks clearly, slow innovation when needed, and remain accountable for long-term outcomes through clear metrics and measurement.

However, even though I continue to believe leaders have a moral responsibility, I also think of myself as part of society and believe that we must be active drivers and guardians of the present and future generations.

PW.

Disclaimer: The views expressed in my posts and articles are my own and do not represent the official positions or policies of any institution or organization with which I am or have been affiliated. All content is shared in a personal capacity.