Artificial intelligence will not have an automatically favorable outcome, nor a destiny already determined by model quality or the pace of investment. That is the message delivered by Brian Chesky at the Goldman Sachs Communacopia + Technology Conference, where the CEO of Airbnb described AI as a force to be actively addressed, so that people can benefit from it rather than be overwhelmed by it.
In his remarks to CNBC, Chesky avoided both unconditional enthusiasm and a purely catastrophic interpretation. His position is clearer on another point: technology gains meaning through the way it is used. To explain the dual nature of the phenomenon, he pointed to nuclear energy, capable of producing profoundly different benefits or harms depending on how it is utilized.
It is a relevant distinction at a time when AI is almost always discussed through two extremes: on one side, promises of productivity, automation, and new services; on the other, fears regarding jobs, power concentration, security, and loss of control. Chesky grounds the issue in less abstract territory. It is not enough to ask what generative systems will be capable of doing: we must determine what role people will retain in the processes those systems will transform.
User centrality in the cognitive revolution
The CEO of Airbnb speaks of a possible “cognitive revolution,” in which humans can become more capable alongside AI tools. The phrasing contains an essential condition: increasing capabilities does not necessarily equate to replacing skills. An assistant capable of summarizing documents, drafting text, or accelerating analysis can speed up work; on its own, it does not determine which information is reliable, which decisions are fair, or which goal is worth pursuing.
For Chesky, therefore, the existential risk associated with AI should not be reduced to a remote scenario or a clash between humans and machines. It can also emerge from the passive use of technology, when individuals and organizations delegate without retaining judgment, accountability, and the capacity to intervene. Hence the imagery of riding the wave rather than being swept away by it: AI is treated as a powerful shift, but not as an authority to which people should simply conform.
This approach highlights a practical problem for companies, workers, and institutions. Adoption is not merely a matter of buying models, integrating chatbots, or cutting the time required to complete a task. What matters is the design of the relationship between the person using a tool and what that tool can decide, suggest, or automate. Without that design, efficiency can translate into dependence on opaque systems or a gradual loss of in-house expertise.
Chesky's statement neither offers a regulatory recipe nor previews specific initiatives from Airbnb. For that very reason, it should be read as an invitation not to mistake a cultural framework for an operational plan. Saying that technology can empower people is an objective, not a guarantee. Making it concrete requires training, error assessment, the ability to contest outputs, and a clear allocation of responsibility whenever a system influences a decision.
Communacopia becomes an observatory on the power of AI
Chesky's remarks came during the tech-focused day of the annual Goldman Sachs conference, which began on Tuesday. Artificial intelligence was among the dominant topics at the event, which convenes executives and investors around major industrial and financial transformations. The significance of the theme is also reflected in the schedule: featured speakers include Jensen Huang, CEO of NVIDIA, Dara Khosrowshahi, CEO of Uber, and Bret Johnsen, CFO of SpaceX.
These figures represent different positions along the value chain. NVIDIA is central to the compute infrastructure underpinning the expansion of AI; Uber operates a platform where algorithms, logistics, and interactions with users and workers are already everyday fixtures; SpaceX brings the perspective of an aerospace company to the stage. The presence of these executives does not imply a shared consensus on AI applications, but it confirms how far the debate has now moved beyond the perimeter of model-building labs.
For the market, AI is simultaneously an infrastructure investment, a lever to transform products and processes, and a risk factor to manage. For society, the consequences are harder to capture in a single indicator: they affect the quality of information, the organization of work, the distribution of economic value, and people’s ability to understand software-assisted decisions. Chesky’s remarks fit into this second tier, without discounting the first.
The limits of metaphors and the test of application
The idea of “riding” AI is effective because it conveys the urgency of learning to manage rapid transformation. However, it has a limit: not everyone starts from the same position. The ability to use artificial intelligence as an amplifier depends on access to tools, available skills, one’s role within a company, and the rules set by those who control platforms, data, and infrastructure. A worker, a small business, and a major platform do not have the same leeway in deciding how to adopt it.
For this reason, highlighting the human contribution, as Chesky suggests, requires more than faith in individual adaptability. Within organizations, it means preventing automation from being evaluated solely on the basis of the number of tasks eliminated; it also means identifying where oversight remains essential, how errors are caught, and which skills must be preserved. At the product level, it requires interfaces and processes that make it clear when an answer is generated by a model and when the user can intervene.
Then there is the question of trust. A system can be useful without being infallible, but those who adopt it must know its limits: inaccurate answers, misinterpretations of context, and outputs requiring human oversight are not marginal details. If the promise is a collaboration between human and artificial capabilities, verification cannot be treated as an incidental cost. It is an integral part of the value AI delivers.
The debate that emerged at Communacopia does not settle these tensions. On the contrary, it shows that the vocabulary of transformation is changing: no longer just how quickly AI will be deployed, but under what conditions it can become a tool for growth for those who use it. Chesky does not express certainty that the outcome will be positive. However, he maintains that there is room for choice, and that this margin depends on the willingness not to hand over decisions and skills to technology without human oversight.
In upcoming sessions at the conference, the stances of leaders from companies involved in infrastructure, mobility, and space activities may offer insights into how this responsibility is translated into industrial strategies. For now, the contribution from the Airbnb CEO sets a useful benchmark for interpreting the sector's promises: AI must be measured not only by what it automates, but by the capacity it leaves for people to understand, choose, and correct.
