What Are We Still Traveling For?

AI will not abolish professional conferences. But it removes their most convenient justification: we no longer need to fly around the world for information that can be summarized, transmitted or generated on demand just as well.

AI Takes the Stage

At the ACSE conference a few days ago, I experienced an AI as conference host for the first time. It announced program items, created transitions and gave the event a professional framework. It felt less spectacular than one might expect—and that was precisely what made it interesting. What had been automated was no minor detail, but a substantial part of the conference’s visible production. An innovative idea from the organizers!

An AI can explain schedules, pronounce names correctly, introduce contributions in multiple languages, cluster questions from the audience and provide a precise summary at the end of the day. It can recommend a personalized agenda and open up missed sessions in a matter of minutes. That makes the conference more efficient. But it also provokes an uncomfortable question: if AI takes over the flow of information, why do we still need the physical event?

Three Functions, Only One Is Scarce

Professional conferences have traditionally sold three things as a package: information, orientation and human connection. AI is increasingly capable of providing the first two. It can turn ten presentations into a reliable update, compare positions and flag contradictions. It can create a presentation from a paper in an instant. Early experiments with semi-autonomous scientific conferences already combine AI-assisted manuscripts, peer review, presentations and archiving.

That does not mean machines are better scientists or speakers. It means only that the pure informational value of many presentations is becoming easier to replace. We know the pattern in scholarly publishing: the same topics, similar slides and often the same people on different stages. Not every twenty-minute market overview justifies a flight with its negative carbon footprint, a hotel room for a few nights and usually several working days. If a contribution can be replaced by a carefully verified AI update without any meaningful loss, perhaps it no longer belongs in the main program.

The third function remains scarce: human connection. Trust does not emerge from a summary. A new idea often reveals its value only through disagreement, a collaboration begins over coffee, and a hiring decision may follow a conversation that appeared on no agenda. Especially for younger colleagues, visibility is not the same as availability on demand. Merely digitizing conferences fails to do justice to this social value. But do we really need eight to ten conferences a year?

Nor is every encounter automatically valuable. A conference circuit that invites the same voices year after year mainly reproduces its own hierarchy. It creates familiarity, but not necessarily insight. AI could even help here—not by choosing people, but by making visible which perspectives are missing, which claims keep recurring and which questions remain unanswered in the program. Editorial decisions must remain with people; the information on which they are based can improve.

Travel Has to Earn Its Place Again

That is precisely why we should take a more critical view of travel. A study of a three-day international scientific conference calculated an average of about 0.57 tonnes of CO₂ equivalent per participant, with travel accounting for the largest share. Another study found that virtual conferences can reduce the carbon footprint by 94 percent. These figures are not a moral footnote. They shift the burden of proof.

The right question is no longer: could this event also take place online? It is: what should happen on site that cannot happen online, in a recording or through a prompt? There may be a convincing answer—but it cannot consist of a series of interchangeable presentations. The era of compulsory online or hybrid formats is over for good. We need something new.

Attendance will therefore probably become more selective. Fewer people will travel for updates alone; those who do attend will expect more than a stage. Hybrid and distributed formats can extend reach without forcing every relationship into a video window. Research on virtual conferences shows both sides: they broaden access and reduce emissions, yet many participants miss informal networking and community. The answer is not a return to the old format. It is a better division of labor.

Information to the Machine, Presence to People

In future, an AI briefing before a conference could establish a shared baseline of knowledge. The stage would then be reserved for argument, genuine opposing positions, workshops, negotiations and decisions. Instead of five keynotes on familiar trends, there would be a precise synthesis followed by a debate about what it means. Early-career professionals would receive curated introductions and mentoring. Free time would not be a gap in the program, but its most valuable component.

The AI host would also have a useful role: as translator, navigator, timekeeper and the event’s memory. But not as an unchecked authority. Anyone who selects questions, formulates summaries and distributes visibility exercises power. Models can invent errors, smooth over dissent, store confidential conversations or reproduce the interests of their operators. Their use must be disclosed, supervised and editorially accountable. AI can assist with moderation; it cannot assume responsibility.

This also changes the economics for organizers. Anyone selling tickets mainly through a long speaker list and as many sessions as possible will increasingly be offering something that is available digitally at almost zero marginal cost. What merits a price is curation, access, working formats and the quality of the relationships that emerge. A smaller conference can then be more valuable than a larger one. And a good hybrid offering is not a cheap transmission of the room program, but a distinct route to information and participation.

This is not a rejection of conferences. It is a rejection of the idea that attendance is a value in itself. AI makes weak events harder to justify and good ones more valuable: less frontal programming, less ritual, less travel out of habit—and more exchange, more surprise and more time for the people in the room.

Perhaps the AI conference host was therefore not a glimpse of a people-free event. Perhaps it was a pointer toward a more human one. If machines take over the updates, we do not need to stop meeting. We simply need to remember why we meet in the first place.