Making Research Accessible, Making AI Interfaces Useful — Part 3 of 3

Understanding enough to explore a paper

A listener finishes the best-paper coverage without understanding every detail. The hosts’ questions keep them listening: they surface questions the listener has half formulated—or would not yet know to ask. In this closing portion of the recap and anniversary recording, the listener describes building familiarity through repeated exposure, even before the research fully makes sense.

That matches the host’s intent for the NeurIPS research recap: select important papers and give listeners a starting point for deeper reading. Listeners could pursue whichever work caught their attention; the hosts did not claim expertise in every topic. Breadth still carried a cost: the edited episode ran roughly three and a half hours, prompting complaints about its length.

The host’s interview preparation draws on a similar accumulation of context. Guest research sits alongside daily professional involvement in AI conversations, which supplies background and reduces the need for extensive preparation specific to each episode.

Prioritizing technical work over cultural reporting

The listener next asks for more documentation of AI builder communities: the conversations in shared spaces that appear only in glimpses on Twitter. The host is open to occasional cultural coverage but prioritizes technical material listeners can apply at work. Personal interviews and the substantial editing needed for narrative reporting would also require a production process the host finds difficult.

AI user experience needs visible examples

The listener identifies as a user experience (UX) designer and asks for substantially more resources on AI and interface design. The practical interest is how the technology can become useful in products.

The host welcomes the subject but identifies a format constraint: audio cannot show the interface. An earlier AI UX meetup is available on YouTube and in the Latent Space newsletter archives, though the host does not recall a podcast edition. Seeing the interaction matters to understanding the proposed experience.

At the time of the conversation, the host points to several places to look for this work: - Product companies: Notion, Vercel, and Retool are expected to be major contributors to AI UX. Interviews with Notion and Retool have already happened; a Vercel interview is planned. - Individual practitioners: Amelia Wattenberger’s AI Engineer Summit presentation is available on YouTube.

Yet the host’s assessment is skeptical: they have seen no new AI UX ideas since the meetup roughly a year earlier, despite the dissatisfaction with chatbots and interest in alternatives they encounter. This is a judgment about the work they have seen, not a measured finding about the field.

The designer sees a participation gap. Designers, product practitioners, and UX researchers should help apply large language models (LLMs) to real problems. They report seeing little of that activity in Sydney or on Twitter and hope to help grow it. The request for more coverage becomes a call for more people to do and share the work.

Organize the next meetup

The host responds with an invitation to organize another AI UX meetup, offering a concrete division of responsibility: - The host offers a venue and connections: community access and introductions can help get the event started. - The listener finds speakers: recruiting presenters remains the organizer’s responsibility.

The listener already has relevant experience: they adapted the earlier AI UX meetup format for a small Sydney event with demos and approximately 20–30 attendees.

Both express interest in proceeding, though no event is scheduled in the exchange. The perceived shortage of AI UX discussion now has a practical response: bring speakers and demos into an existing community.

The anniversary sign-off asks listeners to send feedback and recommend Latent Space to friends. Word of mouth is presented as support for booking guests and attending events as the podcast enters its second year.