Inference: When Will We Solve Intelligence

In 2025, we launched a video interview series called Inference. It was an experiment: was it possible to create not a simple YouTube channel, but a knowledge asset that tries to answer one big question: When Will We Solve Intelligence? Something where the edge is not speed, hype, or exclusivity, but the coherent accumulation of understanding of what we are doing and how we got here. Conversations that help explain how AI is actually evolving, why some ideas keep resurfacing across very different teams, and why others quietly fade once you look closely enough.

Over the year, we spoke with researchers, founders, engineers, and entrepreneurs working on the front lines. It was extremely interesting, and now it’s time to bring Inference to a new level. We’re moving to topic-driven interview cycles.

In 2026, each cycle will focus on one area and unfold across multiple conversations. Think robotics, open source, coding, and other topics that will define 2026, explored one by one, in depth and with continuity. But an interview, even when transcribed, is still an interview. With each cycle, we will now introduce a Topic Report that brings the conversations together, adds context, and surfaces the patterns that only become visible when you step back.

These reports are meant to be kept. This is the kind of analysis you come back to, because it continues to compound in value, helping you understand, decide and act.

2025 was the start. Here are our favorite conversations:

What Intelligence Is Missing

Conversations about gaps: reasoning failures, memory limits, hallucinations, and why current systems still struggle with abstraction, grounding, and transfer. This category asks what intelligence actually requires once scale stops being the answer.

Coding and the Agentic Web

From answers to action. This is about agents, tools, browsers, developer workflows, and the emerging web where AI is an active actor.

Who Is in Control

Questions of trust, oversight, safety, and human involvement. These conversations explore where autonomy should stop, how control degrades, and why keeping humans “in the loop” remains an unsolved design problem.

AI Meets the Real World

Physics, biology, institutions, and domains where failure has consequences.

You can find the rest of interviews here: https://www.youtube.com/playlist?list=PLRRoCwK1ZTNCAZXXOswpIYQqzMgT4swsI

And if you are more of a reader, here are all the transcripts: https://www.turingpost.com/t/Interviews-with-Innovators