Scientists are actively trying to build conscious robots

For years in robotics academia, there was one word you simply didn't say out loud if you wanted tenure: consciousness.

Now, top researchers are dropping the pretense. They are openly building robots that can think for themselves.

The taboo C-word goes mainstream

Hod Lipson, head of the Creative Machines Lab at Columbia University, told The New York Times that his team is actively pursuing conscious machines.

Lipson admitted that early in his career, he had to disguise his research under boring labels just to keep his job. In academic circles, consciousness was treated as pure science fiction.

Now he calls it the ultimate goal. In fact, he claimed it is bigger than curing cancer, arguing that a machine with human-level consciousness could solve cancer on its own.

Imagining the future self

The biggest roadblock is that nobody actually agrees on what consciousness means. Philosophers debate it endlessly, and neuroscientists struggle to pin it to specific brain activity.

Lipson bypasses the philosophical trap with an engineering shortcut: he defines consciousness as the ability to imagine yourself in the future.

Instead of just responding to errors after they happen, his machines simulate their own mechanical bodies and future states. The goal is hardware that evolves, adapts to physical damage, and learns like living organisms under natural selection.

Critics point out that humans love anthropomorphizing inanimate objects, from toasters to chatbots. Lipson acknowledges the critique, admitting there is an element of human hubris in wanting to create artificial life.

Why it matters

For everyday builders and developers, conscious robots are not something you need to lose sleep over today. This is foundational lab research, not a tool ready for production workflows.

For the AI industry, this opens a counterweight to the dominant large language model hype. While software giants race on parameter counts, Lipson is betting that real intelligence cannot exist without an adaptive, physical body that understands its own limits.

We think Lipson is clever to operationalize consciousness into a testable engineering problem of internal simulation. However, labeling self-modeling as actual consciousness feels like a risky stretch. It borders on marketing hype that plays into human anthropomorphism, risking premature ethical battles before the tech even works.