Gom Verdon, who's working on a new type of computing that's so different from what we're used to, it's hard to wrap your head around. He's building a computer that uses the principles of thermodynamics to perform calculations.

Gom and his team started out in quantum computing, where they figured out how to bring differentiable programming to quantum computers. But they quickly realized that noise was a major problem. It's not just a quantum issue, either - as circuits get smaller, voltage noise starts to become a problem in regular computers too.

Most people would try to get rid of the noise, but Gom and his team had a different idea. What if, instead of fighting the noise, they could use it to their advantage? They started exploring alternative approaches to computing, ones that could harness the noise from heat as a resource.

The result is a new type of computing that's not just faster, but also more energy-efficient. It's based on continuous variables and fuzzy values, rather than traditional bits and zeros. This approach is closer to how our brains compute information, and it has the potential to revolutionize the field of artificial intelligence.

Gom showed me some numbers that blew my mind. They're building a computer that can do probabilistic inference and learning by heat dissipation. This approach is so efficient that it could potentially be 100 million times more energy-efficient than current computers.

But what really gets me excited is the potential for this technology to enable new types of intelligence. We're not just talking about faster computers - we're talking about a new way of thinking about intelligence.

Gom and his team are building a full stack for thermodynamic AI, from hardware to software. They're working on developing algorithms and middleware that can map probabilistic ML algorithms onto this new hardware.

One of the most interesting things about thermodynamic computing is its potential to enable new types of machine learning. They're working on developing neural energy-based models that can learn and represent complex distributions in a highly efficient manner.

I asked Gom how this technology could be used in practice. He told me that it could potentially be used for everything from climate modeling to materials science to medicine.

What I love about Gom and his team is that they're not just building a new type of computer - they're creating a new paradigm for intelligence. They're working on developing a new way of thinking about intelligence that's based on the principles of thermodynamics.

This technology is still in its early days, but I'm excited to see where it goes. The potential for thermodynamic computing to enable new types of intelligence is vast, and I think it's one of the most interesting areas of research right now.

Gom and his team are pushing the boundaries of what's possible with computing. They're not just building faster computers - they're creating a new way of thinking about intelligence. And that's something that could potentially change everything.