What core research are you most excited about?

Conscious realism and the cosmological polytope:

https://www.youtube.com/watch?v=reYdQYZ9Rj4

https://arxiv.org/abs/1709.02813

According to Prof. Hoffman, this model of the universe is highly parsimonious in that it can model data from the Large Hadron Collider using a single parameter, while the best incumbent theory (quantum field theory) needs millions.

The implication is that everything we see and experience, including space and time itself, are not fundamental. The unit of reality is consciousness.

This has far reaching applications into every other scientific and non-scientific human endeavor, from neuroscience to philosophy. It's no exaggeration to say that if he's right (and he claims the math shows that he is) it may be the most important discovery in the history of humanity.

Even Albert Einstein appears to have intuited this when he wrote:

"Time and space are modes by which we think and not conditions in which we live."

Deep geothermal energy. Here's the best explanatory video I've seen: https://youtu.be/8erbvqFZ9M8?t=50

Caveat, I have no idea if these guys can do it but someone can. At this stage, it's about getting done it accurately and profitably.

https://newatlas.com/energy/quaise-deep-geothermal-drilling-...

https://www.vox.com/energy-and-environment/2020/10/21/215154...

Machine Learning:

- Image generation

- Multi-modal AI: (one algorithm that can process text, video, images, audio, etc)

Genomics

- liquid biopsy / cfDNA: early cancer detection from a blood test, other diagnostics

- fast (<1 hr) cheap (<$20) sequencing

Machine learning + biology

- AlphaFold++

OS kernel level observability. I shouldn't need to fire up `strace` for that; I should be able to do something like `observe $my_program $my_args` and have streaming results somewhere in `/proc`.

Another thing: CPU-level observability. I want to be able to get statistics about a block of code: which instructions got ran the most, which ones are the slowest etc. Profiling on steroids, basically. If Intel / AMD are not willing to provide that then I'm willing to work with almost any CPU model that does offer that. Anybody knows of such CPUs?

I'm also interested in eBPF but sadly can't find the time to dive deep there.

Fusion is a big one.

Some of the quantum computing research is pretty wild lately. People essentially writing "code" that gets converted by a quantum computer into new states of matter.

https://www.popularmechanics.com/science/a40359762/paired-ti...

I think there's some core research in magnetic control of plasmas / energy transfer from plasma that's been done, and is going to start leaking into more public applications like fusion. The military's "hypersonics" may be a beginning of some of that; bringing out things they've either been sitting on or desperately trying to get working for 30+ yr.

The elaboration of rulial theory: https://www.wolframphysics.org/

It’s such an intriguing idea, and early correspondences seem so strong, that it seems impossible we won’t learn some awesome things about how physical theories relate and arise.

ML-assisted combinatorial optimization and searching in high-dimensional discrete spaces (MILP/SAT/SMT solvers/theorem provers, etc)

ML-assisted continuous regime optimization (e.g., in solid/fluid mechanics problems, chemical problems, etc)

ML-assisted PDE solvers (e.g., weather simulation, etc)

Advances in additive manufacturing. Think 3D printers on steroids.

Malware and WASM/JavaScript

Fusion