autechre

Confirming previous forecast

While synthesis specific is not necessary, if we can synthesize more complex proteins (and this is my prediction w.r.t. >400 amino acid question) then it would be more feasible to make biological agents that target a specific system. It doesn't have to be a specific group in the human population, it could target agricultural systems (the former would be quite more risky than the latter). Vaccines are made ahead of time before deployment and its distribution controlled away from an adversary, requiring the adversary to expend resources into manufacturing its own vaccines or steal / reverse-engineer it.

Chemicals are simpler to make but lack the same level of specificity.

We're only a few stages close in the tech tree. Dependent technologies would include:

  - Artificial Intelligence and misc. RL / ML techniques are perfectly suited for solving non-linear problems such as the arrangement of amino-acids into functional proteins. First of all, systems like Alphafold already demonstrate the effectiveness of such techniques, despite its issues with "hallucinating" protein folds. My hunch is that "hallucination" is a general problem and not something specific to ML-based protein-folding predictors; so the attention towards reducing error-rate in i.e. LLMs would be quickly re-applicable towards making protein-folding predictors more reliable.

- Demonstrated implementation of control systems for stabilizing nano-scale material: I recall reading a paper (can't recall the title) where the scientists used RL (or some other ML technique) to arrange gallium(?) atoms with a (SEM or STM?) microscope. The general principle is that it can be demonstrated for a ML system to create a control loop specifically adjusted for the imperfections of the environment in-which the atoms are arranged, i.e. abrasive / uneven edges of the STM needle (I think it's called?). There are also more methods that we're coming up with in order to control atomic structures. Also stuff like this: https://phys.org/news/2024-05-physicists-atoms-proximity-paving-exploring.html

- Incentive: New drugs, and a step towards nanomachines

- Compounding effect: Being able to synthesize arbitrary proteins could make it easier to build more-and-more complex proteins.

Basically tied to AI progress, which many have constantly mis-predicted so far.

it's been a while. not likely in the next 6 months