Z’s Scribbles
← Index

confused re: how to have an impact in biotech (and aging)

29-Aug-26 — 644 Words — 4 Min Read —#misc

[This is day #14 of my 30-day microblogging challenge. You can and should still join!]

As you know, I’m all about maximizing impact, especially by helping humanity solve disease (and hopefully aging one day!).

Until now, biotech organizations have come in tons of shapes and sizes. They would focus on different diseases or parts of the technological stack (i.e. infrastructure or platform technologies).

But now everything seems to converge on AI and automation (as well as generating and collecting the necessary data to accelerate these two).

Where should we focus?

The most AI-pilled folks seem to think that investing time and effort into tackling any part of the stack directly is a waste of time, since AI will be able to do all of it much faster and more efficiently in a few years. Some seem to believe that AI will automagically solve everything, given the unexpected capability leaps achieved by scaling up compute power. (Scientists with actual lab experience seem much more skeptical.)

Thinking about it from first principles, given current advances in AI, it makes sense to focus on any part of the stack that enables fully automated AI-powered closed-loop systems for generating physical and biological data.

The catch is that these niches already have many well-funded players, and we can expect many more to pop up as new and existing organizations expand into these areas. I’m unsure how competitive a brand-new startup or lab could be.

Talent and expertise

I am also not sure if it would make sense for someone who isn’t already specialized in these fields or adjacent ones to pivot completely or start training from scratch, given the timelines.

It seems likely that specialized scientific talent will be needed to steer these AI models for at least a few years (2-3? I would say <10, unless general AI development is paused or hits a plateau). However, AI adoption and automation will make training junior talent increasingly difficult to justify, and more human talent will gradually become obsolete once the models can handle both cognition and physical execution.

Additional bottlenecks

I believe that AI systems are likely to face roadblocks in the real world, even if the main R&D gaps are solved. These roadblocks might be regulatory, political or economic in nature, and may require human intervention to overcome.

So my instinct is to try to figure out what they might be, and focus on removing them (basically clearing blockers to AI-driven scientific acceleration). (i.e. clinical trials, resource allocation, or permits to build more infrastructure)

Biosafety is another area worth focusing on, especially if the above seems inevitable.

How this affects me

Personally, I’m at a loss. My level of expertise in AI, biology or robotics isn’t deep enough to allow me to contribute meaningfully on the technical side, and a traditional academic program doesn’t seem worth it given the timelines.

I will continue working on Primordia, as I still believe that microgrant programs like it are valuable, as they enable the exploration of new scientific avenues by funding scientists to do field research, invalidate hypotheses, collect helpful data, and upskill talent.

However, I want to contribute more directly. The options I see are:

I would love some advice! If you have more ideas or have been thinking about these topics yourself, I’d love to hear your thoughts.