The future of the bioinformatician
Knowledge workers are in for a rough ride. Models are taking on the most famous problems in mathematics and discovering new biological systems in a day, while serious mathematicians despair and protest. What is happening to software developers and mathematicians now will reach bioinformaticians within the next few years.
This piece looks at how the bioinformatician's job is changing and which skills the market will value in the near future.
Why it's rough: biotech investment isn't growing
Biotech investment fell until 2022, then recovered, and for the last three years it has held at about the same level, around $100B a year. Companies are learning to work with the money they have. That means they invest less in science and invention and think more about sales and production efficiency.
Overall there are fewer job openings in life sciences and bioengineering, mostly because openings for research scientists are shrinking, while openings for physicians and sales managers are growing.
Meanwhile, the whole world needs people who can work with AI, or are at least ready to learn and adapt: organize several agents with different tasks, send them off to work, then collect and check the results. For a manager, building a joint team of people and agents becomes a key skill.
Why it's rough: neural networks keep getting better at code and analysis
Analysis is no longer a profession on its own. Agents generate pipelines, code, statistical methods and plots very fast and pretty well, not to mention reports, presentations and interactive dashboards.
Anthropic discovered a new biological system and confirmed in a real wet lab that it isn't fake. But it was AI that found the candidate: about 950 Claude agents screened more than 200,000 reverse transcriptases in 21 hours, picked 3,500 new candidate systems and wrote reports on the top 20. What comes next? A thousand Mythos 6 agents running for a hundred hours with a robotic lab attached?
A bioinformatician who only runs statistics and draws plots is competing with a swarm of frontier-model agents.
And it will happen fast: Anthropic employees expect AI to be able to do 75% of their current work by 2030, and 91% of them expect AI to make a Nobel-level discovery by the same year.
What to do
If you want to become a bioinformatician, there will most likely still be enough work for our generation, but the profession will change a lot. As models improve and research speeds up, existing markets will grow: oncology, immunology, obesity, mental and women's health. New markets will appear: real longevity, experiments in simulation, implanted chips and robotic transplants. And someone will have to do all of this.
A good specialist's value is shifting from excellent analysis to meta-skills: the very start and launch of a project, and the last mile, which means checking results, shaping value and strategy for the client, and delivering the result.
A bioinformatician with a future
So who is the bioinformatician you'd hire and wouldn't want to replace with a neural network?
A bioinformatician without a future
Denies progress. "These AIs can't do anything properly, and the bosses and the market are stupid."
Neural network operator. Demands a one-hundred-percent clear assignment from a manager, a colleague or a client, just to feed it to a neural network. Doesn't try to understand the problem on their own, doesn't take on more responsibility, isn't curious about neighboring fields. Will be replaced by a neural network.
Doomer. "They'll replace me tomorrow!" Just toxic; you don't want them spoiling your mood.
Conclusions
I want wars to end, people to be wealthy and healthy, investment to flow into biotech, and us to make money. None of this analysis guarantees, of course, that 2035 will turn out exactly like this. But I will still try to be useful and keep learning new skills.
