Hello Bioinformatics lovers,

Tommy here. Like many of you, I am overwhelmed by the advancement of AI too.

Someone asks me every other week whether AI is going to replace bioinformaticians.

A new npj Digital Medicine paper put my answer in print.

Here is the scene that settles the question better than any argument. An analyst runs a single-cell dataset through an AI agent.

Out comes a clean UMAP, cell types labeled, a figure ready for the slide deck. Then I ask why the tumor samples cluster by sequencing batch instead of by subtype. Silence.

That silence is the whole point.

The paper, "Rethinking bioinformatics expertise in the era of artificial intelligence," makes a claim I have repeated for two years. AI accelerates the work.

Whether that speed produces knowledge or garbage depends on expert judgment in four places: design, data curation, interpretation, and governance.

Skip any one of them and you ship the garbage faster.

AI is good at the things you can check by eye. It catches a syntax error, a mislabeled column, a sample that fails QC.

It will not tell you that your batch correction erased the biology along with the noise.

It will not pause before enriching a "novel" pathway that anyone who has stared at TCGA twice would recognize as an artifact.

You have to know. The model has no idea what it is looking at.

This is why the job is changing rather than vanishing.

You spend less time babysitting Cell Ranger and STAR and more time on the parts that decide whether the result is real: framing the question, choosing the right controls, and catching the moment the model produces confident nonsense.

That second list is harder than the first. It is also where the value sits, and where AI cannot follow.

The paper frames the shift the same way. The bioinformatician moves from running workflows toward designing the analysis, driving the hard discovery, and owning the governance of how these tools get used.

The expertise does not disappear. It moves up the stack.

So here is my honest read on the next five years. The people who struggle will not be the ones who adopted AI. They will be the ones who trusted it. The people who do well will use it every day, on real problems, and still know when it is lying to them.

The paper is open access and worth the read: https://www.nature.com/articles/s41746-026-02777-1​

Hit reply and tell me the last time a tool handed you a clean figure that turned out to be wrong. I read every one.

Happy Learning!

Tommy aka crazyhottommy

PS:

If you want to learn Bioinformatics, there are four ways that I can help:

1. My free YouTube Chatomics channel, make sure you subscribe to it.
2. I have many resources collected on my github here.
3. I have been writing blog posts for over 10 years https://divingintogeneticsandgenomics.com/​
4. Lastly, I post daily on Linkedin and you may find useful posts there https://www.linkedin.com/in/%F0%9F%8E%AF-ming-tommy-tang-40650014/recent-activity/all/ (make sure to follow me if you do not want to miss any posts).

Stay awesome!