In response to the UK Government’s Replacing Animals in Science strategy, a familiar refrain has surfaced from some in the research community: that we will “always” need animals in science, and that we will “never” be able to fully replace animal use.

For example, in a letter published in The Guardian, senior UK researchers argued that while alternative methods should be developed, “we are definitely not ready to abandon research with animals” and “for some disciplines we may never get to this.”

The letter was written in response to the government’s plan to accelerate the shift toward non-animal approaches. The idea that we’ll always need animals, or that some boundary is absolutely uncrossable, is a form of catastrophic thinking. It assumes:

• The current method is the only method.
• The barriers we see today are permanent.
• Innovation will stop at a boundary we set.

So let’s examine those assumptions.

Recent scientific history

If we’ve learned anything from the last few decades, it’s that our thoughts at a given time about what is “impossible” are often wrong. Here are three examples that illustrate how what seemed like science fiction twenty years ago, are now science fact.

1. Solving the protein folding problem with AI

For decades, the protein folding problem — predicting a protein’s three-dimensional structure from its amino-acid sequence — was widely regarded as one of biology’s great unsolved challenges. The complexity of combinations made the task appear computationally intractable, and meaningful solutions were not expected in the near term, if at all. https://pmc.ncbi.nlm.nih.gov/articles/PMC8567224/

Then AlphaFold arrived.

In 2020, DeepMind demonstrated that deep learning systems could predict protein structures with accuracy comparable to experimental approaches, releasing predictions for hundreds of millions of proteins. https://deepmind.google/blog/alphafold-five-years-of-impact/

This was not incremental progress — it was a paradigm shift. A problem long assumed resistant to computation turned out to be solvable once the right tools emerged. For animal replacement, the lesson is obvious: transformative insight can arrive suddenly — and reshape every aspect of how research is done.

2. Growing human organoids that model disease

Twenty years ago, the idea that stem cells could self-organise into functional mini-organs in vitro sounded speculative at best.

Today, human organoids and organ-on-a-chip systems are routinely used to study development, cancer, infection, and neurological disorders. They capture aspects of human physiology that animal models often fail to replicate — and are improving rapidly as shown in this report on moving beyond animal testing.

These systems are not perfect. But neither are animal models. What matters is that human-relevant biology can now be studied experimentally in ways previously thought impossible.

3. Precision gene editing with CRISPR

Before CRISPR, genome editing was slow, expensive, and technically demanding. The idea that researchers could routinely and precisely modify DNA sequences in living cells — let alone move such tools into clinical settings — would have sounded optimistic.

CRISPR-Cas9 changed that trajectory almost overnight, moving from discovery to Nobel Prize to therapeutic trials within a decade. https://sciencemediacentre.es/en/story-crispr-last-ten-twenty-and-thirty-years

This shows how new technologies can dramatically accelerate the pace of change.

What this means for replacing animal use

When scientists say we will always need animals, what they often mean is: “I cannot currently see how to do my work without them.” That is honest — but it is not a forecast.

The examples above share a common lesson: Scientific limits are not fixed.

Animal use persists not because it is perfect, but because it is historically embedded, institutionally supported, and familiar. The government strategy acknowledges that technological advances — including AI, organoids, and 3D cell systems — “finally allow us to see a path to changing our reliance on animals in science.”

The trend is clear: tools that were unthinkable are now real.

Looking forward, not backward

If we’ve come this far what might the next two decades bring?

• Human-centric models that predict human responses better than any animal ever could.
• AI systems that design and simulate experiments without animal intermediates.
• Fully integrated multi-organ systems on chips for toxicology and disease research.

And yes — there will still be challenges. There always are. But history shows us that declaring something permanently impossible is not only pessimistic… it’s unscientific.

So let’s debate timelines. Let’s acknowledge complexity and our own ignorance. Let’s be rigorous about validation and uncertainty. But let’s also drop the rhetoric of “always” and “never” because I, for one, believe in the wonder and marvel of science. I believe scientists are capable of achieving the impossible. And that is an evidence-based position if ever there was one.