GP / Blog / Science & Technology

Issue forecast

The responsibility of the scientist: knowledge, morality and foreseeable harm

The classic defence is that science is neutral and only its use is moral or immoral, so blame the general, not the physicist. The 2026 twist is that some of the loudest warnings about AI now come from the very people building it, which makes the old separation of knowledge from consequence harder to hold.

Theme · Science & TechnologyLast set at A-Level · 2020 (on whether a scientist should be concerned with the pursuit of knowledge and not its morality)Foresight, not a prediction

In short

A scientist-responsibility question rewards a student who can argue the neutrality thesis properly before dismantling it. Science as pure inquiry, with morality entering only at the point of use, is a real and respectable position. Argue whether it survives contact with research whose harmful uses are foreseeable at the bench, from gene editing to dual-use AI.

Why this could come up now

This is a hardy values strand. It appeared as human need rather than profit driving research in 2016, animals in research in 2017, and knowledge versus morality in 2020. The examiner returns to the scientist's duty on a steady beat because it forces genuine ethical argument rather than topic knowledge.

The contemporary hook is unusually sharp. In 2026 the people most publicly worried about advanced AI include its own leading researchers, the same scientists whose work creates the risk, which puts the question of a scientist's responsibility for foreseeable harm back in the news an examiner reads.

Framed honestly: a values question of this kind is a frequent visitor rather than a certainty for any one year. It is worth rehearsing because the neutrality-versus-responsibility argument is one most candidates handle clumsily, reaching for Oppenheimer and stopping there.

201320162017202020232025

set at A-Level most recent appearance. A recurring values strand on a few-year beat. The scientist's duty returns because it forces real ethical argument.

What an essay on this would test

These questions test whether you can argue about responsibility, not just gesture at famous harms. The neutrality thesis, that a discovery is morally blank and only its application carries weight, is the position you must engage seriously before you can defeat it. A script that just lists the atom bomb has not made an argument.

They also test the idea of foreseeability. The strongest line is that a scientist owns the uses that are reasonably foreseeable from the work, which is a sharper claim than 'scientists are responsible for everything' or 'scientists are responsible for nothing'. The argument lives in where that line of foresight falls.

Operative angles

value-neutrality: the claim that knowledge itself is morally blank and only use is moral, which the question usually asks you to test
dual-use: research that can heal or harm with the same finding, where the harm is foreseeable at the bench
foreseeability: a scientist may not control every use, but owns the uses a reasonable person could predict

How to answer it: two ways in

Two distinct, defensible routes through the question. A strong script commits to one and uses the other as the concession it answers, rather than sitting on the fence.

Science is neutral

Blame the use, not the discovery

Knowledge itself is morally blank; a finding can heal or harm depending on who wields it, so the moral weight sits with the user and the regulator, not with the scientist pursuing truth.

Restricting inquiry to what might be misused would have blocked discoveries that turned out to save far more than they cost.
The scientist rarely controls the politics and markets that decide how a finding is deployed, so holding them to account for distant misuse is both unfair and impractical.
Progress depends on a culture that prizes open inquiry; loading every researcher with responsibility for every downstream harm chills the work itself.

Worked exampleCRISPR gene editing, the basis of the first approved gene-editing medicine, treats inherited disease and, in different hands, raised the prospect of editing human embryos, the same tool pointed at opposite ends, which is the neutrality thesis in one technology (widely documented, as of 2026-06).

The scientist owns foreseeable use

Knowing the likely harm is part of the work

Neutrality breaks down where the harmful use is reasonably foreseeable at the bench, because a scientist who can predict the misuse and proceeds anyway has made a moral choice, not just a discovery.

Dual-use research, where the same finding heals and harms, is foreseeable by definition, so the 'I only sought knowledge' defence rings hollow.
When the builders themselves issue public warnings, they concede they can foresee the danger, which is precisely the knowledge that creates responsibility.
Self-regulation by scientists, refusing certain experiments, publishing with safeguards, exists because the field already accepts that some responsibility sits before the point of use.

Worked exampleIn May 2023 hundreds of AI researchers, including leading figures who built the field, signed a one-line statement that mitigating the risk of extinction from AI should be a global priority alongside pandemics and nuclear war, an admission by the makers that they foresee the danger of their own work (Statement on AI Risk, Center for AI Safety, as of 2026-06).

The fuel: stats, facts and examples

May 2023

hundreds of AI scientists and leaders signed a statement that AI extinction risk should be a global priority, a warning from the makers themselves

Source: Statement on AI Risk, Center for AI Safety · as of 2026-06

2018

year a Chinese scientist announced the first gene-edited babies, drawing near-universal condemnation and a prison sentence

Source: He Jiankui affair, widely documented · as of 2026-06

2023

the first CRISPR gene-editing therapy was approved, the benign end of the same technology that enabled the gene-edited-babies scandal

Source: US FDA approval of Casgevy, Dec 2023 · as of 2026-06

Facts worth deploying

The scientist He Jiankui was sentenced to three years in prison after announcing gene-edited babies in 2018, a case the scientific community treated as a failure of individual responsibility, not merely of regulation.Source: He Jiankui affair, widely documented, as of 2026-06

Dual-use research, work that can be applied to benefit or to harm, is a recognised category in science governance precisely because the harmful use is often foreseeable when the research is done.Source: science-governance consensus, as of 2026-06

That leading AI researchers publicly warn about their own field is itself the strongest counter to pure neutrality: a maker who foresees the harm cannot fully disown responsibility for it.Source: Statement on AI Risk, 2023, as of 2026-06

FAQ

Can I just use Oppenheimer and the atom bomb?

You can mention it, but if it is your only case you will land where most scripts do. The bomb is overused and easy to wave at. Pair or replace it with current dual-use cases, gene editing, AI safety warnings, where the foreseeability argument is fresher and harder for the examiner to predict.

Is the neutrality argument the weak side I should knock down quickly?

No. Argue it properly first. Neutrality is a serious position, and an essay that strawmans it to set up the responsibility view reads as one-sided. The marks come from granting neutrality real force, then showing precisely where foreseeable harm breaks it.

How is this different from a general 'is science good or bad' essay?

The subject here is the scientist's duty, not science's net effect. Keep the person, the researcher making a choice, at the centre. The question is who bears responsibility for what science produces, which is sharper and more arguable than weighing benefits against harms in the abstract.