Hugin & Munin Journal

Artificial intelligence is no longer only speeding up isolated scientific tasks. In some fields, it is beginning to change how candidate solutions are generated, which experiments are attempted and where researchers direct their attention. AI systems can propose hypotheses, search molecular and mathematical spaces, plan experiments, analyse results and generate synthetic examples. They are not replacing science as a whole, but they are redistributing effort within parts of it. A 2025 perspective in npj Artificial Intelligence surveys the use of large language models across the scientific process, from literature review and hypothesis formation to experimental design and analysis. It also makes an important qualification: AI’s impact on fundamental science, defined as the discovery of new principles or laws, remains limited. Since that perspective appeared, the boundary has moved further. A 2026 Nature paper on Co-Scientist reported AI-generated scientific hypotheses taken throug...

A familiar theory of innovation says that progress becomes harder because the low-hanging fruit is picked first. Early discoveries lie close at hand; later advances require larger machines, larger teams, deeper specialisation and more money. Nicholas Bloom, Charles I. Jones, John Van Reenen and Michael Webb developed an influential empirical version of this argument in “Are Ideas Getting Harder to Find?” . Across semiconductors, agriculture, medicine and firm-level innovation, they found research effort rising substantially while measured research productivity declined. Their most striking example is Moore’s Law: maintaining the historical rate of improvement in chip density requires far more researchers than it did in the early 1970s. There is considerable truth in this. Mature fields accumulate technical difficulty. Once obvious routes have been explored, the remaining work often requires greater precision, more infrastructure and command of a much larger body of knowledge. The m...

Venture capital attracts criticism for familiar reasons: too much money chasing fashionable ideas, companies scaling before they understand their customers, founders and investors protecting themselves while employees discover that their options are worthless. Look only at the visible failures and the verdict seems straightforward. Yet venture capital also finances a kind of company that ordinary debt handles badly. A young firm may have little collateral, uncertain revenue, a prototype rather than a product and a serious chance of producing nothing valuable at all. Requiring regular loan payments while those questions remain unresolved can end the experiment before it produces an answer. The usual defence of venture capital nevertheless states the case too neatly. Venture investors do not absorb failure on behalf of society. They construct portfolios in which losses are expected because a small number of exceptional companies may repay them many times over. The financial write-off...