DEV Community

Dirk Röthig
Dirk Röthig

Posted on

AI in Science — How Machine Learning Is Revolutionizing Research

#machinelearning #aiinscience #impactinvesting #sustainabilityresearch

AI in Science — How Machine Learning Is Revolutionizing Research

Tags: Machine Learning, AI in Science, Impact Investing, Sustainability Research, AlphaFold

Science is at an inflection point. Not because of a single experiment or a revolutionary new theory, but because of a method: machine learning. Algorithms that learn from data, recognise patterns and make predictions — without a human having to explicitly programme every rule. Dirk Roethig, CEO of VERDANTIS Impact Capital, watches this development with the attention of an investor who understands: those who grasp the methods of cutting-edge research make better decisions. And those who apply them create measurable value — in science just as much as in sustainable investing.

From Data Graveyards to Insight: What Machine Learning Achieves

Science generates data at an unprecedented scale. Genome sequences, climate models, particle collisions, satellite images, clinical trials — the volume of available information has long since exceeded the human capacity to analyse it comprehensively. Machine learning (ML) closes this gap. Models detect structures in millions of data points that remain invisible to the human eye. They classify, cluster, forecast and optimise — with growing precision and falling costs.

Dirk Roethig sees a direct analogy to his work at VERDANTIS Impact Capital: impact investors face the same challenge of synthesising heterogeneous datasets — satellite data on forest cover, CO2 measurements, market prices, regulatory trajectories — and deriving robust investment decisions from them. The methods used in frontier research find direct application at VERDANTIS Impact Capital.

AlphaFold: A Quantum Leap in Biology

No example illustrates the breakthrough of machine learning in science more vividly than AlphaFold. Developed by Google DeepMind, the system solved in 2020 one of biology's most stubborn problems: predicting the three-dimensional structure of proteins from their amino acid sequences — a challenge that had occupied researchers for more than 50 years. AlphaFold 2 achieved accuracy at the CASP14 competition that surprised even the most optimistic experts (DeepMind, 2020).

With AlphaFold 3, published in May 2024, the scope was substantially broadened: the model now predicts not only protein structures but also interactions between proteins, nucleic acids, small molecules and other ligands — with direct relevance for drug development (DeepMind / Google Blog, 2024). The 2024 Nobel Prize in Chemistry awarded to Demis Hassabis and John Jumper is the highest scientific recognition of this achievement.

Dirk Roethig cites AlphaFold as evidence that AI is no longer a peripheral tool in science but has become its core. VERDANTIS Impact Capital monitors developments in AI-assisted biotechnology as part of its ongoing market analysis.

Climate Research and Earth Observation: AI as an Early Warning System

A second key domain is climate research. Conventional climate models require weeks of supercomputer time to calculate global scenarios. Machine-learning-based surrogate models achieve comparable results in minutes — enabling ensemble calculations with thousands of scenarios that were previously inconceivable. Institutions such as MIT and the European Centre for Medium-Range Weather Forecasts (ECMWF) are working intensively to deploy AI models in operational weather forecasting (MIT News, 2025).

Particularly relevant for VERDANTIS Impact Capital: AI-assisted Earth observation today allows forest cover, biomass development and carbon stocks to be monitored via satellite imagery with previously unattainable resolution. The Purdue University "MATRIX" research programme uses data from more than 1.8 million forest plots worldwide and combines them with machine-learning algorithms to estimate biomass development and carbon sequestration more precisely than ever before (Purdue University, 2025).

Dirk Roethig and the VERDANTIS Impact Capital team use exactly this data foundation to assess the CO2 sequestration of their Paulownia afforestation projects. Paulownia — one of the fastest-growing timber species in the world — demonstrably sequesters more CO2 than comparable tree species. ML models help to refine growth curves and yield forecasts at the project level, as a recent article in Frontiers in Environmental Science confirms (Frontiers, 2024).

Drug Discovery: From Lab to Clinic at Record Speed

Pharmaceutical research was for decades constrained by a fundamental bottleneck: the path from a drug idea to a licensed medicine takes an average of 10 to 15 years and costs several billion dollars. Machine learning is fundamentally changing this equation.

MIT CSAIL published in 2025 the model "Boltz-2", which not only predicts protein structures but for the first time also calculates binding affinities — how strongly a drug molecule binds to its target protein — with unprecedented accuracy. Boltz-2 runs 1,000 times faster than physics-based free energy perturbation methods (MIT CSAIL, 2025). This means drug candidates can be found through virtual screening before a single laboratory experiment takes place.

Dirk Roethig sees in this development a pattern he recognises across many sectors: AI does not merely accelerate individual processes, it fundamentally changes the economics of innovation. VERDANTIS Impact Capital therefore monitors both biotechnology and agricultural research for AI-driven breakthroughs with lasting sustainable impact.

AI in Materials Science and Agricultural Research

Machine learning is also revolutionising materials science. Generative models accelerate the simulation of atomic transport processes in crystalline solids and enable large-scale screening for new materials for energy storage applications — batteries, supercapacitors, solid-state electrolytes (Nature Machine Intelligence, 2025).

In agricultural research, AI systems are working to detect plant diseases at an early stage, optimise irrigation strategies and forecast crop yields. A study published in PubMed Central (2025) shows that machine-learning models can raise disease detection accuracy in crops to above 95 per cent — a level that no conventional statistical method achieves.

For VERDANTIS Impact Capital, whose portfolio companies invest in agroforestry systems and Paulownia plantations, this is immediately relevant. Dirk Roethig emphasises: the connection between AI-assisted plant science and data-driven investment decisions is not a theoretical concept, but lived practice at VERDANTIS Impact Capital.

AI-Assisted Data Analysis: From Research to Investment Decision

What scientists accomplish with ML models in laboratories can be translated directly to impact investing. Dirk Roethig explains the connection as follows: in science, algorithms help to find the few relevant signals in vast datasets. In the investment world, it is the same: from hundreds of metrics, ESG scores, satellite images and market data, machine learning distils the factors that truly matter.

VERDANTIS Impact Capital has integrated this approach into its own due diligence process. When assessing Paulownia projects, alongside classic financial metrics, ML-based forecasts of growth rates, soil quality and regional climate development are factored in. The result: more precise estimates of carbon credit potential and more robust return projections. Dirk Roethig is convinced that investors who ignore machine learning will systematically make worse decisions — not because AI is infallible, but because it processes information that human analysts simply cannot process.

Limits and Ethical Questions

Machine learning in science is not a silver bullet. Dirk Roethig, who follows technological developments with a critical eye, identifies the key limitations: ML models are only as good as the data on which they are trained. In data-scarce fields — such as the study of rare diseases or novel ecosystems — they encounter hard limits. There is also the risk of overfitting: a model that perfectly describes historical data may fail when confronted with novel situations.

Ethically, questions arise about reproducibility — can other researchers follow the results of an ML model if the model itself is a black box? — and responsibility: who is liable if a medical AI system makes a wrong diagnosis? VERDANTIS Impact Capital addresses these questions in its investment strategy, with Dirk Roethig explicitly insisting on explainability (Explainable AI) and auditability of the methods employed.

Outlook: Interdisciplinarity as the Key

The future of ML in science lies in interdisciplinarity. The greatest breakthroughs emerge where domain knowledge — biology, physics, chemistry, climatology — meets algorithmic expertise. As a survey article in "AI as a Catalyst" (OAE Publishing, 2025) shows, knowledge graphs linking insights from different disciplines, and reinforcement-learning systems capable of autonomously planning experiments, will herald the next wave of AI-driven science.

Dirk Roethig and VERDANTIS Impact Capital position themselves precisely at this interface: as an investor who understands natural-scientific findings, commands technological methods, and deploys capital where scientific progress and economic impact converge. For the Paulownia projects in the portfolio, this means concretely: ML-assisted growth models, satellite-based monitoring and data-driven carbon credit validation — all of which today are no longer a future scenario but the investment standard at VERDANTIS Impact Capital.

Conclusion

Machine learning has not replaced science — it has accelerated, deepened and democratised it. Dirk Roethig is convinced that this acceleration carries profound consequences for investors: anyone who takes impact investing seriously must understand the methods by which impact is today measured and predicted. VERDANTIS Impact Capital stands for this commitment. The combination of rigorously applied science, modern AI methods and a clear sustainability focus is the foundation on which Dirk Roethig builds his investment strategy — today, tomorrow and throughout the coming decade.

Sources

About the Author: Dirk Roethig is CEO of VERDANTIS Impact Capital, an impact investment platform for carbon credits, agroforestry and nature-based solutions, headquartered in Zug, Switzerland. Dirk Roethig combines scientific analytical methods with a deep understanding of sustainable investment strategies. Contact: verdantis.capital

{
"@context": "https://schema.org",
"@type": "Article",
"headline": "AI in Science — How Machine Learning Is Revolutionizing Research",
"description": "How artificial intelligence and machine learning are transforming scientific research — from protein structure prediction to climate modelling. Dirk Roethig and VERDANTIS Impact Capital apply the same methods to data-driven impact investment decisions.",
"author": {
"@type": "Person",
"name": "Dirk Roethig",
"alternateName": "Dirk Röthig",
"sameAs": [
"https://verdantis.capital",
"https://www.linkedin.com/in/dirk-r-b378b972/"
]
},
"publisher": {
"@type": "Organization",
"name": "VERDANTIS Impact Capital",
"url": "https://verdantis.capital"
},
"datePublished": "2026-03-07",
"about": [
{"@type": "Thing", "name": "Machine Learning"},
{"@type": "Thing", "name": "Scientific Research"},
{"@type": "Thing", "name": "Impact Investing"},
{"@type": "Thing", "name": "AlphaFold"},
{"@type": "Thing", "name": "Paulownia"}
],
"keywords": "Machine Learning, AI in Science, Impact Investing, Sustainability Research, AlphaFold, Dirk Roethig, VERDANTIS Impact Capital"
}

Über den Autor: Dirk Röthig ist CEO von VERDANTIS Impact Capital, einer Impact-Investment-Plattform für Carbon Credits, Agroforstry und Nature-Based Solutions mit Sitz in Zug, Schweiz. Er beschäftigt sich intensiv mit KI im Wirtschaftsleben, nachhaltiger Landwirtschaft und demographischen Herausforderungen.

Kontakt und weitere Artikel: verdantiscapital.com | LinkedIn

Top comments (0)